*2.2. Development of Sample Preparation*

#### 2.2.1. Sample Preparation without SPE Clean-Up

Contrary to the vegetable-, fruit- or cereal-based food samples investigated frequently for *Alternarias* earlier (Table S1), the sunflower oil is a very lipophilic matrix and needs a unique sample preparation approach. Therefore, an experimental design was carried out to obtain the appropriate accuracy in different types of sunflower oil samples. A central composition design (CCD) has been planned with the statistical software R, version 3.0.2 for Windows. Two grams of the sample was used for sample extraction with methanol/water mixture; and *n*-hexane was applied for the elimination of lipophilic matrix constituents. This sample weight and these solvents have been found suitable for the *Alternaria* analysis (Table S1). The factors and levels were the following: (I) sample-to-hexane ratio: 1.0, 1.5, or 2.0; (II) methanol content in the extraction medium: 70%, 80%, or 90%; and (III) sample-to-extraction solvent ratio: 2.0, 4.0, or 6.0. A naturally contaminated sunflower oil was used for the experimental design that contained TEA (7.1 μg/kg) and TEN (12.8 μg/kg). An oil sample containing the toxins in much higher concentrations than LOQ would have been better for the CCD, but a sample with greater natural contamination could not be found. An optimal condition was achieved with the sample-to-hexane ratio of 1.0; 80% (*v*/*v*) methanol for extraction, and sample-to-extraction solvent ratio of 4.0 (Figure 3).

**Figure 3.** Response surface of TEN: slice at sample to extraction solvent ratio of 4.0.

In this case, the sample dilution was 4-fold that increased the LOQ value, and AOH could not detected below 10 μg/kg with a signal-to-noise ratio (SNR) higher than 10. Therefore, an aliquot of methanolic extract (6 mL, equal to 1.5 g sample) was evaporated, and the final sample volume was adjusted to 0.5 mL with water that ended up with 3-fold sample pre-concentration (Section 4.2. and Section 4.3.). In this case, all compounds were detected with appropriate SNR at the desired levels. The other aim of sample evaporation was to lower the methanol content of the injection solution, and consequently, the deformation of TEA peak on the chromatogram could be avoided.

One aim of sample preparation is to reduce the ME of LC-MS/MS analysis [37]. ME is caused by the co-eluting matrix constituents and strongly influences the quantification [37]. The ME was studied with the optimal sample preparation conditions and was evaluated using the general approach [37]. Three matrix-matched calibrations were prepared from blank samples (i.e., three different sunflower oils). Blank extracts were spiked with standard mixture: the fortification levels were 10, 20, 30, 40, and 50 μg/kg for TEA, ALT, and TEN; and were 5, 10, 15, 20, and 25 μg/kg for AOH and AME. Low concentration levels were set due to the naturally low contamination of oils with these toxins reported in previous studies (see Table S1). AOH and AME are considered more toxic [4,6], therefore, twice lower levels were set for these two compounds. Calibrants in neat (matrix-free) solvent were also prepared and analyzed. The slopes of matrix-matched calibrations were compared to the slope of neat calibration in order to calculate the absolute ME. ME (%) = (slope in matrix-matched calibration/slope in neat calibration-1) × 100. ME < 0% means ion suppression, and ME > 0% indicates ion enhancement. The relative standard deviation (RSD%) of slopes obtained from the matrix-matched calibrations (*n* = 3) was calculated and evaluated as the relative ME [37]. Therefore, the relative ME means the precision of slopes in different matrix-matched calibrations.

The results obtained without ISTD correction indicated that the high matrix suppression influences the signal of AOH (20–48% ion suppression) and AME (75–88% ion suppression) (Table 2). For TEA, ALT, and TEN, a moderate ME could be seen. The relative ME was also considerable for AOH and AME (22–42%). The ME, however, could be compensated with isotope dilution (Table 2). The calibration evaluated by the ISTD method showed that the ME is greatly compensated, mainly for those two compounds (i.e., AOH and AME) that are considerably influenced by the co-eluting matrix constituents. The relative ME was also improved with ISTD correction. The high ion suppression for AOH and AME, however, indicated that considerable losses of the analytes occurred in the ion source. The reason for high ME is the remaining impurities (i.e., phospholipids) after extraction. Therefore, the SPE clean-up was tested for reducing the number and concentration of matrix constituents, which may lead to lower ME.


**Table 2.** The matrix effect (ME%) and relative matrix effect (RSD% of slopes) evaluated under different sample preparation and evaluation conditions. ME% < 0 means ion suppression, and ME% > 0 means ion enhancement.
