Online Tool Helps Predict Survival in ALS | MedPage Today

An online model based on eight variables reliably predicted the personal survival outcome of patients with amyotrophic lateral sclerosis (ALS), European researchers reported.

Based on a study of 11,475 ALS patients from 14 European ALS centers from 1992 to 2016, the model predicted whether patients had very long, long, intermediate, short, or very short times

between symptom onset and a composite outcome of noninvasive ventilation for more than 23 hours per day, tracheostomy, or death, according to Leonard van den Berg, MD, PhD, of the University

Medical Center Utrecht in the Netherlands, and colleagues.

"We assessed the external validity of our model across several populations, and showed that it had a probability of more than 95% for good performance," they wrote online in the Lancet

Neurology. The model's area under the receiver operator characteristic curve (AUC) concordance statistic for external predictive accuracy was 0.78 and calibration slope was 1.01.

Using baseline clinical, cognitive, and genetic variables defined at diagnosis, the researchers incorporated eight prognostic factors into the model:

The online model is for physicians only, van den Berg's group stated. Medical doctors must register to obtain access "to minimize the risk of potential harm to patients if the prediction

they receive is shorter than expected," they wrote.

While ALS researchers may see the benefits of this model, clinicians should use caution, observed Hiroshi Mitsumoto, MD, of Columbia University Medical Center in New York City in an

accompanying editorial.

"We do not have strong evidence regarding how to discuss diagnosis and prognosis appropriately and effectively," Mitsumoto wrote. "I believe that some uncertainty and hope are important for

patients who have a disease like ALS, and I expect I am not in the minority. When I see patients, I try to find any factors that might be associated with a better prognosis and emphasize

these factors."

It's not known whether this statistical prediction tool outperforms expert clinicians, added Bjorn Oskarsson, MD, of the Mayo Clinic in Jacksonville, Florida, who was not involved in the

study.

"The predictor tool captures a handful of variables and many less tangible aspects of our patients are not reflected," Oskarsson told MedPage Today. Factors like psychological well-being and

social support aren't incorporated, for example.

It also doesn't account for treatments like riluzole (Rilutek), noninvasive ventilation, and gastrostomy placement. "Presumably, since the patients were all being cared for at

multidisciplinary centers of excellence, the use of these interventions was high in the cohort and the application of the model outside of an ALS clinic may not be accurate," he said. And

international variations in healthcare practices, including the use of edaravone (Radicut) which is unavailable in Europe, may affect results.

Oskarsson plans to try the tool, but will share only information that's requested and likely to be useful. "Preserving a realistic sense of hope is always important in the face of a terminal

disease," he said.

van den Berg disclosed support from the Netherlands ALS Foundation, the Netherlands Organization for Health Research and Development, the EU 7th framework programme for the Euro-MOTOR

project, the Netherlands Organization for Health Research and Development, and Baxalta, as well as relevant relationships with Biogen, Cytokinetics, and Baxalta. Co-authors disclosed

multiple relevant relationships with industry including Biogen, Baxalta, Mitsubishi-Tanabe, Cytokinetics, GlaxoSmithKline, Orion Pharma, Teva Pfizer, CSL Behring, Treeway, Chronos

Therapeutics, and Italfarmaco.

Mitsumoto disclosed relevant relationships with Cytokinetics, Mitsubishi-Tanabe, Denali, and Biohaven.