Table 3.

Summary of selected papers in the literature studying EV biochemical composition with FTIR spectroscopy.

Paper	Sample/Purification	Methodological Consideration	Main IR Findings	Impact and Application
Baddela 2015 [132]	Buffalo’s Milk/ Exoquick	Samples were collected from 3 healthy buffaloes. Band assignment was carried out after averaging 3 spectra.	IR spectra display peculiar absorption bands reflecting exosome composition: (i) 1300–1700 cm−1 (amide I–III) and 2700–3500 cm−1 (CH stretching) for protein and lipids; and (ii) 900–1200 cm−1 for nucleic acids and carbohydrates.	The combined use of IR and miRNA profiles allows for the characterization of bioactive compounds in milk.
Mihály J. 2016 [25]	Jurkat T cells/ Centrifugation	Four replicas of the experiment were carried out. The protein–lipid ratio (P/L) was computed as the ratio between the intensity of the amide I–II (1750–1500 cm−1) and the CH stretching (3040–2700 cm−1). ANOVA was used to compare different EV types	Spectra of EVs and parental cells were compared. ABs’ spectra resemble those of parental cells. The following difference among the diverse EV types were observed in the range 1800–1350 cm−1: (i) a shift in the amide I peaks; and (ii) a change in the relative weight of the amide I and II peaks (Figure 5a). The following P/L ratio was measured (Figure 5a): 0.79 ± 0.05 for EXOs; 0.60 ± 0.04 for MVs and 1.20 ± 0.12 for Abs (P < 0.0001)	FTIR provides an effective tool for the classification of different EV types. Classification is based on the shape of amide I–II bands and the P/L ratio. These results impact EV sample control, a key issue in exosome science.
Lee 2017 [137]	THP-1 cells/ Centrifugation	Three replicates of the experiment were carried out. Comparison among spectra was performed considering the 2nd derivative. PCA loadings were computed to highlight significant spectral changes.	Monocyte activation upon lipopolysaccharide stimulation (LPS) can be inferred from the analysis of released MVs. An increase in the integrated areas of the lipid ester, α-helical protein, and uracil bands upon LPS is observed. Similar spectral changes were detected on monocytes, as confirmed with PCA and PCA loadings.	Spectra of MVs provide biochemical insights into the LPS-induced monocyte model of sepsis. Moreover, IR analysis of MVs is an effective tool to monitor cellular phenotypes.
Pereira al. 2018 [138]	CFPAC-1 Cell line and SR4987	Six subjects were recruited for the study, and bone marrow mesenchymal stromal cells were isolated. Spectra were analyzed using the first and second derivatives, and PCA.	The authors studied the influence of culture and time conditioning on exosomes released from human BM-MSCs. Cells were cultured in different media (DMEM and XenoFree). PCA, 1st, and 2nd derivatives showed that IR signatures are more affected by culture conditions than donor or conditioning days.	This paper highlights the role of the different culture conditions in EXO research, showing that great attention has to be paid to this aspect to assure experimental reproducibility.
Romanò 2020 [139]	HT29 cells/ Exoquick	Ten replicates of the experiments were carried out. PCA–LDA was used to classify exosomes. PCA loadings were employed to highlight the most relevant spectral changes. Sensitivity, specificity, accuracy, and recall were estimated	The authors studied the biochemical changes in EXOs obtained from HT29 cancer cells under different culture conditions (well-fed and starved cells). Differences in the spectral shape of the amide I–II bands can be used to classify exosomes extracted from the two groups using PCA–LDA. Classification has very high accuracy, precision, and recall, especially in the amide I and II regions.	FTIR combined with PCA–LDA allows for the automated classification of EXOs derived from cells cultured under different conditions. Most importantly, FTIR spectroscopy on exosomes provides information on the cellular state.
Pascucci 2014 [134]	CFPAC-1/ Centrifugation	Model-free band assignment (see Figure 5d).	The authors applied FTIR to characterize MVs derived from bone marrow mesenchymal cells. MVs were loaded with PTX, an anticancer molecule. Drug loading induces changes in MV spectra between 3000 and 2800 cm−1. These spectral changes show specific features observed in the PTX spectra.	Label-free characterization of EVs with FTIR can provide a quick and effective way of controlling exosome-based nano-cages for drug delivery applications.
Zlotorogski-Hurvitz 2019 [136]	Saliva/ Centrifugation	A total of 21 patients diagnosed with oral cancer (OC) and 13 donors (D) were recruited. Machine learning (ML) techniques (PCA–LDA and support vector machine) were used to classify exosomes. Classification performance was evaluated with ROC curves.	The authors highlighted a significant difference in IR spectra between OC and D at 1072 cm−1 (nucleic acids), 2924 cm−1 and 2854 cm−1 (membranous lipids), and 1543 cm−1 (transmembrane proteins). The difference was highlighted through relative intensity ratios. An ML-based classification model showed a sensitivity of 100%, specificity of 89%, and accuracy of 95%.	The paper first validates, in a complex clinical setting, a liquid biopsy approach based on the IR characterization of exosomes.
Martins 2020 [135]	Serum/ Exoquick	Two cohorts of patients were recruited, with a total of 21 AD patients and 21 controls. The 2nd derivative of FTIR spectra was calculated and a multivariate (PCA, LDA and QDA) and univariate (Mann–Whitney test) analyses were carried out.	EXOs have higher absorbance than serum spectra in the lipid regions (3000–2800 cm–1 and 1483–1423 cm–1) and the nucleic acids/carbohydrates regions (1200–900 cm–1). A multivariate analysis based on 2nd derivative spectra, PCA, LDA, and QDA shows that serum-derived exosomes have better discriminatory properties than serum. A significant difference among the two groups and in both cohorts is measured at 1064 cm−1, a peak assigned to ester C–O–C symmetric stretching of phospholipids and/or ribose C–O stretching (nucleic acids).	A key paper providing clinical validation of an exosome-based liquid biopsy approach for AD diagnosis. This study has wide application in diagnostics, because a blood test for AD is still lacking, despite the large research effort in this field.