Background: Metagenomic next -generation sequencing (mNGS) is valuable for pathogen identification; however, distinguishing between infectious diseases and conditions with potentially similar clinical manifestations, including malignant tumors, is challenging. Therefore, we developed a method for simultaneous detection of infectious pathogens and cancer in blood samples.
Methods: Plasma samples (n = 244) were collected from 150 and 94 patients with infections and hematological malignancies, respectively, and analyzed by mNGS for pathogen detection, alongside human tumor chromosomal copy number variation (CNV) analysis ( >= 5Mbp or 10Mbp CNV region). Further, an evaluation set, comprising 87 plasma samples, was analyzed by mNGS and human CNV analysis, to validate the feasibility of the method. Results: Among 94 patients with hematological malignancy, sensitivity values of CNV detection for tumor diagnosis were 69.15 % and 32.98 % for CNV region 5Mbp and 10Mbp, respectively, with corresponding specificities of 92.62 % and 100 % in the infection group. Area under the ROC curve (AUC) values for 5Mbp and 10Mbp region were 0.825 and 0.665, respectively, which was a significant difference of 0.160 (95 % CI: 0.110 -0.210; p < 0.001), highlighting the superiority of 5Mbp output region data. Six patients with high -risk CNV results were identified in the validation study: three with history of tumor treatment, two eventually newly -diagnosed with hematological malignancies, and one with indeterminate final diagnosis.
Conclusions: Concurrent CNV analysis alongside mNGS for infection diagnosis is promising for detecting malignant tumors. We recommend adopting a CNV region of 10Mbp over 5Mbp for our model, because of the lower false -positive rate (FPR).