Anexos / Apêndices - 165
out in a randomized order. The results were evaluated with the program Design-Expert
®
, Version 7.1.6. (StatEase, Minneapo-
lis, MN, USA).
The effects were calculated presuming a linear model with interaction among the factors. The equation model was defined as:
Y = M + aA + bB + cC + abAB + acAC + bcBC + abcABC (1)
where Y is the model response, M is the mean value, A, B, C are the main factors (A = ethanol concentration; B = extraction
time; C = temperature or presence of a preservative), AB, AC, BC, and ABC are the binary and ternary interactions between
the factors, and a, b, c or ab, ac, bc, and abc are the coefficients of the main factors and interaction factors, respectively,
employing a probability error of p < 0.05. The equation terms were determined by reverse hierarchical regression analysis.
Analysis of variance (ANOVA) was performed to support the polynomial equations and to identify the significance of single
factors and their interactions.
3. RESULTS AND DISCUSSION
3.1. Analytical method validation
Reversed-phase high performance liquid chromatographic (HPLC) was proposed as a suitable method for the simultaneous
determination of CGA and CFA in ES. The method developed was then validated in terms of accuracy, precision, linearity,
quantification limit and detection limit, in accordance with the ICH guidelines [23]. The nature of the sample determines which
parameters should be evaluated, especially when the samples are complex biologic matrices, as in the case of plant extraction
solutions [24]. Figure 1 shows a typical chromatogram for the ES. The retention times investigated (3.5 min for CGA and 6.0
min for CFA) showed a sharp and symmetrical peak, with good baseline resolution and minimal tailing, thus facilitating the
accurate measurement of peak area ratios.
3.1.1. Precision and accuracy
The precision and accuracy were determined by spiking the sample with a known quantity of the standard. The mean recovery
was calculated on one assay for each standard. Good accuracy was observed with satisfactory recovery in the range of 98.00 –
102.30% (Table 2). Measurement of intra- and inter-day variability was used to determine the precision of the newly devel-
oped method. The intra-day variation (repeatability) was determined by analyzing in triplicate the mixed standard solution
three times within 1 day. While, for the inter-day variability test (intermediate precision), the solution was examined in
triplicate on 3 separate days. The percentage relative standard deviation (% RSD) was taken as a measure of precision. The
results for the precision showed low values (less than 5.0%) for intra and inter-day % RSD as shown in Table 3.
3.1.2. Calibration curves, linearity and detection and quantification limits
The calibration curves were found to be linear over the range of 2.5 – 200 µg/mL for CGA and 2.5 – 100 µg/mL for CFA. The
ranges and correlation coefficients are given in Table 4. All calibration curves showed good linear regression in the range of r
2
= 0.999974 – 0.999990. The DL and QL values for CGA were found to be 0.3460 and 1.05µg/mL, and for CFA were 0.0732
and 0.22µg/mL, respectively (Table 4). In order to observe the presence of interference, extraction solution curves were used to
determine the amount of samples (8d20, 6d50 and 8d80) where linearity was found. The regression equations and the correla-
tion coefficients of the extraction solutions are shown in Table 5. Excellent linearity was obtained for both phenolic acids in
the extraction solution 8d20 and CFA in 6d50, demonstrated by the correlation coefficients of the curves. However, in the case
of 6d50 and 8d80, there was linearity deviation for the CGA curves with correlation coefficients of 0.9642 and 0.9229,
respectively (Table 5). On the other hand, for ES concentrations higher than 30% (6d50) and 22% (8d80), the correlation
coefficients were, respectively, 0.9917 and 0.9919 for CGA as shown in Table 6. In the case of CFA in 8d80 we could not
determine linearity. The reason for this result might be a combination of a low concentration of caffeic acid and a high
percentage of EtOH, changing the polarity of the sample.
3.2. Statistical analysis for stability study
Initially, CGA was chosen as the chemical marker because phenolic acid has been reported as the major compound in aqueous
and/or butanolic extracts of Cecropia species [6, 25, 26, 27, 28, 29]. However, in this study, mixtures of ethanol and water
were chosen as nontoxic and environmentally friendly solvents, which have been shown to be effective in the extraction of
polyphenolic compounds. Since the type and magnitude of the extraction variables can affect analyte recovery, a suitable
design (2
3
full factorial) design was used. The experimental parameters analyzed initially were ethanol concentration and time.
Surprisingly, in the 20% ethanolic extracts we observed a considerable amount of some other compound that was not detected
in the extracts with higher ethanolic concentrations. Later, we identified it from the DAD spectrum as being caffeic acid (Fig.
2), and confirmed this finding through it having the same retention time as the CFA standard solution. In addition, when
comparing CGA concentrations at 4d20 and 8d20 ES, a significant reduction in its yield from 4 to 8 days was observed. This
effect might be related to a chemical and/or enzymatic and microbiological degradation, considering that chlorogenic acid is an
ester of caffeic acid and quinic acid [30]. However, there are no reports of the degradation of CGA during the extraction
process. In order to investigate this possible degradation, the influence of temperature and the presence of a preservative on
CGA content in the extraction solutions were evaluated. For this, these two variables, each with two levels, were introduced
into the experimental design.
The experimental design allows the maximum amount of information to be obtained from the data collected in the smallest
number of experimental runs. The basic idea is to change all relevant factors simultaneously over a set of planned experiments
and then connect and interpret the results using mathematical models [31]. To provide a statistical verification of the response
curvatures, two center points were added to each design. Also, each experiment was repeated twice. This type of design was
used to provide additional degrees of freedom and to measure the experimental error [31]. The CGA and CFA content of the
18 runs of each experimental design are presented in Table 7. The experiments resulted in concentrations of CGA ranging from
62.76 to 173.59 µg/g and of CFA from 4.90 to 46.21 µg/g.
3.2.1. Influence of Extraction Temperature