Biogen Inc. (BIIB) Earnings Call Transcript & Summary

Hello. My name is Laura Nisenbaum, I am the Head of the Diagnostic Pathways Group at Biogen and lead the biomarker strategy for aducanumab. It's my pleasure to present our work on cerebrospinal fluid biomarker concordance with amyloid PET from EMERGE and ENGAGE, our Phase III studies of aducanumab in patients with early Alzheimer's disease. Before moving into the results from my presentation, I provide here the disclosures from the authors of this work. I want to note that aducanumab is an investigational compound and is not yet approved. Finally, please note that this presentation may include forward-looking statements as detailed here. The data described in this presentation come from the EMERGE and ENGAGE Phase III aducanumab trials. These studies were identical in design, randomized, 18-month double-blind and placebo-controlled. Patients were in the early symptomatic stages of disease where symptoms are still limited. Patients were classified clinically as Mild Cognitive Impairment, or MCI, due to Alzheimer's disease and mild Alzheimer's disease dementia. All enrolled patients had confirmed beta amyloid pathology as assessed by PET, with 1 of 3 approved tracers, based on visual read. The primary and secondary endpoints for the studies are listed here. Of most relevance to this presentation, there were 3 sub-studies, one of which included collection of cerebrospinal fluid, or CSF, in which we measured biomarkers associated with Alzheimer's disease pathology. Our objectives for this work were to assess the use of CSF biomarkers, namely Abeta-42, Abeta-40, phospho-tau 181, or p-tau 181, and total tau, or t-tau as an alternative to amyloid PET imaging. We conducted a concordance analysis using the amyloid PET visual read data and the CSF biomarker data from the EMERGE and ENGAGE studies. By demonstrating concordance between amyloid PET and CSF biomarkers, we may provide alternatives to the existing diagnostic toolbox used for patient screening as part of clinical trials, and ultimately, clinical practice. CSF samples collected during screening from a total of 350 subjects who consented to the CSF sub-study and had an evaluable amyloid PET scan were included in this concordance analysis. Amyloid PET scans were performed using 1 of the 3 approved PET tracers, florbetapir, flutemetamol and florbetaben. Lumbar puncture was performed and CSF was collected via the gravity method using an atraumatic needle. CSF biomarker data were generated for the analytes listed here using the Lumipulse G1200 fully automated chemiluminescent enzyme immunoassay system at Covance Central Labs. Receiver Operator Characteristic, or ROC, curves for each biomarker or ratio against the amyloid PET visual read and the area under the concentration time curve were generated. Sensitivity, specificity and the overall agreement were calculated using the optimal cutoff value that maximize the Youden J index. Baseline characteristics from the population used for the concordance analyses are presented in this table. Of the total 350 patients, 308 were amyloid PET positive, whereas 42 were amyloid PET negative. The average age for both groups was 69 years old. And the male to female split was roughly 54% male and 46% female, with slightly higher male percentage in the amyloid negative group than in the amyloid positive group. As expected, there was a higher percentage of APOE4 carriers in the amyloid positive group than in the amyloid negative group, such that the amyloid positive group had close to 70% APOE4 carriers, whereas the amyloid negative group was about 20% APOE4 carriers. The clinical stage breakdown was similar across the 2 groups, with 83% to 84% in the MCI stage and 16% to 17% in the mild AD stage. Receiver Operating Characteristic, or ROC, curve analysis was performed. The area under the curve, or AUCs, were calculated, and are shown along with 95% confidence intervals for each biomarker. For the individual biomarkers, Abeta 1-42 and p-tau had the highest accuracy with AUCs of 0.9 and 0.87, respectively. The plot with the biomarker ratios on the right shows that the combination of Abeta 1-42 with the second analyte increased the accuracy versus the individual analytes and that these ratios demonstrated robust concordance with the amyloid PET results. Following the generation of the ROC curves, the sensitivity or positive percent agreement, specificity or negative percent agreement and the overall agreement for the individual analytes were calculated using the optimal cutoff value. This optimal cutoff was identified by maximizing the Youden J index, which is indicated by the vertical line in these plots. The Youden J index is defined as the sensitivity plus specificity minus 1. As can be seen in the table, Abeta-42 showed the best test performance among the single biomarkers. Specifically, the optimal cutoff value for Abeta-42 was determined to be 664 picograms per mL, which yielded a sensitivity of 90.5%, specificity of 81% and overall agreement of 89.4%. Turning our attention to the CSF biomarker ratios, similar analyses produced the cutoffs and performance characteristics as shown here. For these ratios, the plots showed plateau stages, especially for the Abeta 42-40 ratio, indicating that a wide range of cutoffs yielded similar Youden indices. The optimal cutoffs for the biomarker ratios were 0.062 for Abeta 42-40, 0.066 for p-tau over Abeta-42 and 0.548 for t-tau over Abeta-42. The performance of all 3 biomarker ratios showed overall percent agreement above 90% with the sensitivity and specificity of the Abeta 42-40 ratio being 94% and 88%, respectively. These scatter plots illustrate the agreement between the amyloid PET visual read and CSF biomarker ratios for the individual subjects. Each individual is represented in the plots based on the amyloid PET visual status at screening with a negative value indicated by a blue circle and a positive value indicated by a green plus. As an example, the plot on the far left shows the individual data points for a particular subject as a function of the Abeta 42 and 40 values. The diagonal line in each plot identifies the optimal cut point, below which individuals are identified as amyloid positive based on the CSF biomarker results. The optimal cutoff value for the individual and ratio biomarkers are listed in this table, along with the cutoff values obtained in 2 other published studies using the Lumipulse Automated Analysis platform. The Abeta 42-40 ratio shows the most consistency across the different studies. The variation observed in the cut points across the 3 studies for the individual CSF biomarkers may be related to differences in the study collection protocols and/or the composition of the cohort, such as the amyloid positivity rate. The test performance for the individual and ratio biomarkers are listed in this table along with those obtained in the 2 other published studies using the Lumipulse Automated Analysis platform. As with the cutoff values, the test performance for the CSF biomarkers on the Lumipulse platform is relatively similar across the 3 studies. In summary, we have demonstrated robust concordance between CSF biomarker results as assessed in the screening samples and amyloid PET visual read in the EMERGE and ENGAGE Phase III aducanumab trials. The results from this analysis showed that the CSF biomarker ratios had higher concordance with amyloid PET than the single CSF biomarkers and that the sensitivity and specificity for the Abeta 42-40 ratio is comparable to the performance of approved PET tracers. Finally, the optimal cutoff values for the CSF biomarkers, in particular, the Abeta 42-40 ratio were similar to other studies analyzed using the Lumipulse platform. Lastly, I would like to acknowledge and express my heartfelt thanks to all of the patients and caregivers who participated in the aducanumab EMERGE and ENGAGE studies. Thank you for your attention.