, 2013 and Wijekoon RG 7204 et al., 2011). Comparison of all evaluated extractions with methanol and acetone aqueous solutions revealed that most of the acetone solutions extracted more phenolic compounds than the

hydro-methanolic solutions. The optimisation procedure was conducted in order to simultaneously maximise the total phenolic content, total flavonoids, and antioxidant capacity measured by FRAP and also to minimise DPPH values. The final result for this optimisation suggested that extraction with 84.5% methanol for 15 min, at 28 °C, and extraction with 65% acetone for 20 min, at 10 °C were the best solutions for this combination of variables. These new extractions were submitted to the same experimental analytical procedures as those applied from the beginning of this study. The observed and predicted values, along with the computed absolute errors (AE) for methanolic extraction were: total phenolics (mg/100 g) (observed: 590.82 ± 5.54; predicted: 588.81; AE = 0.34%), total flavonoids (mg/100 g) (observed: 165.55 ± 1.39; predicted: 164.47; AE = 0.66%), DPPH (mg/100 g) (observed: 2439.89 ± 72.55; predicted: 2441.10; AE = 0.05%), FRAP (μM/100 g) (observed: 1863.78 ± 24.67; predicted: 1835.31; AE = 1.55%). For extraction with the acetone solutions, the observed and predicted values, along with the computed

absolute errors (AE), were: total phenolics (mg/100 g) C646 cell line (observed: 738.23 ± 10.52; predicted: 711.59; AE = 3.74%), total flavonoid content (mg/100 g) (observed: 334.45 ± 2.72; predicted: 325.09; AE = 2.88%), DPPH (mg/100 g) (observed: 1856.00 ± 19.90; predicted: 1958.06; AE = 5.20%), FRAP (μM/100 g) (observed: 1960.13 ± 54.43; predicted: 1934.36; AE = 1.33%). Because of the low absolute error values obtained by the comparison between observed and predicted values, the proposed model could be used to predict the response value. The phenolic profile of the extracts was determined in the best conditions of extraction for phenolic and antioxidant capacity (Table 5). The chromatograms of phenolic compounds analysed are shown in

Fig. 1. Gallic, coumaric and caffeic acid, phloretin, quercetin, kaempferol and myricetin were not detected in the samples analysed by HPLC. Except for chlorogenic acid and phloridzin, the extract from the acetone 17-DMAG (Alvespimycin) HCl solution had the highest content (p ⩽ 0.05) of the individual phenols analysed. These results showed that the recovery of phenolic compounds is influenced by the polarity of the solvent used, as reported in other studies ( Kchaou et al., 2013 and Wijekoon et al., 2011). Methanol and acetone seem to have different specificities in the extraction of phenolic compounds. Total phenolic compounds and total flavonoids in methanolic extractions had a significant (p ⩽ 0.05) correlation with antioxidant capacity measured by the DPPH (r = −0.75; r = −0.52, respectively) and FRAP (r = 0.62; r = 0.53, respectively) assays.