Safety Monitoring for CNS-Active Therapies: Before the SAE Becomes a Problem

Safety monitoring in CNS clinical trials is the integrated operational infrastructure that detects, escalates, and documents adverse events specific to CNS-active compounds, including ARIA in anti-amyloid programs, suicidality signals via the C-SSRS, seizure risk, and psychiatric adverse events, in real time and with a continuous audit trail that meets regulatory inspection standards.

In August 2025, the FDA issued a Drug Safety Communication requiring an additional MRI monitoring scan prior to the third infusion of lecanemab (Leqembi), citing 101 cases of serious ARIA-E reported to the FDA Adverse Event Reporting System and six deaths – four of which occurred after the third treatment. The existing monitoring schedule had covered MRI before the 5th, 7th, and 14th infusions. The gap between infusion 2 and infusion 5 was not a monitoring gap that appeared in hindsight. It was a window the safety data eventually made visible. For CNS programs, the question is not whether serious safety signals will emerge during the trial – it is whether the monitoring infrastructure is configured to see them before they become SAEs.

ARIA (Amyloid–Related Imaging Abnormalities) monitoring is the scheduled MRI surveillance process used to detect ARIA–E (edema/effusion) and ARIA–H (microhemorrhages/siderosis) in patients receiving anti–amyloid therapies. ARIA–E is usually asymptomatic, which means the site cannot rely on participant–reported symptoms to trigger a safety review. Detection depends entirely on scheduled MRI at protocol–defined intervals, graded by a trained neuroradiologist using a standardized scoring system, with results escalated to the medical monitor within the protocol’s reporting window. The FDA’s August 2025 safety communication requiring an additional pre–infusion–3 MRI was a direct response to the finding that 24 serious ARIA–E events and four fatal events occurred in the early infusion window that the original monitoring schedule did not cover.

The C–SSRS is the FDA–recognized standard for suicidality monitoring in CNS trials. It is a rater–administered instrument that requires raters to make clinical judgments about ideation severity and plan specificity. A rater who underscores on the ideation items: recording a lower severity level than the participant’s actual response warrants; does not generate a data quality query in a generic pharmacovigilance workflow. The data appears clean. In a connected monitoring environment, C–SSRS item 4 or 5 responses trigger immediate escalation regardless of the rater’s interpretation. Underscoring on the C–SSRS delays safety escalation for a threshold response, which is a patient safety failure with potential regulatory consequence.

The escalation latency gap is the time between an adverse event being entered at a site and the medical monitor receiving notification. In a disconnected safety environment where events are batched into weekly safety reports, a C–SSRS threshold response entered on Monday may reach the medical monitor on Friday. For a Grade 2 ARIA–E or a suicidal ideation item 4 response, that latency is a patient safety failure and a regulatory documentation problem – the 24–hour and 72–hour reporting clocks run from the event, not from the batch report delivery.

In a connected monitoring environment, the medical monitor’s dashboard aggregates adverse events by type, severity, and site in real time. Three Grade 1 ARIA-H events at three sites in three different countries, reviewed individually in a disconnected system, appear as three separate records. In a connected environment, they appear as a cluster that may trigger a safety review and potentially a protocol amendment before the fourth event occurs. Individual AE reports rarely surface a pattern; cross-site aggregation in real time is the mechanism that converts individual events into safety signals early enough to act on.

A regulatory–ready CNS safety audit trail must be continuous and contemporaneous, linking each adverse event to its detection mechanism, the medical monitor notification, and the sponsor response in a single traceable chain, without requiring reconstruction from multiple source systems. An FDA inspection of a CNS program will review not just whether adverse events were reported on time but whether the monitoring workflow produced a traceable record meeting this standard. Audit trails reconstructed from emails, batch exports, and manual log entries across multiple disconnected systems fail this standard. Continuous audit trails produced by a single integrated safety and EDC environment pass it.

1. For anti-amyloid programs, is the MRI monitoring schedule, including the additional pre-infusion-3 scan required by the FDA’s August 2025 lecanemab safety communication, enforced in the CTMS, or does it rely on site coordinator compliance?
   
2. How are ARIA grading results transmitted from the neuroradiology read to the medical monitor, and what is the documented latency between image read and monitor notification for a Grade ≥2 finding?
   
3. Does C-SSRS threshold response data trigger an immediate alert to the medical monitor, or does it enter a weekly safety batch review queue?
   
4. For programs with cognitive decline as a safety endpoint, are MMSE or CDR scores monitored against protocol-defined threshold values in real time, or are threshold breaches identified at scheduled data reviews?
   
5. How does the safety monitoring workflow distinguish between disease-related cognitive changes and drug-related adverse events in a program where both are expected to appear in the same instrument data?
   
6. Is the C-SSRS data quality monitored through the same endpoint oversight process as efficacy scales? Underscoring on the C-SSRS is a safety data gap, not a data quality query.
   
7. What is the SAE reporting clock workflow? How is the 24-hour serious/unexpected SAE reporting requirement tracked, and where does that tracking live relative to the initial site report?
   
8. How are safety signals aggregated across sites in real time? Can the medical monitor identify a cluster of same-type events across three sites in different countries without manually pulling individual site reports?
   
9. For programs in elderly or cognitively impaired populations, is there a specific monitoring workflow for falls, sedation events, and medication-related cognitive changes separate from the general AE reporting process?
   
10. Does the safety monitoring infrastructure produce a continuous audit trail linking each event to its detection mechanism, escalation notification, and sponsor response, or does that chain require reconstruction from multiple source systems?

Q1: What is ARIA in anti-amyloid Alzheimer’s disease clinical trials?

ARIA (Amyloid-Related Imaging Abnormalities) is an MRI-detected adverse event associated with anti-amyloid therapies for Alzheimer’s disease, categorized as ARIA-E (edema/effusion, brain swelling or fluid buildup) or ARIA-H (microhemorrhages and superficial siderosis). ARIA-E is usually asymptomatic, which is what makes it operationally dangerous in a clinical trial context: detection depends entirely on scheduled MRI at protocol-defined intervals rather than participant-reported symptoms. The FDA’s August 2025 Drug Safety Communication required an additional MRI prior to the third lecanemab infusion following a review of 101 cases of serious ARIA-E and six deaths, four of which occurred after the third treatment.

Q2: How does C-SSRS scoring error create patient safety risk in CNS trials?

The C-SSRS is the FDA-recognized gold standard for suicidality monitoring in CNS trials and is a rater-administered instrument subject to the same scoring error risks as other CNS scales. A rater who records a lower ideation severity level than the participant’s actual response warrants, due to insufficient training, scale version confusion, or time pressure, does not generate a data quality alert in a generic pharmacovigilance workflow. The data appears clean. But the underscored threshold response means the medical monitor is not notified, the participant continues in the study without a safety evaluation, and the regulatory reporting clock does not start. C-SSRS data quality monitoring through ongoing rater oversight is a patient safety function, not a data management activity.

Q3: What FDA requirements apply to ARIA monitoring in lecanemab clinical trials?

The FDA’s August 28, 2025 Drug Safety Communication updated MRI monitoring requirements for lecanemab (Leqembi), requiring an additional MRI scan prior to the third infusion in addition to the previously required scans before the 5th, 7th, and 14th infusions. This change was based on the FDA’s review of 101 cases of serious ARIA-E and six deaths, with four fatal cases occurring after the third infusion. The updated requirement was added to the prescribing information’s monitoring schedule (Section 2.3). Sponsors running anti-amyloid programs should verify that their CTMS-enforced MRI monitoring schedule reflects this updated requirement.

Q4: What is the difference between pharmacovigilance and medical monitoring in a CNS clinical trial?

Pharmacovigilance covers the collection, assessment, and reporting of adverse events across the study lifecycle, including regulatory submissions such as expedited SAE reports, PSURs, and DSURs. Medical monitoring is the real-time clinical oversight function: reviewing individual participant safety data, interpreting adverse events in the context of the protocol, and making escalation decisions on an event-by-event basis. In a CNS program, both functions require indication-specific expertise. When pharmacovigilance and medical monitoring operate within the same integrated platform and the same CNS-specialized team, the escalation chain is shorter, the clinical interpretation is more consistent, and the audit trail is continuous rather than reconstructed.

Q5: What does a continuous safety audit trail require in a CNS clinical trial for regulatory inspection?

A continuous safety audit trail for regulatory inspection must link each adverse event to its detection mechanism (scheduled MRI, C-SSRS administration, site report), the medical monitor notification with timestamp, and the sponsor response, in a single traceable chain without reconstruction from multiple source systems. An FDA inspection of a CNS program reviews not just whether adverse events were reported on time but whether the monitoring workflow produced a contemporaneous, traceable record meeting this standard. Audit trails assembled from emails, batch exports, and manual log entries across disconnected systems require reconstruction and fail this standard.

Sitero has supported more than 230 CNS, neurology, psychiatry, and behavioral health studies across 67 countries. Safety monitoring for CNS-active therapies, ARIA surveillance, suicidality monitoring, cognitive adverse event detection, CNS-specific pharmacovigilance, requires infrastructure built before first patient in, not assembled after the first safety signal fires.

Learn more on Sitero’s clinical operations capabilities, or click the button below to talk to a neurology trial expert:

1. Sitero. CNS/Neurology Program Operational Data. Internal dataset. sitero.com/neurology/
2. FDA. FDA to recommend additional, earlier MRI monitoring for patients with Alzheimer’s disease taking Leqembi (lecanemab). Drug Safety Communication. August 28, 2025. https://www.fda.gov/drugs/drug-safety-communications/fda-recommend-additional-earlier-mri-monitoring-patients-alzheimers-disease-taking-leqembi-lecanemab
3. Centers for Disease Control and Prevention. Alzheimer’s Disease and Dementia. https://www.cdc.gov/alzheimers-dementia/about/alzheimers.html.
4. van Etten, E. S., Mahinrad, S., Grill, J. D., Salloway, S., Atri, A., Cogswell, P. M., Benzinger, T. L. S., Iwatsubo, T., Iadecola, C., Lemere, C. A., Nicoll, J. A. R., Greenberg, S. M., Carrillo, M. C., Jack, C. R., Jr, & Sperling, R. A. (2026). Amyloid-related imaging abnormalities (ARIA) in anti-amyloid therapies for Alzheimer’s disease: An update from the Alzheimer’s Association ARIA workgroup. Alzheimer’s & dementia : the journal of the Alzheimer’s Association, 22(4), e71361. https://doi.org/10.1002/alz.71361
5. Jeong, S. Y., Suh, C. H., Lim, J.-S., Shim, W. H., Heo, H., Choi, Y., Kim, H. S., Kim, S. J., & Lee, J.-H. (2025). Incidence of amyloid-related imaging abnormalities in phase III clinical trials of anti-amyloid-β immunotherapy: An updated meta-analysis. Neurology, 104(8), e213483. https://doi.org/10.1212/WNL.0000000000213483
6. Meglio, M. (2025, March 13). Insights from the TRAILBLAZER-ALZ program: Elevated ARIA risk with donanemab highlights the need for vigilant monitoring. NeurologyLive. https://www.neurologylive.com/view/insights-from-trailblazer-alz-program-elevated-aria-risk-donanemab-highlights-need-vigilant-monitoring