Micro-Raman spectroscopy study of blood samples from myocardial infarction patients

Blood samples are well suited for any disease diagnosis, due to its easy access, it transports oxygen, carbon dioxide, proteins, hormones, toxins, amino acids, metabolic products, nutrients etc. throughout the body and hence has the ability to reflect the biochemical changes accompanying with the progress of any disease [21]. Raman spectroscopic technique has the ability to provide spectral fingerprints of biomolecular changes during the occurrence of any disease which originated from the vibrational frequencies of molecular bonds within molecule/molecules. This technique has been of paramount importance in providing valuable insights to clinicians in disease diagnosis and prognosis of therapy [22, 23]. Averaged Raman spectra of whole blood (Fig. 2a) and plasma (Fig. 2b) collected from MI(BM) and MI(AM) patients and healthy volunteers are plotted in Fig. 2, where, each spectrum is an average of 60 spectra recorded from the different sample categories (MI(BM),MI(AM) and clinically Normal).

Significant spectral variations were observed in the Raman spectrum that correspond to oxygenation and deoxygenation characteristic markers of whole blood from MI patients with respect to control samples. Raman peaks at 1211 cm⁻¹ and 1224 cm⁻¹ arising from C-H methine deformation were regarded as the standard markers for hemoglobin oxygenation and deoxygenation states respectively. Another peak observed at 1634 cm⁻¹ was also reported as a good representative of oxygenated hemoglobin. Similarly, spin marker region also holds interest in blood investigations due to the presence of oxy (1546 cm⁻¹) and deoxyhemoglobin (1563 cm⁻¹) Raman peaks as indicators. In addition, the hemoglobin oxygenation state can also be evaluated by monitoring the spectral signatures of pyrrole stretch frequencies (1375 cm⁻¹ and 1398 cm⁻¹) and the Fe-O₂ stretch frequency(568 cm⁻¹) [24, 25]. The assignments of the Raman peaks of the whole blood sample that had undergone changes are given in Table 1.

The patients were admitted to the emergency department along with oxygen supplementation. The medical oxygen support has been continued during the treatment period when the patients were undergoing treatments in the ICU. Data given here for the normal samples were collected from healthy volunteers with 25–40 years of age group. It is clearly evident from Fig. 2 that all the oxygenation characteristic peaks were shown intensity enhancement in Raman spectra of MI patient in comparison with the control samples. Studies have shown that oxygen supply through face mask can elevate the oxygen level in the blood. As the viscosity of the blood decreases during the treatment, the resistance to blood flow reduces with overall effect of increased blood flow to and from the heart, resulting in an increased oxygen supply [26, 27]. In the present study, the MI samples have been collected from the patients who were already under external oxygen supply through facial mask as a part of the emergency treatment. This may be the causative factor for the increase in hemoglobin oxygenation found in case of MI blood samples.

The plasma Raman spectra of MI and Normal samples showed significant variations in terms of intensity of spectral band, which may be possibly due to changes in biological components in blood plasma. The assignments of the Raman peaks of the plasma sample showing changes are presented in Table 2. The increase in Raman band 898 cm⁻¹ in MI sample is assigned to C–O–C starching. The Raman spectral band corresponding to 754 cm⁻¹ and 1264 cm⁻¹ corresponds to protein and amide III respectively, and is found to be increasing in case of MI to that of normal sample. The enhancement in the Raman peak at 1318 cm⁻¹ (guanine) and at 1337 cm⁻¹ (guanine and adenine) corresponding to DNA base pairs is observed in case of MI samples [28]. The increase in the concentration of cell free DNA in plasma is suggested to be a marker for MI. The cell free DNA originates as a result of cellular injury and it explains the degree of cell damage [29, 30]. The rise in Raman intensity of the peaks at 742 cm⁻¹ and 1447 cm⁻¹, corresponding to phospholipids is observed in MI samples. The decrease in intensity of Raman signal at 1404 cm⁻¹ corresponding to glutathione is also seen in MI samples [31]. Glutathione is an important anti-oxidant and its role in oxygen radical scavenging thus preventing the cellular damage has been well established in earlier studies. The lower plasma level of glutathione, in various clinical conditions such as CVD, diabetes, arthritis and malignancies suggests the defensive role of glutathione against cellular damages [32].

The principal component analysis (PCA) was carried to visualize variation in MI and normal Raman plasma spectra. The Figure shows the Score 1 versus Score 2 plot of the plasma Raman spectra, where Score 1 covers 53.93% of variance and Score 2 having 63.74% of variance. The PCA results convey clear discrimination between the MI (BM) and MI (AM)) and Normal samples. The PCA results consolidate, relative changes observed in the Raman spectra of two categories of samples.

Phe-Tyr ratio has been regarded as a vital indicator for inflammatory marker and immune activation in clinically abnormal patients. The inflammation and immune activation hinders the effective hydroxylation of phenylalanine to tyrosine in individuals suffering from cardiac disorders. In view of this, intensity of Raman peaks at 825 cm⁻¹ and 1000 cm⁻¹ that corresponds to tyrosine and phenylalanine respectively has been estimated. The Phe-Tyr ratio, peak intensities obtained for normal individuals and MI patients are given in Table 3 and Table 4. The bar plots obtained by calculating Phe-Tyr ratio (1000 cm⁻¹/825 cm⁻¹) for normal, MI (BM) and MI (AM) patients from the Raman spectra of blood and plasma are shown in Fig. 3a and b. As evident from the figure, the MI patients exhibited an increase in Phe-Tyr ratio with respect to healthy controls. In addition, the MI (AM) group has shown a decrease in this ratio as compared to MI (BM) samples. As mentioned earlier, previous literatures have linked higher Phe-Tyr ratio with the immune activation marker neopterin and C-reactive protein (CRP). Higher neopterin concentration in blood serum is correlated with an increased mortality rate in cardiovascular disease patients [33, 34]. The higher Phe-Tyr ratio indicates abnormal PAH enzyme activity, which is responsible for hydroxylation of phenylalanine to tyrosine. The PAH enzymatic activity is affected due to the deficiency in BH4 cofactor [13, 14]. The higher Phe-Tyr ratio observed in case of BM samples in whole blood and plasma indicates the scarcity of BH4 availability. The lower BH4 availability not only effects the PAH activity but also other codependent process namely, glyceryl ether monooxygenase and nitric oxide synthases (NOS) [35‐37].

The pictorial representation showing the correlation between the Phe-Tyr ratio with inflammation and immune activation is illustrated in Fig. 4. The preliminary step in the production of BH4 is through enzyme GTP-cyclohydrolase I (GCH). The stimulations of proinflammatory stimuli’s, for instance cytokine tumor necrosis factor α (TNF—α) and lipopolysaccharide (LPS) on GCH causes the production of immune activation marker, neopterin in place of BH4. Literatures have also reported that the anti-inflammatory drugs such as salicylic acid, aspirin etc. suppress the neopterin level, thus altering the BH4 availability [38‐40]. This may be the reason for the decrease in Phe/Tyr ratio observed in AM patients with respect to the BM sample. In brief, the increase in Phe-Tyr ratio observed in case of BM samples depicts possible inflammatory and immune activation process in myocardial infarction (MI). Similarly, the reduction in inflammation due to medication to patients may be the probable reason behind the decrease in the ratio.

The principal component analysis (PCA) was also performed to visualize the classification of the three groups of samples (Normal, MI (BM) and MI (AM)) as different by selecting the Raman spectral region corresponding to tyrosine (810–840 cm⁻¹) and phenylalanine (990–1020 cm⁻¹) using Unscrambler X software. Figure 5 shows the Score of Factor 1 versus Score of Factor 2 for whole blood (Fig. 5a) and plasma (Fig. 5b) samples. The scatter plot shows fairly good discrimination between MI (BM) samples and Normal samples with minimal overlap. Whereas significant overlap of samples was observed between MI (AM) samples and Normal samples. The observation strongly supports the earlier findings that there exists significant variation in phenylalanine and tyrosine concentrations in all the three categories of samples along with the tendency of mixing of MI(AM) samples with Normal resulting from the reduction in the inflammatory condition (lower Phe-Tyr ratio) while the MI patient is undergoing treatment.

The receiver operating characteristic (ROC) curve is plotted to evaluate the success of the diagnostic method. It gives the relation between sensitivity and specificity of a particular diagnostic test and also the optimum threshold value [41, 42]. Figure 5c and d shows ROC curve for whole blood and plasma samples obtained by considering Raman intensity ratio of phenylalanine to tyrosine. The optimum threshold value of Phe-Tyr ratio for whole blood and plasma samples are found to be 4.19 and 2.6 respectively. The area under the curve (AOC) for the ROC for whole blood and plasma samples is observed to be 100% and 98% respectively. The ROC curve for whole blood and plasma shows good diagnosis outcome indicating the success of selecting Raman intensity ratio of peaks corresponding to phenylalanine and tyrosine as a measure of inflammatory condition. The ROC curve also predict reliability of method in clinical use and the result obtained highlights the capability of the Raman spectroscopy analysis of whole blood and plasma in the diagnosis of MI.

Whole blood and plasma Raman spectra were subjected to Match/No match test. Procedure for Match/No match test involves a calibration set. In this case, the calibration set consists of thirty clinically certified MI sample Raman spectra. The spectra of the calibration set is then subjected to PCA to determine the parameters which are highly specific to the MI sample Raman spectra. For Match/ No match test, three parameters were considered namely factors, spectral residuals and M-distances. In PCA, initial factors cover maximum variation, in this case factor 3 gave good results for whole blood and plasma samples. The sum of difference between original spectrum and predicted spectrum from factors gives spectral residual [43]. The distance between the test sample point and the mean of all the other points in the class is determined as the M- distance, where distance is measured in terms of standard deviation in all dimensions for range of variation in a class.

M- distance matrix is given by,

M- distance, D for any test sample is given by

M is an f x f M- distance matrix and S is an n x f matrix of calibration set sample’s PCA scores, where n & f are number of samples and PCA scores respectively [44]. The result of the Match/No match test is shown in Table 5. In whole blood and plasma samples up to sample number 17 showed Match with the MI calibration set, whereas three samples gave No-Match leading to false result. Similarly, in normal samples, samples up to 19 had No Match with MI calibration set, but one sample Matched with MI calibration set in whole blood sample, whereas in plasma sample, all twenty samples had No Match with MI calibration set. The Match/No Match test results showed the same 85% sensitivity for whole blood and plasma samples, while, specificity is observed to be 95% and 100% for whole blood and plasma samples, respectively.

Direct detection of BH4 in blood sample demands complex preanalytical procedures which limits its use in clinical application [39, 45]. In view of this, the Phe-Tyr ratio can be used as an indirect marker of BH4 availability, since it regulates the PAH enzyme activity. This indicates the diagnostic importance of probing Phe-Tyr ratio to evaluate the inflammation process occurring in cardiac patients. This ratio can also be exploited for evaluating the effectiveness of anti-inflammatory drugs on subjects by acquiring Raman spectra of blood samples with minimal sample preparation. Prospective rigorous validation using more number of clinical samples is necessary to establish Raman-spectroscopy as a rapid and objective tool for inflammation monitoring in routine clinical usage.