Cerebrospinal fluid biomarkers for assessing Huntington disease onset and severity

Abstract

The identification of molecular biomarkers in CSF from individuals affected by Huntington disease may help improve predictions of disease onset, better define disease progression and could facilitate the evaluation of potential therapies. The primary objective of our study was to investigate novel CSF protein candidates and replicate previously reported protein biomarker changes in CSF from Huntington disease mutation carriers and healthy controls. Our secondary objective was to compare the discriminatory potential of individual protein analytes and combinations of CSF protein markers for stratifying individuals based on the severity of Huntington disease. We conducted a hypothesis-driven analysis of 26 pre-specified protein analytes in CSF from 16 manifest Huntington disease subjects, eight premanifest Huntington disease mutation carriers and eight healthy control individuals using parallel-reaction monitoring mass spectrometry. In addition to reproducing reported changes in previously investigated CSF biomarkers (NEFL, PDYN, and PENK), we also identified novel exploratory CSF proteins (C1QB, CNR1, GNAL, IDO1, IGF2, and PPP1R1B) whose levels were altered in Huntington disease mutation carriers and/or across stages of disease. Moreover, we report strong associations of select CSF proteins with clinical measures of disease severity in manifest Huntington disease subjects (C1QB, CNR1, NEFL, PDYN, PPP1R1B, and TTR) and with years to predicted disease onset in premanifest Huntington disease mutation carriers (ALB, C4B, CTSD, IGHG1, and TTR). Using receiver operating characteristic curve analysis, we identified PENK as being the most discriminant CSF protein for stratifying Huntington disease mutation carriers from controls. We also identified exploratory multi-marker CSF protein panels that improved discrimination of premanifest Huntington disease mutation carriers from controls (PENK, ALB and NEFL), early/mid-stage Huntington disease from premanifest mutation carriers (PPP1R1B, TTR, CHI3L1, and CTSD), and late-stage from early/mid-stage Huntington disease (CNR1, PPP1R1B, BDNF, APOE, and IGHG1) compared with individual CSF proteins. In this study, we demonstrate that combinations of CSF proteins can outperform individual markers for stratifying individuals based on Huntington disease mutation status and disease severity. Moreover, we define exploratory multi-marker CSF protein panels that, if validated, may be used to improve the accuracy of disease-onset predictions, complement existing clinical and imaging biomarkers for monitoring the severity of Huntington disease, and potentially for assessing therapeutic response in clinical trials. Additional studies with CSF collected from larger cohorts of Huntington disease mutation carriers are needed to replicate these exploratory findings.

Keywords: CSF; Huntington disease; biomarkers; neurofilament light; proenkephalin.

Graphical abstract

Figure 1

Comparison of CSF protein levels across disease stages. Box and whisker plots comparing normalized CSF protein levels between controls (n = 8), preHD (n = 8), early/mid HD (TFC >5; n = 8), and late HD (TFC <5; n = 8) individuals. Intergroup differences were assessed using ANCOVA including age as a covariate and summary statistics are shown at the top of each plot. Post hoc tests were performed using Tukey’s test to correct for multiple comparisons. (A) C1QB (*P = 0.010 compared with controls, ^##P = 0.008 compared with early/mid HD), (B) CNR1 (^##P = 0.008 compared with early/mid HD), (C) IDO1 (*P = 0.020 compared with controls), (D) IGF2 (*P = 0.012 compared with controls), (E) IGHG1 (**P = 0.004 compared with controls), (F) NEFL (**P = 0.002 compared with controls), (G) PDYN (*P = 0.012, **P = 0.004, ***P = 0.0003 compared with controls) and (H) PENK (*P = 0.012, **P = 0.004, ***P = 0.0002 compared with controls). Individual data points are plotted for each group. Boxes show 25th to 75th percentiles, the central band denotes the median, the plus sign denotes the mean, and the whiskers show the minimum and maximum values. ANCOVA = analysis of covariance; a.u. = arbitrary units; preHD = premanifest Huntington disease.

Figure 2

Discriminatory potential of CSF protein markers for HD. ROC curves showing the top 3 individual CSF proteins with highest discriminatory power for distinguishing (A) HD mutation carriers from controls, (B) preHD from controls and (C) manHD from preHD groups. sPLS-DA models for assessing the discriminatory potential of all 26 CSF markers for distinguishing (D–F) HD mutation carriers from controls, (G–I) preHD from controls and (J–L) manHD from preHD groups. Two-dimensional score plots showing segregation of (D) controls and HD mutation carriers , (G) controls and preHD, and (J) preHD and manHD. The top 10 CSF proteins ranked based on their relative importance for discriminating (E) HD mutation carriers from controls, (H) preHD from controls and (K) manHD from preHD disease. Grey bars represent the percentage of events a variable was selected by sPLS-DA in the bootstrapped samples. ROC curves generated using sPLS-DA models showing discriminatory ability to distinguish (F) HD mutation carriers from controls, (I) preHD from controls and (L) manHD from preHD individuals. Age-adjusted values were used for all analyses. AUC = area under the curve; manHD = manifest Huntington disease; preHD = premanifest Huntington disease; ROC = receiver operating characteristic; sPLS-DA = sparse partial least squares discriminant analysis.

Figure 3

Multi-marker CSF protein panels for stratifying subjects based on HD mutation status and disease severity. ROC curves showing individual and combinations of CSF proteins with the greatest discriminatory accuracy for distinguishing between (A) HD mutation carriers and controls, (B) preHD and controls, (C) manHD and preHD, (D) early/mid HD and preHD, (E) late HD and early/mid HD. Age-adjusted values were used for all analyses. AUC = area under the curve; manHD = manifest Huntington disease; preHD = premanifest Huntington disease; ROC = receiver operating characteristic.