A diagnostic classifier for gene expression-based identification of early Lyme disease

Abstract

Background: Lyme disease is a tick-borne illness that causes an estimated 476,000 infections annually in the United States. New diagnostic tests are urgently needed, as existing antibody-based assays lack sufficient sensitivity and specificity.

Methods: Here we perform transcriptome profiling by RNA sequencing (RNA-Seq), targeted RNA-Seq, and/or machine learning-based classification of 263 peripheral blood mononuclear cell samples from 218 subjects, including 94 early Lyme disease patients, 48 uninfected control subjects, and 57 patients with other infections (influenza, bacteremia, or tuberculosis). Differentially expressed genes among the 25,278 in the reference database are selected based on ≥1.5-fold change, ≤0.05 p value, and ≤0.001 false-discovery rate cutoffs. After gene selection using a k-nearest neighbor algorithm, the comparative performance of ten different classifier models is evaluated using machine learning.

Results: We identify a 31-gene Lyme disease classifier (LDC) panel that can discriminate between early Lyme patients and controls, with 23 genes (74.2%) that have previously been described in association with clinical investigations of Lyme disease patients or in vitro cell culture and rodent studies of Borrelia burgdorferi infection. Evaluation of the LDC using an independent test set of samples from 63 subjects yields an overall sensitivity of 90.0%, specificity of 100%, and accuracy of 95.2%. The LDC test is positive in 85.7% of seronegative patients and found to persist for ≥3 weeks in 9 of 12 (75%) patients.

Conclusions: These results highlight the potential clinical utility of a gene expression classifier for diagnosis of early Lyme disease, including in patients negative by conventional serologic testing.

Keywords: Bacterial host response; Bacterial infection; Diagnostic markers; RNA sequencing; Targeted resequencing.

Fig. 1. Flowchart of the approach used to develop and validate a 31-gene Lyme disease classifier panel for identification of early Lyme disease.

DEGs differentially expressed genes, KNNXV k-nearest neighbor cross-validation, TREx targeted RNA expression sequencing.

Fig. 2. A 31-gene Lyme disease classifier derived using the generalized linear model machine learning algorithm.

A Chart of misclassification error versus number of genes considered and related log(lambda) statistic. B Receiver-operating characteristic (ROC) curve of the performance of the LDC using a training set of 44 Lyme seropositive and 93 “non-Lyme” control samples. The cutoff for positivity according to Youden’s J statistic is 0.3. C Violin plots of the LDC score for an independent test set of 63 samples. D Violin plots of the LDC score for the training set of 137 samples. E 2 × 2 contingency tables of LDC test set performance overall and for seropositive (serologically confirmed) and seronegative Lyme cases. F Pie chart of signaling pathways associated with the 31 genes comprising the LDC panel.

Fig. 3. Longitudinal testing of clinical Lyme patients using the Lyme disease classifier.

A comparison between the LDC score and results from two-tiered Lyme serologic testing for Lyme seronegative and Lyme seropositive (both early and late seroconversion) patients at 0 and 3 weeks. Patients testing Lyme seropositive at 0 week did not get repeat serologic testing. CDC criteria for a positive Lyme serology include a positive screening ELISA and either ≥2 of 3 bands on reflex IgM testing (in patients with signs and symptoms lasting <30 days) or ≥5 of 10 bands on reflex IgG testing. LDC Lyme disease classifier, ELISA enzyme-linked immunosorbent assay, WB western blot, IgM immunoglobulin M, IgG immunoglobulin G.

Fig. 4. Lyme disease classifier scores from longitudinally collected patient samples.

Plots of the LDC score in 18 Lyme disease patients from available samples collected at 0 week, 3 weeks, and 6 months. An LDC result is considered positive if the LDC score is above the 0.3 cutoff as determined from training set data using Youden’s J statistic. Patients are labeled P1–P18.