Differentiating Between Malignant Mesothelioma and Other Pleural Lesions Using Fourier Transform Infrared Spectroscopy

Our plain-English summary of Differentiating Between Malignant Mesothelioma and Other Pleural Lesions Using Fourier Transform Infrared Spectroscopy, by F. Sadiku-Zehri, O. Gamulin, M. Škrabić, et al (Applied Spectroscopy, 2020).

The study

• Authors: F. Sadiku-Zehri, O. Gamulin, M. Škrabić, A. Qerimi-Krasniqi, F. Sedlić, A. Šepac, L. Brčić, L.B. Vuletić, S. Seiwerth
• Journal: Applied Spectroscopy
• Published: 2020;74(7):808-818
• Cited by: 1 article, as counted by NIH iCite
• Read it: PubMed · Publisher

Abstract

Histopathology, despite being the gold standard as a diagnostic tool, does not always provide a correct diagnosis for different pleural lesions. Although great progress was made in this field, the problem to differentiate between reactive and malignant pleural lesions still stimulates the search for additional diagnostic tools. Our research using vibrational spectroscopy and principal component analysis (PCA) statistical modeling represents a potentially useful tool to approach the problem. The objective method this paper explores is based on the correlation between different types of pleural lesions and their vibrational spectra. Obtained tissue spectra recorded by infrared spectroscopy allowed us to categorize spectra in different groups using a created PCA statistical model. The PCA model was built using tissues of known pathology as the model group. The validation samples were then used to confirm the functionality of our PCA model. Student's t-test was also used for comparing samples in paired groups. The PCA model was able to clearly differentiate the spectra of mesothelioma, metastasis and reactive changes (inflammation), and place them in discrete groups. Thus, we showed that Fourier transform infrared spectroscopy combined with PCA can differentiate pleural lesions with high sensitivity and specificity. This new approach could contribute in objectively differentiating specific pleural lesions, thus helping pathologists to better diagnose difficult pleural samples but also could shed additional light into the biology of malignant pleural mesothelioma.

In plain English

This laboratory study tested whether infrared spectroscopy (FT-IR) combined with a statistical pattern method (principal component analysis, PCA) could tell malignant pleural mesothelioma apart from other pleural lesions. The authors used archived tissue from 32 patients at the University of Zagreb: 10 with mesothelioma, 10 with adenocarcinoma that had spread to the pleura from the lung, colon or kidney, 10 with pleural inflammation and 2 with normal pleura, recording 670 spectra from tissue sections. Asbestos exposure was not studied. A model built from 150 spectra per disease group was tested on the remaining 60 per group. The authors estimated that more than 95% of these test spectra were correctly identified, with no overlap between mesothelioma and the other two groups. The authors stated that sections from the same patients were used both to build and to test the model, raising the chance of false positive results, and that a larger patient group is needed before clinical use. The authors concluded that the method could be a valuable tool in mesothelioma diagnosis.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Laboratory diagnostic-method study on archived tissue samples, comparing infrared spectra of different pleural lesions using Student's t-test and principal component analysis (PCA), with validation of the model and two additional classification methods (PLS-DA and k-nearest neighbor)
• Who was studied: Archived paraffin tissue blocks from 32 different patients, from the Department of Pathology, University of Zagreb School of Medicine (Croatia): malignant mesothelioma of the pleura (10 patients), pleural inflammation (pleuritis, 10), adenocarcinoma metastatic to pleura (10; originating from lung, colon and kidney clear cell) and normal pleura (2). A total of 670 sections were used for infrared spectroscopy. The article does not give the years the samples came from.
• Exposure: None. The study did not examine asbestos or any other exposure. It tested a diagnostic method (FT-IR spectroscopy with PCA) on tissue samples.
• Main results:
  • The 10 mesothelioma samples were 6 epithelioid, 3 sarcomatoid and 1 biphasic type
  • The 10 metastasis samples came from lung, colon and kidney (clear cell). The text gives 7 from lung, 1 from colon and 2 kidney clear cell; Table II gives different colon and clear-cell counts.
  • Average spectra of the four tissue groups looked very similar. Student's t-test (significance level p less than 0.01) on the six pairs of groups found some statistically significant differences between the spectra of the different tissue types.
  • In a PCA model using all 670 spectra, spectra from mesothelioma, metastasis, inflammation and normal pleura fell into distinguishable groups, with a few spectra overlapping or lying close to other groups
  • For validation, normal pleura was left out. A model was built from 150 randomly chosen spectra from each of the mesothelioma, metastasis and inflammation groups, and the remaining 60 spectra per group (180 in total) were used to test it.
  • The authors estimated from their graph that the model correctly identified more than 95% of the validation spectra
  • Validation spectra from the mesothelioma group did not overlap with the metastasis or inflammation groups. There was a slight overlap between the inflammation and metastasis validation groups.
  • More than 90% of spectra from the inflammation group fell inside the inflammation region. This group was more widely spread than the others.
  • When the test set was instead made from the spectra of two random patients per tissue group, the authors reported almost the same results but did not show the data
  • Two other classification methods were also tested; the article does not say which tissue each 'class' is. In the confusion tables, PLS-DA correctly classified 205, 208 and 206 spectra in classes 1, 2 and 3, and KNN correctly classified 189, 153 and 194. The authors described both as showing extremely high accuracy.
• The authors' conclusion: The authors concluded that validation showed the PCA model's accuracy was higher than 95%, and that FT-IR spectroscopy combined with PCA could be a valuable tool for the difficult problem of mesothelioma diagnosis. They added that more spectra recorded with more precise protocols are needed to build a reliable clinical model.
• Limitations the authors state:
  • Sections from the same patients were used in both the training set and the testing set, which increases the possibility of false positive results
  • Clinical use would require a model built with tissue spectra from more patients; the authors plan to test the model on a larger patient group
  • More spectra recorded with more precise recording protocols are needed to build a reliable clinical model
  • Some spectra overlapped between tissue groups, which the authors attributed probably to small differences in recording conditions or sampling procedure
  • The model used the whole infrared spectrum, although not all parts of the spectrum contribute equally to the differences

Topics

The National Library of Medicine indexes this study under subjects our site covers: Mesothelioma.

Medical terms in this study, explained in our Lexicon: Malignant Mesothelioma, Infrared Spectroscopy (IR), Histopathology, Pathology, Mesothelioma, Metastasis, Inflammation, Pleural Mesothelioma.

Related studies

• Quantitative analysis of asbestos in drinking water and its migration in mice using fourier-transform infrared spectroscopy and inductively coupled plasma optical emission spectrometry — B. Zheng, L. Zang, et al. (2019)
• Diffuse Reflectance Infrared Fourier Transform Spectroscopy for the Determination of Asbestos Species in Bulk Building Materials — G. Accardo, R. Cioffi, et al. (2014)
• Chemical differences between long and short amosite asbestos: differences in oxidation state and coordination sites of iron, detected by infrared spectroscopy — A. Graham, J. Higinbotham, et al. (1999)
• Determination of low levels of free fibres of chrysotile in contaminated soils by X-ray diffraction and FTIR spectroscopy — E. Foresti, M. Gazzano, et al. (2003)
• Infrared and nuclear magnetic resonance studies of some surface properties of asbestos-albumin interactions — R. Dumitru-Stănescu, C. Mandravel, et al. (1994)
• Persistent Pleural Lesions and Inflammation by Pulmonary Exposure of Multiwalled Carbon Nanotubes — D. Liao, Q. Wang, et al. (2018)

Browse return to medical research.