VGF in cerebrospinal fluid, when combined with conventional biomarkers, enhances prediction of conversion from mild cognitive impairment to Alzheimer’s Disease

Sensitive and accurate biomarkers for the prediction of conversion from mild cognitive impairment (MCI) to Alzheimer’s Disease (AD) are needed to both support clinical care and enhance clinical trial design. Here, we examined the potential of cerebrospinal fluid (CSF) levels of a peptide derived from a neural protein involved in synaptic transmission, VGF (a non-initialism), to enhance accuracy of prediction of conversion from MCI to AD. The performance of conventional biomarkers (CSF Aβ1-42 and phosphorylated tau +/− hippocampal volume) was compared to the same biomarkers with CSF VGF peptide levels. It was observed that VGF peptides are lowered in patients with AD compared to controls and that combinations of CSF Aβ1-42 and phosphorylated tau, hippocampal volume and VGF peptide levels outperformed conventional biomarkers alone (hazard ratio = 2.2 vs. 3.9). VGF peptide levels were correlated most strongly with total tau levels, but not hippocampal volume, suggesting that they serve as a marker for neuronal degradation, but not necessarily in the hippocampus. The latter point suggests that VGF may serve as a more general marker of neurodegeneration. Future work will be needed to determine the specificity of VGF for AD vs. other neurodegenerative diseases.


65
Alzheimer Disease (AD) is characterized by a long prodromal course during which a number of 66 pathological changes occur prior to the onset of clinical symptoms. Classically, these changes 67 include the deposition of amyloid beta (Aβ) and phosphorylated tau (pTau) into the brain,  AD biomarkers may be used to 1) achieve earlier diagnoses for patients, 2) predict which 76 individuals are most likely to clinically worsen over time, 3) help to identify and stratify subjects 77 enrolling in AD-related clinical trials and 4) serve as outcome measurements in AD-related 78 clinical trials [5][6][7]. For example, there is a 10-15% misdiagnosis rate when AD is diagnosed on 79 clinical grounds only. This high rate of misdiagnosis has substantial cost implications [8][9][10][11] and 80 if such misdiagnosed subjects are enrolled into clinical trials, they could obscure the impact of 81 disease-modifying therapy. In addition, prediction of clinical decline in subjects with early-stage 82 disease will permit the institution of aggressive interventions, such as physical exercise or 83 pharmacologic therapy, to stave off AD symptoms. Finally, novel biomarkers or combinations of 84 biomarkers could be used to enrich MCI clinical trials with subjects with high conversion rates to 85 shorten and diminish the cost of clinical trials [12,13]. Therefore, a better understanding of how 86 biomarkers delineate disease classes and predict progression is needed. shown to be useful in the prediction of MCI to AD conversion [14][15][16][17][18][19][20]. For example, we used a 91 4 hypothesis-free bioinformatics approach to identify a panel of 16 peptides in CSF initially 92 identified as showing high diagnostic accuracy for AD vs. control, that was highly predictive of 93 conversion from mild cognitive impairment (MCI) to AD in an independent group of subjects 94 and outperformed conventional CSF markers such as Aβ, tau derivatives and their ratios [20]. 95 These studies highlight non-canonical pathological cascades that may both provide useful tools 96 for clinical practice and clinical trials purposes, and may also reveal new insights about disease 97 mechanisms underlying AD.

99
One of the peptides identified using this hypothesis-free approach to separate AD from normal 100 (NL) controls was VGF [20]. VGF (a non-initialism) has recently received significant attention 101 because of its role in learning and memory and potential role in the pathophysiology of AD [21, Therefore, in the current study, we examined the potential for VGF in the CSF, when combined 117 with conventional biomarkers of CSF Aβ1-42, total tau (tTau) and pTau-181 and hippocampal 118 volume, to enhance the diagnostic and prognostic accuracy of these markers. The focus of this 119 work is on the VGF peptide fragment with sequence NSEPQDEGELFQGVDPR ("VGF.NSEP") 120 since it previously emerged as a strong predictor in a panel of peptides that predict MCI to AD 121 conversion [20], though other VGF peptide fragments are also examined. Unlike our previous 122 5 studies involving hypothesis-free approaches to identify optimal peptides to include in biomarker 123 signatures [20,33,34], the current study was focused on the utility of VGF. Using data from two 124 independent groups in the ADNI cohort: one group of AD and control subjects and a separate 125 group of MCI subjects, it was found that VGF, when combined with conventional biomarkers, 126 enhanced both the diagnostic accuracy of these markers and the ability of these markers to 127 predict MCI to AD conversion. This research was focused on the relationship between VGF, conventional biomarkers (CSF 144 amyloid/tau and MRI hippocampal volume [HV]) and therefore, only those subjects whose 145 values for these markers were available at baseline were included in this study. Ultimately, this 146 dataset included 287 subjects across the three diagnostic categories (AD, MCI and NL). NL 147 subjects were defined as those without memory complaints and had a clinical dementia rating 148 (CDR) score of 0. MCI subjects had CDR scores of 0.5, had an abnormal score on Wechsler 149 Memory Scale Revised-Logical Memory II and did not have significant functional impairment.

150
AD subjects had functional decline and a CDR score of 0.5 or 1.0. HV was chosen given its robust ability to predict MCI to AD conversion [35,36] and its 154 incorporation into proposed schema to classify AD subjects [37]. HV was obtained from MRI 155 scans (mostly 1.5T; 25% in this dataset had 3.0T scans) and was computed using FreeSurfer 156 software. Please see "UCSF FreeSurfer Methods" PDF document under "MR Image Analysis" in 157 the ADNI section of https://ida.loni.usc.edu/) for details as well as [38][39][40].  Once an optimal signature for differentiating NL from AD was derived, it was tested on a 185 different group of 135 MCI subjects from the ADNI dataset. Baseline values for Aβ1-42, tTau, 186 pTau-181, HV and VGF peptides for each MCI subject at baseline were used to classify each 187 subject as being "signature positive" (i.e., similar to the profile found in AD) or "signature 188 negative" (i.e., similar to the profile found in NC). PPV, NPV and accuracy were then computed 189 by comparing the actual outcome (conversion or not to AD over 36 months) to the predicted 190 outcome (signature positive/negative which would predict conversion/nonconversion, 191 respectively). Exact McNemar's test was used to compare PPV, NPV and accuracy. 192 193 In addition to measuring the performance of whether MCI subjects would convert over 36 194 months, time to conversion was also computed using available data up to 10 years after the initial 195 evaluation. Potential markers for this analysis were grouped into categories:   p=0.285, Chi-squared test). The likelihood that an APO-E4 allele was present was higher AD 218 than in other subjects (present in 71.2% AD, 50% MCI and 24.4% NL subjects, p < 0.0001, Chi-219 squared test) and was a relatively weak risk factor for the conversion of MCI to AD (present in  Prediction of the likelihood of MCI to AD progression: As described above, for disease state classification, no advantage was found when adding the 246 VGF.NSEP peptide to the conventional markers (overall accuracy of 76.3% vs. 75.7%, p > 0.05).

247
However, the combined biomarkers signature (HV+AT+VGF) significantly outperformed 248 conventional biomarkers (HV+AT) for the prediction of MCI to AD conversion over 36 months 249 (p=0.00013). Most of the impact of the addition of VGF was in increasing the NPV (from 70.2% 250 to 79.2%, p<0.0001) while the impact on PPV was more modest (60.2% to 62.1%, p=0.008).

251
The signature derived from the conventional and novel markers took a simple form based on 252 only a few markers, with a cut-point on each of them; HV < 7.81 cm 3 , pTau < 16.18 pg/mL, ratio 253 of tTau to Aβ1-42 > 0.29 and VGF.NSEP peptide < 20.39 intensity units. Thus, the addition of a 254 novel VGF peptide to the conventional AD markers provides a simple biomarker that improves 255 the prediction of 36-month disease progression in MCI subjects at baseline.

257
Prediction of time to AD progression from MCI: 258 Using available information containing 3-10 year follow-up clinical data, future time to 259 progression was computed using the optimal signatures defined above. Table 3 Table 4 and Figure 6). However, 288 the differences were greater in the overall time course of progression that resulted in larger 289 hazard ratios (4.1 and 4.7). Thus, the considerable improvement we see in the prediction of MCI 290 to AD progression by including VGF with the conventional markers is consistently evident for 291 all three peptide fragments of VGF, and not isolated to a specific peptide fragment. Here, we examined the ability of CSF VGF-derived peptides, in combination with conventional 296 AD biomarkers (Aβ1-42, tTau, pTau-181, their ratios and HV) to serve as a disease-state marker 297 to distinguish between AD and NLn subjects, and to predict conversion from MCI to AD in a 298 separate group of subjects. We observed that CSF levels of a VGF peptide, on its own, are lower 299 in AD subjects than NLs and that lower levels predict MCI to AD conversion. When combined 300 with conventional biomarkers, the VGF peptide significantly increased the ability of a 301 combination of conventional biomarkers to predict MCI to AD conversion, with the hazard ratio 302 increasing from 2.2 to 3.9. These data suggest that VGF may play a previously under-recognized 303 role in the pathophysiology of AD and that CSF VGF may be useful to help predict MCI to AD 304 conversion.

11
Total tau vs. phosphorylated tau in predicting MCI to AD conversion: 306 It is notable that, when combined with HV, Aβ1-42 and VGF.NSEP, CSF tTau was found to 307 more strongly predict MCI to AD conversion than pTau-181. tTau, but not pTau-181, elevations 308 in the CSF have been observed in many non-AD conditions involving neuronal injury, including 309 stroke, traumatic brain injury, Creutzfeldt-Jacob disease, multiple sclerosis as well as vascular 310 dementia [50-55], suggesting that tTau is a general marker of neuronal injury, while pTau-181 311 better reflects AD pathology. The finding in the current study that tTau is more strongly 312 predictive of MCI-AD conversion than pTau-181 is consistent with previous data showing that 313 total tau is more predictive than pTau-181 in predicting subsequent cognitive decline in MCI and 314 AD [56, 57]. These findings suggest that while pTau-181 may be more useful as a disease-state 315 marker, particularly when making a differential diagnosis, that tTau may be a better marker of  these findings were reproduced in the current study (Table 2). In addition, recently a number of 364 non-Aβ, non-tau CSF markers have been found, often using proteomic approaches, that separate 365 AD from NL subjects, and these markers have been implicated across a number of metabolic, 366 inflammatory and synaptic physiology pathways [25-29, 31, 84-90]. A small number have also 367 shown the ability to predict MCI to AD conversion. For example, heart fatty acid binding 368 protein, chemokine receptor 2, neurogranin, calbindin, IL-1, thymus-expressed chemokine have 369 all individually been shown to predict MCI to AD progression [14][15][16][17][18][19][20]. In addition, we and 370 others identified panels of peptides that predict MCI to AD progression [19,20]. These data 371 point to a range of potential pathophysiological mechanisms implicated in AD outside of the 372 classical amyloid-driven cascade. It will be important to replicate the findings in this study as 373 well as others in independent cohorts. In addition, like most of the previous work, the current 374 study did not examine non-AD dementia or other neurologic disease. This absence is particularly 375 important in the current study which shows VGF levels that correlate with tTau levels (a marker 376 of neurodegeneration, as described above) but not hippocampal volume (Figures 3C and D).

377
These data suggest that VGF levels may correlate with a more general neurodegenerative 378 phenotype. Therefore, it will be important in future studies to include non-AD dementias as well 379 as other neurological illness such as stroke or encephalitis, to determine the specificity of VGF as 380 a biomarker for AD and predictor of MCI to AD progression.  Figure 2B. Patients meeting the signature criterion that 398 includes the VGF.NSEP peptide experience 3.9-fold faster progression to AD (hazard ratio = 399 3.9), relative to the 2.2-fold faster progression without this peptide.