A Simplified Diagnostic Classification Scheme of Chemotherapy-Induced Peripheral Neuropathy. (29th January 2020)

• Record Type: Journal Article
• Authors: Huang, Han-Wei; Wu, Pei-Ying; Su, Pei-Fang; Li, Chung-I; Yeh, Yu-Min; Lin, Peng-Chan; Hsu, Keng-Fu; Shen, Meng-Ru; Chang, Jang-Yang; Lin, Chou-Ching K.
• Other Names: Peluso Marco E. M. Academic Editor.
• Is Part Of: Disease markers. Volume 2020(2020)
• Publication Date: 2020-01-29
• Subjects: Diagnosis -- Periodicals; Biochemical markers -- Periodicals; Pathology -- Periodicals; 616
• DOI: 10.1155/2020/3402108
• Languages: English
• ISSNs: 0278-0240

Abstract

Abstract : Background and Objective . The main purpose of this study was to develop a simple automatic diagnostic classification scheme for chemotherapy-induced peripheral neuropathy. Methods . This was a prospective cohort study that enrolled patients with colorectal or gynecologic cancer post chemotherapy for more than 1 year. The patients underwent laboratory examinations (nerve conduction studies and quantitative sensory tests), and a questionnaire about the quality of life. An unsupervised classification algorithm was used to classify the patients into groups using a small number of variables derived from the laboratory tests. A panel of five neurologists also diagnosed the types of neuropathies according to the laboratory tests. The results by the unsupervised classification algorithm and the neurologists were compared. Results . The neurologists' diagnoses showed much higher rates of entrapment syndromes (66.1%) and radiculopathies (55.1%) than polyneuropathy (motor/sensory: 33.1%/29.7%). A multivariate analysis showed that the questionnaire was not significantly correlated with the results of quantitative sensory tests (r = 0.27 ) or the neurologists' diagnoses (r = 0.2 ). All of the patients were classified into four groups by the unsupervised classification algorithm. The classification corresponded to the severity of neuropathy and correlated well with the neurologists' diagnoses and the scales of neurological examinations. The overall correct rate of classification by the unsupervised classification algorithm was 78.8% (95% confidence interval: 73.1%-88.3%). Conclusion . The results of our unsupervised classification algorithm based on three variables of laboratory tests correlated well with the neurologists' diagnoses.