Validation of Methodology for Quantifying Caffeic and Ferulic Acids in Raw and Roasted Coffee Extracts by High-Performance Liquid Chromatography

Abstract

Caffeic and ferulic acids are critical phenolic compounds in coffee beans, known for their antioxidant properties and influence on coffee’s acidity. This study developed and validated a rapid and simple analytical method to quantify these acids in raw and roasted coffee extracts using high-performance liquid chromatography with a violet detector (HPLC-UV). Parameters such as linearity, accuracy, precision, robustness, limit of quantification, and detection were analyzed. The two acids were quantified in five Coffea arabica cultivars (Red catucai, Siriema, IPR 98, Yellow catuai, and Yellow catucai), and the results were within the standards recommended by ANVISA and INMETRO. The concentrations of caffeic acid ranged from 1.43 to 1.93 mg/g in roasted grains and from 0.16 to 0.38 mg/g in raw grains; for ferulic acid, the concentrations ranged from 0.13 to 0.16 mg/g in raw grains and from 0.27 to 0.50 mg/g in roasted grains. According to the Tukey test, only cultivar IPR 98 showed a statistically significant difference in roasted grains. The proposed method offers a rapid and accurate analysis, contributing to the study of bioactive properties and the relationship between beverage quality and the presence of these compounds in coffee beans.

1. Introduction

Coffee beans contain a wide variety and quantity of organic compounds, the most studied of which are bioactive compounds due to their antioxidant, anti-inflammatory, and antifungal properties [1]. Among the main bioactive phenolic compounds present in coffee beans, caffeic acid and ferulic acid stand out because, in addition to their nutritional properties, these acids contribute to the quality of coffee by influencing acidity characteristics [2] and are compounds that have a protective effect against UV radiation [3]; because of these and other properties, phenolic compounds have been extensively studied. The concentrations of these compounds can vary depending on the type of grain, roasting process, and other factors inherent to the raw material and processing [4].

Caffeic acid is a compound from the class of hydroxycinnamic acids [5]; this class of compounds is found in beverages of plant origin such as coffee and yerba mate and in various foods such as apples, plums, and other fruits; cruciferous vegetables; and cereals, among others [6]. In a study carried out by Soares (2002), it was shown that the antioxidant activity of caffeic acid on the preservation of fatty foods was superior to some other phenolic acids such as cinnamic, p-coumaric, protocatechuic, and vanillic acids, and this more significant activity was attributed to the hydroxyls in positions 3 and 4 of caffeic acid; this is because the hydroxyls in ortho positions provide greater stability to the phenoxy radical through inductive and resonance effects [7,8,9].

Ferulic acid is also a member of the hydroxycinnamic acid class. What distinguishes these two acids is a methoxy group in position 3 of ferulic acid instead of the hydroxyl in position 3 of caffeic acid. Their structural similarity gives them similar bioactive properties, like the other hydroxycinnamic acids [10]. Some work has been conducted to investigate the antioxidant activity of ferulic acid [11] and how this acid acts on free radicals, causing an increase in cellular response through the positive regulation of cytoprotective systems, and it was found that ferulic acid can inhibit the expression of cytotoxic enzymes (cyclooxygenase-2, inducible nitric oxide synthase, and caspases), which can be used to treat diseases such as Alzheimer’s, diabetes, skin diseases, and cardiovascular diseases [12].

The presence of these two acids is higher in roasted coffee beans because the roasting process hydrolyzes chlorogenic acids and transforms them into other compounds, increasing the concentration of hydroxycinnamic acids, such as caffeic acid and ferulic acid, directly affecting coffee’s taste. Caffeic acid is one of the compounds that affect the quality of coffee, and a higher concentration of this acid is associated with more excellent acidity and astringency [13]. With this in mind, the Specialty Coffee Association (SCA) has created a roasting process for all types of coffee that provides the best roasting conditions to enhance the aromas and flavors produced during the process [14].

Chromatography is one of the most efficient and widely used techniques for separating and analyzing organic compounds. Coupling chromatography with other analytical techniques, such as mass spectrometry or ultraviolet or refractive index detectors, makes this method even more robust and allows for the identification and quantification of analytes. Liquid chromatography is the most popular because it does not require the additional sample preparation steps, such as derivatization, that gas chromatography requires to analyze samples with nonvolatile groups. To analyze some classes of organic compounds, such as phenolic acids, liquid chromatography with an ultraviolet detector is a robust and effective method, and the analysis is rapid, depending only on the sample preparation time [15].

Some studies are reported in the literature on the determination of organic acids and cinnamic acid derivatives in coffee, such as that of Yilmaz and Kolak (2017), who validated a methodology using SPE-HPLC to determine chlorogenic and phenolic acids in coffee, with a chromatographic analysis time exceeding 50 min. Other studies are found, such as that of Klikarová and Česlová (2022), who analyzed organic acids by HPLC in coffee, seeking to assess authenticity, however, without the validation of the methodology, whereas Gani and Istyastono (2021) validated the methodology but only for the determination of caffeic acid in coffee powder [16,17,18].

Considering that there is no specific validated methodology for the determination of caffeic and ferulic acids in raw and roasted coffee samples, this study aims to validate a rapid and low-cost methodology for the simultaneous analysis of these two acids in raw and roasted coffee extracts using high-performance liquid chromatography with an ultraviolet detector (HPLC-UV), since these compounds are of great importance due to their organoleptic activities and because they influence the quality of coffee beverages. In addition, they have the potential to be applied to other plants that have considerable concentrations of these acids, such as Camellia sinensis (green tea) and Ilex paraguariensis (yerba mate).

2. Materials and Methods

2.1. Standards and Reagents

Acetonitrile (Sigma-Aldrich, St. Louis, MO, USA) HPLC grade 99.9%, glacial acetic acid (Fmaia, São Paulo, Brazil) UV/HPLC 99.7%, and high-purity water were obtained from Milli-q-Plus system (Millipore Corporation, Darmstadt, Germany). Standard solutions of caffeic acid (Sigma-Aldrich, St. Louis, MO, USA) >95% purity and ferulic acid (Sigma-Aldrich, St. Louis, MO, USA) 99% purity were also used, both in solid state.

2.2. Equipment

An Agilent 1260 HPLC liquid chromatograph equipped with a G7111B 1260 Quat pump, a G7129A 1260 Vial sampler autosampler, and a G7117C 1260 DAD HS detector were used (Agilent, Santa Clara, CA, USA). The column used was a Luna Omega Polar C18 LC column with internal diameter 4.6 mm, length 150 mm, particle size 5 µm (Phenomenex, Aschaffenburg, Germany) connected to an Eclipse XDB-C18 pre-column with internal diameter 4.6 mm, length 12.5 mm, particle size 5 µm (Agilent, Santa Clara, CA, USA).

2.3. Coffee Samples

Beans from 5 coffee tree genetic materials (Red catucai, Yellow catucai, Yellow catuai, Siriema, and IPR 98, raw and roasted beans) were used, with three biological repetitions, from the 2021 harvest of the Francisco Melo de Palheta Variety Field of the Federal University of Viçosa Rio Paranaíba campus.

2.4. Sample Preparation

All the samples underwent the same post-harvest processing and roasting, following the SCAA protocol for specialty coffees [19]. The beans were ground using a cryogenic mill (IKA A11basic S032) using liquid nitrogen to make it easier to break the beans. After milling, the samples were sieved through a 20-mesh stainless steel 304 sieve to standardize the grain size, stored in adequately sealed falcon tubes, and refrigerated at −80 °C until the analysis day.

2.5. The Figures of Merit Used in the Validation of the HPLC Method

In validating the methodology for quantifying caffeic acid and ferulic acid in extracts of raw and roasted coffee beans by high-performance liquid chromatography with an ultraviolet detector with a diode array (HPLC-DAD-UV), the following parameters were evaluated: linearity, precision (recovery), precision (reproducibility), robustness, matrix effect, limit of quantification (LOQ), and limit of detection (LOD).

2.5.1. Linearity

Linearity was analyzed by using an analytical curve in triplicate on three consecutive days, at concentrations of 0.25, 0.5, 1.0, 2.0, 4.0, 8.0, 12.0, 16.0, 20.0, and 24.0 µg mL⁻¹. High-purity water was used to prepare the standard solutions of the respective reference substances (caffeic acid and ferulic acid).

2.5.2. Accuracy

Accuracy was assessed by adding the standards to a sample with a known concentration of the analytes (in raw grain—caffeic acid: 0.5 µg mL⁻¹ and ferulic acid: 0.5 µg mL⁻¹ and in roasted grain—caffeic acid: 3.2 µg mL⁻¹ and ferulic acid: 1.2 µg mL⁻¹). The following nominal concentrations were added: 4.0, 8.0, 12.0, 16.0, and 20.0 µg mL⁻¹, all carried out in triplicate.

2.5.3. Precision

Precision was studied using a reproducibility test on three alternate days and in quintuplicate, with standard solutions at concentrations of 5.0, 10.0, 15.0, and 22.0 µg mL⁻¹. The results were analyzed using the coefficient of variation (CV) of the series of measurements.

2.5.4. Robustness

Robustness was assessed using the Plackett–Burman method to check the effects of multiple variations (

Table 1. The parameters and variations used in the planning to assess robustness. Acetonitrile concentration in the mobile phase; (p2) column temperature (°C); (p3) mobile phase flow rate (mL min⁻¹); (p4) wavelength (nm); (p5) acetonitrile brand; (p6) acid concentration in the mobile phase; (p7) acid type.

	Conditions		Combination of Factors
Parameters	Nominal (1)	Variation (−1)	1	2	3	4	5	6	7	8
p1	15	16	−1	−1	−1	1	−1	1	1	1
p2	40	42	1	−1	1	1	−1	−1	−1	1
p3	1	1.1	−1	1	1	−1	−1	−1	1	1
p4	272	274	−1	1	−1	1	1	−1	−1	1
p5	Êxodo Científica	Dinâmica Química Contemporânea LTDA	1	−1	−1	−1	1	−1	1	1
p6	1	1.5	−1	−1	1	−1	1	1	−1	1
p7	acetic	formic	1	1	−1	−1	−1	1	−1	1

) on the chromatographic method [20]. The variation in the results on area and retention time was analyzed using Lenth’s Student’s t-test with a significance level of 5%. A standard solution containing the two analytes was used.

2.5.5. Limit of Quantification

The limit of quantification was determined from the lowest value of the calibration curve used in the linearity test. The signal/noise ratio was also analyzed using OpenLab software (Agilent LC 1260) to verify a value greater than 10:1.

2.5.6. Limit of Detection

The detection limit was determined from the signal/noise ratio greater than 3:1, using OpenLab software (Agilent LC 1260), by analyzing three replicates of the blank measured on each day of the method’s validation.

2.5.7. Matrix Effect

Two calibration curves were constructed using different concentrations of caffeic acid, one based on the linearity of the method at concentrations of 4.0, 8.0, 12.0, 16.0, 20.0, and 24 µg mL⁻¹ and the other considering the addition of analyte standard (doping) at the same concentrations. The relationship between the area obtained and the concentration was analyzed by linear regression, and the angular and linear coefficients were compared by means of coincidence and intercept tests.

2.6. Extraction of Bioactive Phenolic Acids

The bioactive compounds were extracted in a water bath at 90 °C for 5 min using 0.5 g of the sample, ground and sieved, and 50 mL of distilled water. The mixture was filtered through No. 4 filter paper and a 0.45 mm hydrophilic syringe filter. The filtrate was immediately injected into the liquid chromatograph.

2.7. Analysis by HPLC-UV

The method was initially improved for the best chromatographic conditions in order to avoid the co-elution of analytes and other substances and to obtain a shorter chromatographic run time. The identification of acids was performed by analyzing the retention time of the analytes through standard solutions of the respective acids. The chromatographic condition used was gradient mode, with the mobile phase consisting of a 1% solution of acetic acid in water (solvent A) and acetonitrile (solvent B), following the following proportion: 75% solvent A and 25% solvent B for 5 min and for cleaning the column and 84:15 v/v for a further 10 min, with the detector at a wavelength of 272 nm, an oven temperature of 40 °C, an injection volume of 10 µL, and a flow rate of 1.0 mL min⁻¹.

3. Results and Discussion

3.1. Chromatograms of Raw and Roasted Coffee Extracts

To verify the quality and resolution of the chromatograms of the raw and roasted coffee extracts, an analysis of different parameters was carried out in both extracts, such as the following: retention factor (k’), number of theoretical plates (N), symmetry (S), resolution (Rs), and selectivity (Se). The chromatogram of the raw and roasted coffee extracts shows the elution of the analytes in less than five minutes (Figure 1).

In both chromatograms, the theoretical plates were above 2000 according to the acceptance criteria of the Food and Drug Administration [21,22]. Considering the symmetry results (>0.58) together with the resolution values (1.84–1.88 for caffeic acid and 2.33–2.44 for ferulic acid) and what was proven by the accuracy results, the determination of the compounds was not affected. The values of symmetry, resolution, and selectivity were influenced by the chromatographic conditions, such as the composition and flow rate of the mobile phase and column pressure, among others [23]. The k’ values indicate a moderate interaction between the stationary phase and the analytes (k’ of 1.37–1.41 for caffeic acid and k’ of 2.57–2.59 for ferulic acid), which means there is a good separation between the bands.

3.1.1. Linearity Results

The analytical curves showed homoscedastic behavior (Cochran’s test). The tabulated C value (0.616) was higher than those calculated for the two analytes, so the curve was constructed using ordinary least squares. The appropriate linear behavior can be seen from the correlation and regression coefficient values above 0.990, by ANVISA’s 2017 RDC 166 guide (

Table 2. Results of regression statistics for caffeic and ferulic acids.

	Caffeic Acid	Ferulic Acid
Correlation coefficient	0.9980	0.9978
R2	0.9960	0.9948
Linear coefficient	142.87	132.34
Angular coefficient	130.60	120.45
Limit of quantification (µg mL−1)	0.250	0.250
Limit of detection (µg mL−1)	0.075	0.075
C calculated	0.511	0.385

).

An ANOVA was carried out to assess the quality of the linear regression, in which the angular coefficient and intercept for both analytes were highly significant, with a p-value below 0.05 (Table S1, Supplementary Materials).

The analysis of the intercept using Student’s t-test showed a p-value of less than 0.05 (both acids), thus rejecting the null hypothesis (intercept equal to zero) at a significance level of 5%. The calculation values were higher than those in the table (2.12). Therefore, we assume that the linear coefficient differs from zero (Table S2, Supplementary Materials).

3.1.2. Accuracy Results

The average recovery (

Table 3. Average acid recovery values in roasted coffee extracts.

Compound	Caffeic Acid			Ferulic Acid
Theoretical Concentration (µg mL−1)	Experimental Concentration (µg mL−1)	Recovery (%)	Coefficient of Variation (%)	Experimental Concentration (µg mL−1)	Recovery (%)	Coefficient of Variation (%)
4	3.82	95.5	2.29	4.16	104.0	3.32
8	7.25	90.6	0.81	9.03	112.8	1.45
12	11.26	93.8	0.11	12.06	100.5	0.84
16	15.63	97.7	2.16	15.74	98.38	3.33
20	19.22	96.1	3.51	19.52	97.6	1.37
Average	-	94.7	-	-	102.66	-

) in the roasted coffee extract samples showed a minimum value of 90.6% for caffeic acid and 97.6% for ferulic acid; only one value exceeded 10% of the added concentration, and as it was slightly higher than that in the INMETRO guideline (90–107%), the method is considered to have good accuracy in terms of the extract of roasted beans.

In the raw coffee extract (

Table 4. Average acid recovery values for raw coffee extracts.

Compound	Caffeic Acid			Ferulic Acid
Theoretical Concentration (µg mL−1)	Experimental Concentration (µg mL−1)	Recovery (%)	Coefficient of Variation (%)	Experimental Concentration (µg mL−1)	Recovery (%)	Coefficient of Variation (%)
4	4.50	112.5	3.32	4.18	104.5	3.71
8	8.15	101.8	1.86	8.50	106.3	1.13
12	11.75	97.9	3.05	11.80	98.3	3.01
16	16.68	104.3	0.49	15.21	95.1	0.31
20	20.25	101.3	2.04	19.10	95.5	0.72
Média	-	103.56	2.15	-	99.94	1.78

), the average recovery showed a maximum value of 112.5%, referring to the lowest concentration of caffeic acid, and a minimum value of 95.1% for ferulic acid. The only value that exceeded 10% of the theoretical concentration added refers to the lowest concentration of caffeic acid; however, in the other concentrations, there is no considerable variation, which does not affect the method’s accuracy [24].

3.1.3. Precision Results

The reproducibility test (

Table 5. Values for evaluating the accuracy of the chromatographic method for determining caffeic and ferulic acids.

Compound	Caffeic Acid		Ferulic Acid
Theoretical Concentration (µg mL−1)	Experimental Concentration (µg mL−1)	Coefficient of Variation (%)	Experimental Concentration (µg mL−1)	Coefficient of Variation (%)
5	4.96	0.17	5.22	3.63
10	9.39	0.86	9.66	2.93
15	14.22	1.54	14.11	1.75
22	21.55	2.08	21.18	3.30

) showed coefficient of variation values between 0.17 and 2.08% for caffeic acid and 1.75 and 3.63% for ferulic acid, values which are lower than 5.3% according to the INMETRO guidelines (2020), indicating the good precision of the method [25].

3.1.4. Robustness Results

Regarding retention time (Figure S1, Supplementary Materials), no effect was significant, and the effect values were below the margin of error (ME), with an ME value of 1.0833 for caffeic acid and 0.9735 for ferulic acid. There was no significant influence in relation to area, with ME values of 1.1883 for caffeic acid and 1.1943 for ferulic acid (Figure S2, Supplementary Materials). The results of Student’s t-test using Lenth’s method (

Table 6. The effects of the variations obtained by Plackett–Burman planning and Student’s t-test. (p1) Acetonitrile concentration in the mobile phase; (p2) column temperature (°C); (p3) mobile phase flow rate (mL min⁻¹); (p4) wavelength (nm); (p5) acetonitrile brand; (p6) acid concentration in the mobile phase; (p7) acid type.

Parameters	Caffeic Acid		Ferulic Acid
	p-Value for RT	p-Value for Area	p-Value for RT	p-Value for Area
p1	0.8783	0.5219	0.5173	0.7739
p2	0.7109	0.7322	0.7288	0.8175
p3	0.2769	0.5038	0.3319	0.3417
p4	0.2150	0.0892	0.3788	0.1254
p5	0.4316	0.6108	0.5649	0.5649
p6	0.8565	0.3213	0.7896	0.3893
p7	0.5649	0.6600	0.5649	0.6322

) showed no significant values for any of the acids for the seven effects studied. The method is, therefore, considered to be robust.

3.1.5. Matrix Effect Results

The coincidence test, which assesses whether the slopes of the curves are significantly different, resulted in a value of t = −0.04 and p = 0.9685. Since the p-value is much higher than the significance level of 0.05, there is no statistical evidence to reject the hypothesis of the equality of the slopes. This indicates that doping does not alter the sensitivity of the method. The intercept test showed a displacement between the lines, with t = −53.16 and p = 0.0000, indicating a statistically significant difference between the intercepts, which shows that the curves are parallel and not coincident. The curves for caffeic acid can be seen in Figure S3 in the Supplementary Materials.

3.2. The Quantification of Caffeic and Ferulic Acids Using the Methodology Developed in Five Different Varieties of Raw and Roasted Coffee

The validation of the method showed values within the analytical validation standards for a safe and efficient validation for all parameters analyzed. The two acids were therefore quantified in five different genotypes of Coffea arabica, including raw and roasted beans (

Table 7. Concentration of caffeic acid and ferulic acid in raw and roasted coffee extracts using proposed analytical method, with Tukey’s test results.

	Red Catucai	Yellow Catucai	Yellow Catuai	Siriema	IPR 98
	Caffeic acid (mg g−1)
Roasted	1.564 Aa	1.646 Aa	1.429 Aa	1.541 Aa	1.934 Ab
Raw	0.197 Bc	0.156 Bc	0.224 Bc	0.233 Bc	0.384 Bd
Training fee	693.91%	955.13%	537.95%	561.37%	403.65%
	Ferulic acid (mg g−1)
Roasted	0.296 Ab	0.303 Ab	0.276 Ab	0.354 Ab	0.504 Aa
Raw	0.135 Bc	0.162 Bc	0.142 Bc	0.127 Bc	0.141 Bc
Training fee	119.26%	87.04%	94.34%	178.74%	257.45%

Lowercase letters compare the means between the varieties, and uppercase letters compare the means between the raw and roasted processes. Values with the same letter do not differ according to Tukey’s test.

). The caffeic acid levels found in raw coffee (0.156–0.384 mg g⁻¹) are similar to those reported in the literature (0.25 mg g⁻¹) [26]. The levels of ferulic acid in the roasted coffee extracts (Table 7) ranged from 0.29 to 0.50 mg g⁻¹, values close to those of caffeic acid in raw beans. Due to the difficulty of quantifying these two acids, their low concentration, and the large number of compounds present in coffee extracts, few studies have been published on determining these acids in coffee.

Roasted beans have a higher concentration of both acids and are statistically different from raw beans according to the Tukey test. The more significant variation in caffeic acid is seen because the chlorogenic acid with the highest concentration in raw coffee beans (5-caffeoylquinic acid) is formed by the esterification of quinic acid and caffeic acid, and in the roasting process, hydrolysis occurs, forming caffeic acid, among other products [27]. The variation in concentration is not uniform, which can be explained by the complexity of the reactions that occur in the roasting process and the fact that caffeic acid and ferulic acid are formed by the hydrolysis of other chlorogenic acids, caffeoylquinic acids generating caffeic acid and feruloylquinic acids generating ferulic acid [28], and they therefore depend on the concentration of other chlorogenic acids in raw coffee beans.

The highest concentrations of caffeic acid (raw beans) were found in Siriema and IPR 98 (with only IPR 98 showing a statistical difference from the other varieties), which showed rust tolerance. Some studies have shown that the content of chlorogenic acids may be related to the resistance of the coffee plant to the occurrence of rust [29]. Still, there is no published article relating caffeic acid and ferulic acid to the plant’s resistance to the pathogen Hemileia vastatrix, which causes rust. Future work could correlate the occurrence of pathogens and the presence of these two acids in the coffee plant.

4. Conclusions

The methodology developed for the analysis of caffeic and ferulic acids in raw and roasted coffee extracts using HPLC-UV showed linearity, accuracy, robustness, and precision at the concentrations studied, according to the ANVISA DRC 166/2017, INMETRO (2020), and ICH (2005) regulations for the validation of analytical methods. This indicates the accuracy of the proposed analytical method.

The variation in the concentration of these compounds between different coffee genotypes is significant only for the IPR 98 variety, which may indicate that the resistance of this variety to pathogens may be related to the concentration of these two acids. The methodology developed represents a good tool for studying the composition of these phenolic acids, which will allow for a better understanding of the bioactive properties in coffee and how different factors influence their concentration in other cultivars, as well as how the quality of the drink is related to the presence of these compounds.