Allergenic Fragrances Analysis in Brazilian Perfumes by Headspace Solid Phase Microextraction and Gas Chromatography-Mass Detector ( HS-SPME-GC-MS )

Perfumes are products mainly consisting of ethyl alcohol, water and fragrance. These fragrances are responsible for characterizing the pleasant and unique odor of each perfume. Among the fragrances, we highlight a group of fragrances which can cause contact allergy, leading to dermatitis. Brazilian and the European law state that when these concentrations of allergenic fragrances exceed the limit of 0.01% for non-rinse products and 0.001% for products with rinsing, the manufacturer is obliged to discriminate on its label their presence. This work aims to quantify allergic fragrances in original and Brazilian perfume using solid-phase microextraction and analyze by gas chromatography-mass detector.


Introduction
Perfumes are products of great importance in the cosmetic industry.Their essential fragrance composition is formed by ethanol and water, in which each product has a unique formulation which allows unique characteristics. 1 Fragrances are volatile organic compounds and semi-volatile compounds which have pleasant scent characteristics.For this reason, they are used in perfumes or scented products of different purposes. 2Some of these fragrances when used by some individuals may manifest allergic reaction when they come into direct contact with skin. 3 Clinical studies on this allergic reaction were carried out first in European Union and then Brazil to adopt more stringent legislation regarding these types of compounds. 4,5ome perfume components at high concentrations can cause allergic reactions.Among these compounds, stands out a group of allergic fragrances that, according to Resolution No. 3/2012 of the National Health Surveillance Agency (ANVISA, Brazil), 5 when found in excess 0.001% (m/m) and 0.01% (m/m) concentrations in products without and with rinsing, respectively, the dermocosmetic product should discriminate in its label their presence.
The gas chromatography mass spectrometry (GC-MS) is the most widely used technique in the analysis of fragrances. 6,7Other techniques have also been recently used such as two-dimensional gas chromatography coupled with flame ionization detector, 8,9 electronic nose 10 and electrospray ionization coupled to mass spectrometer. 2,11he literature reports some works on analysis of allergic fragrances in various matrices such as baby bathwater, 4 fragrances, 8,9 cosmetics, [12][13][14] shampoo, 15 toys, 16 water types (pool and sewage), 17 indoor air 18 and fragrance oils. 19ue to the lack of studies in Brazil on products that can cause allergy and in order to assist the current health legislation, given the large market for skin cosmetics, we sought to investigate and quantify these compounds in perfume samples.
The aim of this work is to propose an analytical methodology for identification and quantification of allergic Vol. 28, No. 11, 2017   fragrances in Brazilian perfumes with minimal amounts of samples, using solid phase microextraction (SPME) and GC-MS.

Samples
A total number of 25 fragrances were acquired: 10 dealer perfumes of authorized Brazilian brand A (PA), 5 perfumes of authorized brand B (PB) and 10 similar perfumes of brand A (PSA) found in informal trade as shown in Table 1.

Gas chromatography-mass spectrometry (GC-MS)
The determination of the allergenic fragrance was performed on a gas chromatograph coupled with a mass detector type quadrupole GC-MS-QP2010 Plus model (Shimadzu, Japan) equipped with a DB5 column (Agilent, USA) (30 m length, 0.25 mm internal diameter and 0.25 µm film thickness, 5% phenyl and 95% polydimethylsiloxane).Helium (99.99%) was used as the carrier gas at constant flow of 1 mL min -1 .
The mass spectrometer conditions were set as follows: ionization mode: electron ionization (EI), 70 eV; ion source temperature 220 °C and transfer line temperature 280 °C.Quantitative analysis was performed in the selected ion monitoring (SIM) mode based on the use of one quantitative fragment and two or three qualitatives fragments.Tables 2 and 3 show the temperature program and the conditions of GC-MS and analyzed fragments.Table 3 shows the retention times of allergic fragrances and their quantitative and qualitative fragments used in SIM method.

Analysis of allergic fragrance perfume by HS-SPME-GC-MS
Previously, a multi-element stock solution in methanol of 10,000 µg L -1 allergenic fragrances was prepared.The following dilutions of the multi-element stock solution, 1-50 µg L -1 , were prepared in milli-Q solvent water.The extraction of the compounds was carried out using 10 mL of solution by adding 2 g NaCl and transferred to a 40 mL vial.The sealed vial was stirred 5 min at 100 °C.After this period, the fiber was exposed in a headspace (HS) for 20 min.Shortly after the analytes extraction, the fiber was taken and brought to the remaining gun for 15 min for complete desorption of analytes in the GC injector.This method is based on the procedure adopted by Lamas et al. 4 and Becerril et al. 17

Method validation
Twenty one allergenic fragrances in perfumes were analyzed by HS-SPME-GC-MS.Validation parameters such as linearity, selectivity, limits of detection (LOD) and quantification (LOQ), accuracy and precision were determined according to guidelines of ANVISA-2003 and ABNT NBR 14029. 20,21The external standard analytical curves were plotted with proper amounts of standard solutions at the concentration range of 1-50 µg L -1 .The significance test of the curves calibration parameters was based on a hypothesis test applying the parameter t test (equations 1 and 2): (1) (2)   where A is the experimental angular coefficient; α = 1 is the theoretical angular coefficient; t calc,A is the t value calculated for the slope; S A is the standard deviation of the slope; B is the experimental linear coefficient; β = 0 is the theoretical linear coefficient; S B is the standard deviation of the intercept; t calc,B is the t value calculated for the intercept. 22,23OD and LOQ values were estimated in the SIM mode analysis as the lowest concentration injected.The values for LOD and LOQ were calculated taking into account the standard deviation (SD) of repetitions (n = 7) of the chromatographic analysis of the lowest point (1 µg L -1 ) of the curve and the slope equation (AC) based on ANVISA 20 (equations 3 and 4).
(3) (4) Precision of the method was evaluated by repeatability (intraday) and intermediate precision (interday) of sample solutions.The intermediate precision assays were performed in three levels of 1, 10 and 30 µg L -1 for three consecutive days (n = 3) and repeatability tests at a level of 30 µg L -1 (n = 6).The results were expressed as %RSD of the measurements.Accuracy of the method was tested with recovery experiments, performed with five replicates of blank samples spiked with 21 allergic fragrances (5, 10 and 30 µg L -1 ) according to guidelines of ANVISA. 20

Selectivity
In literature studies 7,9,24,25 have been shown that for allergic fragrances analysis, GC-MS in SIM mode operating system is good option to the resolution of co-elution of some compounds, however, the two-dimensional chromatographic system has shown promise.
As can be seen in the Figure 1, the scan mode was not efficient to solve the problems of co-elution.Due to this, was chosen in order to work in SIM, because the it solved the co-elution problems and increase the sensibility, as shown in Figures 2-5.
The SIM method solves the problem of co-eluting compounds, for example, as can be seen in Figures 2 and 3, between the citral (m/z 84 and 94) and hydroxycitronellol (m/z 59).In Figures 4 and 5 can be seen between anisyl alcohol (m/z 138) and citronellol (m/z 41, 67 and 69).Table 4 shows the retention time and identification and quatitation fragments of the analytes (see Suplementary Information, Figures S1-S17).

Linearity
The calibration curves of the compounds related to allergic 21 fragrances and their correlation coefficients (R) are given in Table 5.It can be noted that all curves have an appropriate correlation coefficient value according to ANVISA. 20The curves obtained by external standard showed good linearity as well as the results obtained by Lamas et al. 4 and Becerril et al. 17 On the other hand, Debonneville and Chaintreau 9 and Leijs et al. 19 investigated with success the use of two internal standards (1,4-dibromobenzene and 4,4-dibromobiphenyl) with direct injection.
Moreover, results of Table 5 showed that the technique of applying HS-SPME for those compounds did not require the use of an internal standard.However, the application of HS-SPME technique provides low limit of detection (at µg L -1 ) for determination of these compounds within a very satisfactory linear range.The contrast was observed by studying only the headspace process as demonstrated by Sanchez et al. 14 which did not achieve a good linearity to the level of µg L -1 .These results are important from an analytical point of view, because, when working with level of concentration in µg L -1 , small amounts of solutes can be quantified in analytical sample, minimizing the matrix effect, in addition to preserve the integrity of the instrument.

Parameter significance testing of calibration curves
Ideally, the regression lines obtained for the calibration curve should pass through the origin, that is, the intercept of the curve passing through the point (0, 0) of the Cartesian axes.In order to evaluate the statistical significance of the regression parameters A and B, the models obtained experimentally for the calibration curves, it is useful to      The significance test of the curves calibration parameters is based on a hypothesis test applying the parameter t test.Vol. 28, No. 11, 2017   To evaluate the statistical significance of each regression parameter, standard deviations were obtained for the slopes and intercepts of the regression equations and values of the statistical parameter t were calculated for each slope and intercept, according to equations 1 and 2. 22,23 The calculated value of t (t calc ) for the parameters was compared with the tabulated critical value of t (t crit ), to a confidence level of 95% and the degree of freedom (Df = N − 2) for each calibration.When t calc is smaller than t crit the hypothesis that the difference between the calibration parameters obtained experimentally and theoretically expected value is accepted statistically insignificant, and then the experimental calibration parameters are considered equal to the theoretical value (α = 1 or β = 0).The results are shown in Table 6.
As can be seen in Table 6, the results of the statistical analysis of significance of the regression parameters showed that all inclinations are significant if t calc,A > t crit .The intercept, with the exception of the compound of linalool curve, is not significant, considering that t calc,B < t crit and was considered as statistically equal to β, which has a value of zero reference.Therefore, the equations of the curves applied to the calculations of the analytes concentrations are the ones shown in Table 6, with a 95% confidence level.Among the tested compounds, the only compound which shows the statistically significant linear coefficient is linalool.

Limits of detection (LOD) and quantification (LOQ)
The results for the limits of detection (LOD) and quantification (LOQ) are shown in Table 7.The LOD and LOQ values calculated in this work took into account the standard deviation (SD) of repetitions (n = 7) of the chromatographic analysis of the lowest point (1 µg L -1 ) of the curve and the slope equation (AC) based on ANVISA. 20

Precision
The analysis of the coefficient of variation values (CV%), repeatability and intermediate precision showed values ranging from 2.6 to 19.2%, as seen in Table 8.According to ANVISA, an acceptable value is around 5%; however, because of the magnitude of these compounds in the sample to the level 1 µg L -1 , other references 20,21,26 also accept a variation coefficient value of up to 20%.Therefore, our results can be considered acceptable, because, according to the norm of ABNT NBR 14029, 21 which stipulates that for a chemical analysis of the magnitude 10-100 µg L -1 , the value of coefficient of variation between 23 to 32% is accepted, although this rule is for pesticide analysis at this level of concentration.
Our results of intermediate precision presented values lower than 20%, however the literature reports results lower than 10%.However, values obtained in this study (1, 10 and 30 µg L -1 ) do not follow a quantitative standard for intermediate precision as expected, but it was also observed in other studies. 17Analyzing the CV% for the repeatability study in Table 8, it can be seen that the values were above 15%.These values are acceptable according to the standard 14029 of ABNT-NBR. 21However, when compared to literature data, 17 it is noted that the CV values obtained are less than 10%, but these results are for concentration of 20 µg L -1 .Therefore, as in this work the concentration of the compounds studied was 30 µg L -1 , then the difference in the CV values may be due to this fact.

Recovery
Recovery results were satisfactory for most of the 21 compounds analyzed according to ANVISA standard. 20ooking at Table 9, the compounds which showed poor recovery were cinnamic alcohol (10 and 30 µg L -1 ) and lilial (30 µg L -1 ), but considering the RSD these results are consistent with current regulations of 70-120%.The comparison of the recovery values obtained with literature data 4,17 indicates that our values are lower.This can be explained by the complex nature of the perfume composition, which can interfere with the recovery rate of the studied compounds.

Analysis of perfumes
From the results of the analyzes of perfumes (Table 10), it can be observed the presence of some of these fragrances in high concentrations that can cause allergy and thus they need to be informed in the products' labels for consumers' knowledge and meet the current legislation.Among the fragrances studied it was observed that the original perfumes (brands A and B) present allergenic fragrances in high concentrations for all the range of molar weight of the compounds.However, for the similar perfumes was not observed for fragrances with high molar weight, such as benzyl cinnamate and hexylcinnamic aldehyde.This can   be explained due to the low capacity of scent fixation of similars perfumes when compared to originals.

Conclusion
This technique can be used for analysis of allergic fragrance perfume using small amount of sample for analysis.This study may be a help to ANVISA in the control of allergic fragrances in perfumes.The results indicate that the similar perfumes do not present allergenic fragrances with high molecular weight, which influences an inferior scent fixation when compared to the original perfumes.

Table 1 .
Perfume sample: brand A, brand B and similar brand A

Table 4 .
Retention time and identification and quatitation fragments of the analytes

Table 5 .
Calibration curves, correlation coefficient (R) of the compounds studied in a concentration range 1-50 µg L -1

Table 6 .
Results of the statistical test of significance of the parameters of the calibration curves a : standard deviation of the slope; S b : standard deviation of the intercept; t calc,A : t value calculated for the slope; t calc,B : t value calculated for the intercept; Df: degree of freedom; t Df;95% : Student's t-test with confidence level of 95%.

Table 7 .
Limit of detection (LOD) and limit of quantification (LOQ) of allergic fragrances

Table 8 .
Intermediate precision and repeatability of allergic fragrances

Table 9 .
Recovery of allergic fragrances

Table 10 .
Result analysis fragrances in original perfumes and similar