Response Surface Methodology optimization of HS-SPME-GC-MS method for the analysis of pentene dimers and terpenes in extra virgin olive oil

The volatile fraction of virgin olive oil (VOO) includes hundreds VOCs, present in concentrations ranging from ng/kg to mg/kg, affects the sensory characteristics of VOOs and is affected by several factors. It is recently widely used for the quality control of VOO. Several volatile hydrocarbon such as pentene dimers (linear and branched C10 hydrocarbons from the LOX-pathway) and mono- and sesquiterpenes have been detected in the volatile fraction of VOO but are less investigated to date, and are thought to contribute to the pleasant notes of EVOOs. Terpenes might be varietal and geographical differentiators. Some analytical methods for analysis of hydrocarbons were very time consuming; therefore, a suitable method for the simultaneous analysis of pentene dimers and terpenes in EVOO is required. HS-SPME-GC-MS is the most common approach for analysis of VOO volatile profile, being cost-effective, solvent-free, easy to adopt, fast and versatile, and not requiring sample preparation. The Design of Experiments (DoE) with Response surface methodology (RSM) can be suitable for optimization of the HS-SPME pre-concentration step, in particular using the method operable design region (MODR), which is the zone where the requirements are fulfilled with a certain probability. The aim was developing a HS-SPME-GC-MS method for the simultaneous analysis of pentene dimers and terpene hydrocarbons in EVOOs. RSM has been applied for the optimization of the critical process parameters (CMPs) of VHCs pre-concentration by HS-SPME, thus obtaining the MODR and leading to the selection of a working point to be used for routine analysis. A quantitation method was set up using a number of external and internal standards, it was then validated and applied to a group of monovarietal EVOOs. Samples: Stock solutions of external (ESTD) and internal standards (ISTD) of terpenes and pentene dimers were prepared in a refined olive oil. Eight levels of calibration scales were prepared by mixing the same amount of ISTD and increasing amounts of ESTD. Four monovarietal EVOO were used for preliminary trials. A pooled sample of Coratina and Altomira cvs was used for the Doehlert Design experiments. Eight EVOOs were analyzed using the validated method: 4 of the Moraiolo and 4 of the Tonda Iblea cultivar. Sample amount, extraction time, extraction temperature and desorption time of HS-SPME step were optimized using a Doehlert Design experiment. A 50/30 μm DVB/CAR/PDMS 1-cm SPME fiber was employed for extraction of VHCs from the HS of 20-ml screw vials at the selected conditions. The VHCs were desorbed at 260 °C in a 6890N GC system with a 5975-MS detector ( all from Agilent, Palo Alto, CA, USA), and separated in a HP-Innowax capillary column (50m×0.2mm id, 0.4 μm ft). Oven: 2 min at 40°C; to 156°C at 4 °C/min; to 260°C at 10°C/min. MSD worked in scan mode at m/z 29-350 Th, IE energy 70 eV. Peaks were identified using commercial standards when available; in the other cases the retention index evaluated analyzing C9–C30 linear alkanes and the NIST08/Wiley98 library were used. Eight-levels calibration lines were built, and the response factors were calculated after normalizing the peak area using the ISTDs. Validation by considering repeatability, LOQ, LOD, linear range of calibration, accuracy, sensitivity and selectivity was performed. 65 VHCs were identified, including pentene dimers and terpenes. HS-SPME preconcentration step was optimized by RSM, making previsions all throughout the experimental domain. The domain of the CMPs were: sample amount (SaAm), 2.1500-8.6000 g; extraction time (ExTi), 20-80 min; extraction temperature (ExTe), 30-90 °C and desorption time (DeTi), 1-5 min. The responses included both cumulative areas of groups of VHCs and areas of individual compounds of interest. The responses related to sesquiterpenes content were selected for building the MODR. Quadratic polynomial models relating the factors to the responses were hypothesized and the coefficients of the model were calculated by means of a Doehlert Design, a matrix with high efficiency (i.e., low n° of experiments). Each factor was studied at a different number of levels uniformly distributed for an experimental plan with a total of 23 experiments. Three replicates at the center were performed, enabling the estimation of the experimental variance. Logarithmic transformations of responses and model refining were done excluding the factors that were found to be not significant, obtaining very good results in terms of quality of the models. All the models were significant, while validity was verified for the majority of the responses. All the models were considered acceptable due to the small residuals and to the high values of Q2, which indicated a good prediction quality. Graphic analysis of effects made it possible the direct evaluation of the significant effects of the CMPs on the responses. The trend of the predicted values of the responses could be easily visualised by drawing the four-dimensional contour. The conditions which made it possible to optimize both these responses corresponded to the red zone, located at high ExTi, medium ExTe, low SaAm and low DeTi values. Taking into account Doehlert Design results, target values for the three critical method attributes CMAs related to sesquiterpenes were defined and the sweet spot plots with the zone where all the CMAs are fulfilled were drawn. In the next step, the MODR was defined, which includes any combination of the variables that provide assurance of quality of the data produced by the method. The MODR around the set-point is in green in Fig. 2. Inside the MODR, the working point was chosen as the same set-point originally selected. It was at: SaAm, 3.27 g; ExTi, 65 min; ExTe, 90 °C and DeTi 1.70 min. Using these optimized conditions, a quantitative method was developed and validated and applied to samples of the Moraiolo and Tonda Iblea cultivars, which showed different VHCs profiles.