Comparative profile of bioactive compounds in arabica and canephora coffees at different roast levels

Green coffee contains several bioactive compounds of interest, such as caffeine, trigonelline and chlorogenic acids (CGAs). Among the CGAs, caffeoylquinic acids (3-CQA, 4-CQA, and 5-CQA) account for at least 70% the total CGA content. Coffee species is one of the main factors affecting the concentration of these compounds. Because coffee is consumed after roasting, it is also essential to evaluate the effect of this process, which promotes the degradation of several constituents. Therefore, this study aimed to compare the concentrations of these compounds in arabica and canephora coffees subjected to different roast levels. For this purpose, 9 arabica and 9 canephora coffee samples, kindly provided by Bourbon Specialty Coffees, comprised three crop years and different roast levels (light, medium, and dark). Extractions were performed in triplicate using 100 mg of sample and a 50:50 (v/v) methanol:water solution containing 0.1% formic acid, followed by sonication for 1 hour. The extracts were then centrifuged at 3,600 rpm for 5 minutes and filtered through a 0.22 µm PTFE membrane. The target compounds were separated by reversed-phase HPLC-DAD on a Thermo Vanquish Core system, using a HICHROM 3.5 C18 column and gradient elution with mobile phase A (water containing 0.1% formic acid) and mobile phase B (acetonitrile containing 0.1% formic acid). Quantification was performed using calibration curves prepared with reference standards. The concentrations were then evaluated as a function of roast level. Caffeine content was not affected by roasting, remaining significantly higher concentration in canephora coffee (2.2±0.1%) than in arabica (1.25±0.06%). Trigonelline and CQAs decreased linearly with roast degree (observed R²≥0.92): trigonelline (0.19–0.68%), 5-CQA (0.10–1.15%), 4-CQA (0.05–0.50%) and 3-CQA (0.04–0.40%). Given this apparent dependence of concentration with roast degree, ANCOVA was performed to compare species while adjusting for roast degree, which requires the assumption of homogeneity of slopes. For CQAs, this assumption was met, ANCOVA revealed no significant effect of species (p<0.05) indicating that CQAs content did not differ significantly between species. For trigonelline, the homogeneity-of-slopes assumption was violated, indicating that each species has a distinct linear trend. Consequently, the two species were modeled separately. Across the entire experimental range (Agtron 94.3-34.5), arabica consistently exhibited higher contents than canephora. The smallest difference was observed at Agtron 34.5, where the respective regression models predicted contents of 0.17±0.02% for Canephora and 0.24±0.03% for Arabica. No effect of harvest year was observed. Although CGAs are generally reported at higher concentrations in canephora green coffee, roasting virtually equalized their concentrations between the species. These findings indicate that the chemical differences between arabica and canephora become less pronounced after roasting for some bioactive compounds, particularly CQAs, the major class of CGAs.

Funding sources: FAPESP 2022/03229-8, CAPES 88887.713856/2022-00, and CNPq 140751/2023-8.