Potential and Limitations for Determining Lycopene in Tomatoes by Optical Methods

Lycopene in tomatoes and tomato products is routinely determined by extraction into organic solvents and spectrophotometric quantification. The direct determination of lycopene, by measuring the color or other optical properties of a tomato homogenate, would be an attractive alternative. We have evaluated two instruments for this purpose. The color of tomatoes and tomato products is routinely determined with a reflectance colorimeter such as a LabScan (Hunter Lab). We have examined whether the parameters determined with this instrument correlate sufficiently well with the lycopene content of raw tomato juice to allow for the prediction of lycopene levels. CIE “a” values were not linearly related to lycopene content but could be fit to a logarithmic regression line. A better, but still logarithmic, fit was obtained when the parameter optical density at 560 nm minus the optical density at 700 nm, was plotted versus lycopene. The absorbance of light transmitted through tomato juice was also measured with a Hunter Lab UltraScan. The parameter, absorbance at 560 nm minus absorbance at 700 nm was linearly related to the lycopene content of the tomato juice. The slope of the regression line, however, was affected by the method used to homogenize the tomato. Measured values for (A560-A700) were also substantially lower in cooked versus raw tomato juice samples that contained equal amounts of lycopene. Evidently, factors other than the total lycopene content greatly affect the absorbance values measured by the UltraScan. INTRODUCTION In tomatoes, lycopene is the principal carotenoid, comprising about 95% of the total, and is responsible for the red tomato color. Intuitively it seems apparent that the intensity of tomato color should correlate with the total lycopene content. Correlations between color, usually reported as CIE a value, and lycopene content have been reported and proposed as a way to estimate the lycopene content (Arias et al., 2000; D'Souza et al., 1992). In addition, direct measurements of light absorbance by tomato juice, at the wavelengths of maximum absorbance by lycopene, have also been shown to correlate with lycopene content. This also has been proposed as an alternative method for estimating lycopene in tomatoes (Davis et al., 2003). Here we have evaluated these methods and examined some of the factors that may affect their usefulness. MATERIALS AND METHODS Tomatoes used to prepare fresh tomato homogenates were both processing types (evaluated in annual variety evaluation program; specific varieties not specified) and fresh market type (‘Early Girl’). Homogenates were prepared either by grinding in a blender with added water (in some cases at a 1:1 ratio of water to tomato, in others at a 3:1 ratio), or by grinding in a mortar and pestle followed by passing through a screen to remove any large pieces of skin. Various dilutions of these homogenates were then examined. When homogenized in a blender, samples were allowed to stand for one hour prior to measuring to allow any air bubbles to completely dissipate. Microwave hot-break juice from tomatoes was prepared and stored frozen as described (Barrett and Anthon, 2001). For 243 Proc. 9 th IS on the Processing Tomato Ed. W.J. Ashcroft Acta Hort. 724, ISHS 2006 optical measurements 5 g of juice was diluted to 20 mL with water. Absorbance measurements of the tomato homogenates and hot-break juice samples were made with an UltraScan XE (Hunter Associates Laboratory, Reston, VA), as described by Davis et al. (2003). Measurements were made in transmittance mode using a 20 mL, 1 cm path-length cuvette. The absorbance difference between 700 nm and 560 nm was then determined. The same samples were also analyzed with a LabScan (Hunter Associates Laboratory, Reston, VA), which measures reflected rather than transmitted light. From these measurements both the optical densities and CIE a values were determined. All measurements with both instruments were made in triplicate and averaged. Lycopene contents of the homogenates were determined by extraction in hexane:ethanol (3:4) and spectrophotometric quantification (Barrett and Anthon, 2001). RESULTS AND DISCUSSION The ability of the UltraScan to determine lycopene levels was examined by preparing a series of homogenates from both fresh market and processing type tomatoes. Tomatoes were ground in a blender with various amounts of added water to produce homogenates with a range of lycopene contents, as determined by solvent extraction and spectrophotometric quantification. Following the procedure of Davis et al. (2003), the absorbance difference between 560 and 700 nm (A560-A700) was then measured in the UltraScan and compared with the lycopene levels (Fig. 1A). A linear relationship between the absorbance difference and the lycopene content was found, in agreement with the results of Davis et al. (2003). The slope of the regression line for homogenates prepared in a blender (0.0269) also agrees with the value reported by Davis et al. (0.030) for fresh tomato homogenates. One difficulty encountered when preparing fresh tomato homogenates in a blender was the presence of air bubbles. Numerous air bubbles, which increase the light scattering and thus the measured absorbance of the sample, are present immediately after blending. To eliminate any effect these would have on the absorbance measurements, samples were allowed to sit for at least an hour between blending and absorbance measurements to allow the bubbles to dissipate. After this amount of time bubbles were no longer visually apparent and the measured absorbance values were stable. To avoid the introduction of bubbles we tried homogenizing the tomatoes with a mortar and pestle rather than a blender. Tomatoes homogenized in this way also showed a linear relationship between the absorbance difference and the lycopene content (Fig. 1B) but the slope of this line is only 0.0179 or 63% of that obtained when a blender was used (Fig. 1A). Apparently, for a given lycopene content in a tomato, the absorbance of a homogenate, measured in the UltraScan, depends on the method used to homogenize the tomato. Factors other than the lycopene content, such as the particle size and other physical properties of the solution, must also exert a substantial effect. The biggest apparent difference between these two sets of samples was the finer particle size of the blender samples. A comparison of the complete absorbance spectra of a sample prepared in a mortar and pestle with that of one prepared in a blender gives some indication as to the origin of the discrepancy. Since blender samples give higher (A560-A700) values for a given amount of lycopene (Fig. 1), it was possible to select two samples, one prepared in a mortar and pestle and one prepared in a blender, that had very different lycopene contents (60.3 and 43.4 mg/L respectively), yet gave similar (A560-A700) values of approximately 1.0. The complete absorbance spectra of the two samples (Fig. 2) confirms that the absorbance difference between 560 nm and 700 nm is nearly the same (1.041 and 1.075) in each. The big difference between the two spectra is that the background absorbance in the blender-prepared sample is higher and rises much more sharply with decreasing wavelengths. This rising baseline indicates that, in this sample, more of the absorbance measured at 560 nm is actually from background absorbance not lycopene. This would give a larger (A560-A700) value and explain the similarity in this value for the two samples despite the difference in lycopene contents. The same samples that were analyzed with the UltraScan were also examined with