Discrimination of subtypes and stages of non-Hodgkin lymphomas using serum laser-induced breakdown spectroscopy†

Abstract

Non-Hodgkin lymphomas (NHLs) are a major group of lymphomas. Accurate diagnosis of subtypes and stages of NHLs plays an important role in clinical practice for precise treatment and prognosis evaluation of the disease. In this work, serum laser-induced breakdown spectroscopy (serum LIBS) was applied to discriminate two major subtypes of NHLs, namely, diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (FL). Serum samples from confirmed patients with DLBCLs and FLs, and healthy normal controls (NCs) were collected for LIBS measurements. For DLBCLs, both germinal center B-cell (GBC) and activated B-cell (ABC) originated subgroups were included, each subgroup containing samples at different progressive stages. The serum samples were deposited onto silicon substrates, naturally dried, and ablated with a focused Nd:YAG laser to generate laser-induced plasmas. The spectral intensities of typical elemental emissions from the plasmas were compared among DLBCLs, FLs and NCs, between the GBC- and ABC-originated DLBCL subgroups, and among different stages. The k nearest neighbors and support vector machine classifiers were used to discriminate the two subtypes of NHLs, to classify the originating B-cell subgroups of DLBCLs, and to stage within the GBC- and ABC-originated DLBCL subgroups. Accuracy of about 90% was achieved for discrimination of DLBCLs and FLs, about 88% for classification of the GBC- and ABC-originated DLBCLs, and about 92% and 85% for staging within the GBC- and ABC-originated DLBCL subgroups, respectively. The results show that the serum LIBS supported by chemometric methods could be a potential solution for discrimination of subtypes of NHLs, and for diagnosis and staging of DLBCLs. The serum LIBS may help to diagnose the NHLs in a minimally invasive, rapid and cost-effective way, and to provide valuable references for precise treatment and prognosis prediction of the disease.