Figure 32. Difference spectrum obtained by subtracting the average pyrolysis mass spectrum of normal muscle samples from the average spectrum of dystrophic muscle samples.
hydroxyproline (9%) (see Atlas). Indeed, the relative contribution of collagen, an important constituent of connective tissue, strongly increases with progressive wasting of the muscle tissues in DMD.
The negative part of the difference spectrum appears to contain different components. The typical protein-derived peaks at m/z 34 (HgS), 48 (CHgSH) and 117 (probably indole) seem to point to a relative decrease in a protein component rich in cysteine, methionine and tryptophan. Although it is tempting to associate this with the strong relative decrease in muscle proteins, e.g. myoglobin, in DMD, the situation is far less clear than with collagen. Also, it should be kept in mind that the presence of an abnormal positive component tends to lead to relatively low signals for all normal components present in spite of the elimination of m/z 67 during the normalisation procedure. Indeed the negative part of the difference spectrum shows a rough resemblance to the spectrum of normal muscle tissue in Figure 25. Finally, the presence of a typical series of carbohydrate peaks at m/z 31, 32, 43, 70, 84, and 85 merits attention. When examining the intensity distributions of these peaks, a marked degree of inter-individual variation is noted, especially among controls. The DMD cases show relatively low intensities, as evidenced by the appearance of these masses in the negative part of the difference spectrum. This tendency is shown in the intensity distribution of m/z 32 in Figure 27. The strong inter-individual variation points to a carbohydrate component present in varying concentrations. Thus, it is tempting to identify this component as muscle glycogen, the concentration of which will be strongly dependent on the physiological status of the muscle prior to the biopsy. Of course, many more samples would have to be analysed before such conclusions could be confirmed. Therefore, these data are given primarily as an example of how to attempt a limited degree of chemical interpretation with complex samples. Such a limited degree of interpretation can be extremely helpful in pointing the way to more specialised biochemical studies. Also, once the identity of a given pattern has been firmly established, e.g. the supposed collagen pattern, then the pyrolysis mass spectrum may provide a unique way of obtaining a quantitative estimate of the amount of the component present in further samples.
Apart from performing simply subtractions, computer techniques can be used for more sophisticated support of attempted chemical interpretation. For instance, if doubts exist as to the significance of features discovered in difference spectra, the computer may be used to calculate all the variances involved in order to arrive at an estimate of statistical significance of the features shown.
A different approach is necessary when more than two classes of samples are compared and the differences between these classes are caused by multiple components. Using factor analysis techniques, the contribution of each of the different components
(factors) to each of the different sample classes can be determined (ref. 126, 127), provided the number of different spectra available is several times larger than the number of components involved. In many ways, this requirement is similar to the requirement for solving equations with multiple unknowns, namely that the number of different equations available should be equal to or greater than the number of unknowns. Apart from determining the contribution of each factor to each class of spectra, factor analysis procedures also allow the determination of the contribution of each mass peak to each factor. The resulting factor spectra can be regarded as characteristic of the corresponding components. If sufficiently large numbers of observations are available, the number of factors involved is fairly low (e.g. <10), and the components do not mutually interfere in the analytical procedure, the factor spectra should show a close resemblance to the spectra of the individual compounds. Thus far, experience with factor analysis in pyrolysis mass spectrometry is limited. Figure 33 shows the factor spectrum calculated for the main factor derived from the series of pyrolysis mass spectra of DMD muscle samples described in the earlier sections of this Chapter. The factor analysis program used was part of the SPSS package (ref. 128). Windig et al. (ref. 101) used factor analysis techniques for qualitative comparison of pyrolysis mass spectra of standard biopolymers on changing the pyrolysis parameters (see Section 5.3). Van Graas et al. (ref. 73) demonstrated that from the pyrolysis mass spectra of coals of different coalification stages a factor can be calculated which is strongly correlated with the rank of the coal.
Was this article helpful?