How risky is risk scoring?

Here at PREDICT we are proponents of using evidence-based risk scoring to screen populations in an attempt to find those at the earliest stages of neurodegenerative disease. As my previous blog post shows, we are not the only team devising sensitive methods for risk statification. The drive to identify people at the early stages across the spectrum of neurodegenerative disease has become a major preoccupation of contemporary neuroscience. In keeping with this theme, earlier this year, the Movement Disorder Society (MDS) published research criteria to diagnose "probable prodromal" parkinson's disease.

But we have to remember the pitfalls of the approach, and this paper from an international group which seeks to evaluate the MDS criteria illustrates some of the difficulties. Whilst we hope to see disease modifying treatment for Parkinson's in the future, currently being "at-risk" confers relatively little advantage and is likely to cause anxiety and distress. Compounding this is the fact that some of those identified as high risk will still not go on to get the disorder, depending how high the risk cut-off score is set. In this combined study of 1348 participants, 2/5 who were identified as "probable prodomal" Parkinson's Disease eventually developed Parkinson's Disease. When the bar was set lower i.e. "probable prodromal" Parkinson's Disease, only 7/24 who fulfilled criteria went on to develop the disease. The participants had been followed up for a maximum of 6 years by the end of the study and with further follow up we may see larger numbers being diagnosed but this should strike a note of caution, and shows the score lacks sensitivity.

Risk scores in Parkinson's are still very much a research tool and clearly there is a long way to go before we would want to introduce them to the clinic. But there are great opportunities; whilst we need to refine scoring systems by continuing to identify the most sensitive markers, the differences in outcome in the high risk groups will also be worth studying for possible protective factors.

https://www.ncbi.nlm.nih.gov/pubmed/28509336

Mov Disord. 2017 Jul;32(7):1025-1034. doi: 10.1002/mds.27035. Epub 2017 May 16.

Application of the movement disorder society prodromal Parkinson's disease research criteria in 2 independent prospective cohorts.

Pilotto A, Heinzel S, Suenkel U, Lerche S, Brockmann K, Roeben B, Schaeffer E, Wurster I, Yilmaz R, Liepelt-Scarfone I, von Thaler AK, Metzger FG, Eschweiler GW, Postuma RB, Maetzler W, Berg D

Abstract

BACKGROUND:

The research criteria for prodromal PD of the MDS propose a new approach for the assessment of the individual probability of prodromal PD. These criteria require a testing of their reliability in different prospective cohorts.

OBJECTIVES:

The objective was to evaluate the MDS prodromal PD criteria in 2 independent prospective studies.

METHODS:

Prodromal PD probabilities of the Tübingen Evaluation of Risk Factors for Early Detection of Neurodegeneration cohort (TREND study, n = 650, recruited by the presence of probable rapid eye movement sleep behavior disorder, depression, and/or hyposmia or none of these at baseline and 2-, 4-, and 6-year follow-up) and the population-based Prospective Evaluation of Risk Factors for Idiopathic Parkinson's Syndrome cohort (PRIPS Tübingen subsample; n = 715, baseline and 3- and 5-year follow-up) were calculated. Baseline posttest probabilities, time to PD diagnosis, marker constellations, and longitudinal changes of prodromal PD probabilities were analyzed.

RESULTS:

Incident PD cases (TREND, n = 10; PRIPS = 7) showed significantly higher likelihood ratios of risk and prodromal markers at baseline when compared with nonconverters. Only 2 of 17 incident PD cases met the criteria for probable prodromal PD (ie, posttest probability > 80%) and 5 had possible prodromal PD (ie, > 50%) 1.4 to 3.8 years before diagnosis. The criteria showed high specificity and negative predictive values (>98%), but low sensitivity (TREND, 30%; PRIPS, 14%) and positive predictive values (TREND, 19%, PRIPS, 50%). The individual risk for prodromal PD in incident PD cases showed an inverse correlation with the time to conversion (Spearman rho = .80, P = .006) and unlike in nonconverters, increased during follow-up.

CONCLUSION:

The MDS prodromal criteria provide a practical framework for the calculation of prodromal PD risk. Although specificity of the criteria is high, most patients will not meet the criteria before diagnosis unless testing is thoroughly performed with numerous and highly specific markers objectively assessed. © 2017 International Parkinson and Movement Disorder Society.