Overview
Novartis is one of the world's largest pharmaceutical companies, with a marketed portfolio spanning more than 300 active products across oncology (Kisqali, Pluvicto, Scemblix), immunology (Cosentyx, Kesimpta), cardiovascular (Entresto, Leqvio), neuroscience (Aimovig, Mayzent), ophthalmology (Lucentis, Beovu), and gene therapy (Zolgensma). This breadth of portfolio generates one of the highest volumes of periodic safety reporting obligations in the industry.
Novartis's Global Drug Safety (GDS) organization was producing more than 150 Periodic Safety Update Reports (PSURs) and Periodic Benefit-Risk Evaluation Reports (PBRERs) annually, each requiring 6 to 8 weeks of manual effort by cross-functional teams of pharmacovigilance scientists, medical writers, biostatisticians, and regulatory affairs professionals. The cumulative operational burden was consuming the equivalent of 85 full-time employees annually and costing an estimated $20M per year in direct labor and external vendor fees.
ArcaScience partnered with Novartis to deploy a portfolio-wide PSUR automation platform that transformed the periodic reporting process from a labor-intensive manual exercise into a streamlined, AI-assisted workflow with continuous signal monitoring and automated document generation.
The Challenge
The scale and complexity of Novartis's periodic reporting obligations presented challenges that no incremental process improvement could solve:
Massive volume with rigid deadlines. With 150+ PSURs due annually, Novartis's GDS organization was operating in a near-continuous state of PSUR production. The EMA's fixed International Birth Date (IBD) schedule and EURD list requirements meant that deadline compression was not an option. A single missed deadline could trigger regulatory non-compliance proceedings, risk management plan modifications, or in extreme cases, marketing authorization variations. In the 18 months prior to the ArcaScience engagement, two PSURs had been submitted with less than 48 hours of margin, requiring emergency weekend staffing to avoid deadline breaches.
Inconsistent quality across therapeutic areas. Each therapeutic area team had developed its own PSUR preparation workflows, resulting in significant variation in analytical depth, formatting, and benefit-risk presentation. The oncology team's PSURs for Kisqali and Pluvicto used quantitative MCDA-based benefit-risk frameworks, while the cardiovascular team's PSURs for Entresto relied primarily on qualitative assessments. EMA PRAC assessors had noted this inconsistency, with assessment reports for three products in the immunology portfolio requesting "more structured and quantitative benefit-risk evaluation" in the most recent cycle.
Resource-intensive data compilation. Each PSUR required data extraction from Novartis's Argus safety database, clinical trial databases (Novartis Clinical Trial Results Database), published literature (PubMed, Embase), post-authorization safety studies, and risk management plan effectiveness data. For high-volume products like Cosentyx (secukinumab), with over 300,000 patient-years of exposure, the data compilation phase alone consumed 3 to 4 weeks of a pharmacovigilance scientist's time. Manual line listing generation, tabulation of adverse events by System Organ Class, and interval/cumulative summary calculations were repeated identically for each reporting period.
Signal detection fragmentation. Signal detection activities were performed separately from PSUR preparation, often by different teams. The signal management team used empirica Signal for disproportionality analysis on the Argus database, while PSUR authors conducted independent literature-based signal assessments. This created redundancy and, more critically, gaps where signals identified in one workflow were not consistently reflected in the other. A 2023 internal audit found that 12% of signals evaluated by the signal management team were not referenced in the corresponding product's most recent PSUR.
Escalating costs with no scalability path. Novartis's pipeline included 40+ late-stage assets expected to reach market within five years, each of which would add PSUR obligations upon approval. The existing manual model would have required hiring an additional 25-30 FTEs to accommodate portfolio growth, at an incremental cost of $6-8M annually, with a 12-18 month recruitment and training cycle for qualified pharmacovigilance professionals in a highly competitive labor market.
The ArcaScience Solution
ArcaScience deployed a portfolio-wide PSUR automation platform that integrated directly with Novartis's existing pharmacovigilance infrastructure. The implementation was executed in four phases over 24 weeks, beginning with a 10-product pilot and scaling to the full portfolio of 300+ products.
Phase 1: Data Layer Integration
The ArcaScience Data Intelligence Engine was connected to Novartis's Argus safety database via validated API, establishing automated extraction of ICSRs, follow-up reports, and case narratives. Parallel connections were established to the Novartis Clinical Trial Results Database, post-authorization study repositories, and risk management plan tracking systems. The platform automatically calculated interval and cumulative case counts, patient exposure estimates (using both clinical trial enrollment data and commercial sales-based exposure algorithms), and reporting rates by MedDRA System Organ Class, High-Level Term, and Preferred Term.
Phase 2: Automated PSUR Content Generation
ArcaScience's Regulatory Outputs module was configured to generate complete PSUR/PBRER documents following the ICH E2C(R2) template structure. For each product, the system automatically produced: worldwide marketing authorization status tables, estimated patient exposure calculations with methodology documentation, summary tabulations of serious adverse events by SOC and PT, interval and cumulative line listings, signal evaluation summaries cross-referenced with the signal management database, and benefit-risk analysis sections using standardized MCDA frameworks. Medical writers reviewed and refined the AI-generated content rather than authoring from scratch, shifting their role from document creation to quality assurance.
Phase 3: Portfolio-Wide Signal Detection Dashboard
The Decision Intelligence module deployed a unified signal detection dashboard that ran continuous disproportionality analyses across the entire portfolio simultaneously. The dashboard applied product-specific expected event profiles, adjusting signal thresholds based on each drug's mechanism of action and known class effects. For Cosentyx, the system monitored immunosuppression-related signals including serious infections, malignancies, and inflammatory bowel disease exacerbation. For Entresto, it tracked hypotension, hyperkalemia, and angioedema against historical ACEI/ARB class rates. Cross-product signal correlation identified potential class effects early, flagging them for multi-product PSUR coordination.
Phase 4: Portfolio Orchestration & Deadline Management
The platform implemented an intelligent scheduling system that tracked all 150+ PSUR deadlines, automatically initiated data lock points 8 weeks before each submission date, assigned workflow tasks to pharmacovigilance scientists and medical writers, and escalated overdue tasks to team leads. A portfolio-level dashboard provided GDS leadership with real-time visibility into the status of every active PSUR, resource allocation across therapeutic areas, and predictive deadline risk scoring based on historical completion patterns.
Platform Modules Used
Implementation Timeline
24 weeks
10-product pilot + full portfolio rollout
Key Products
Cosentyx (secukinumab)
Entresto (sacubitril/valsartan)
Kisqali (ribociclib)
Kesimpta (ofatumumab)
Pluvicto (lutetium Lu 177 vipivotide)
+ 295 additional marketed products
Regulatory Jurisdictions
FDA, EMA, Swissmedic, PMDA, NMPA, Health Canada, TGA, ANVISA, SAHPRA, and 60+ additional markets
Results & Impact
PSUR Preparation Time Reduction
Average PSUR preparation time decreased from 6-8 weeks to 2.5-3 weeks across the portfolio. For high-volume products like Cosentyx and Entresto, the reduction was even more pronounced: data compilation that previously required 3-4 weeks was completed in 2-3 days through automated extraction and tabulation. Medical writers now spend their time on clinical interpretation and quality review rather than document assembly and data formatting.
Regulatory Deadline Misses
In the 18 months since full deployment, Novartis has achieved a 100% on-time submission rate across all 150+ annual PSURs. The intelligent scheduling system provides 12-week advance visibility into upcoming deadlines, automatic resource allocation based on product complexity scores, and real-time escalation when milestones are at risk. The previous pattern of emergency weekend staffing to meet tight deadlines has been entirely eliminated.
Annual Cost Savings
The automation platform reduced direct PSUR-related labor costs by $8.4M annually (equivalent to 42 FTEs redeployed to higher-value pharmacovigilance activities including signal evaluation and risk management) and eliminated $3.6M in external medical writing and biostatistics vendor fees. The total $12M annual savings represents a 60% reduction from the pre-implementation baseline of $20M. Return on investment was achieved within 14 months of the initial deployment.
Improvement in PRAC Assessment Outcomes
EMA PRAC assessment outcomes improved significantly following the standardization of benefit-risk analysis across the portfolio. The proportion of PSURs receiving "no regulatory action" increased from 62% to 90%, while the rate of major objections requiring supplementary data submissions decreased by 45%. Assessors specifically commended the consistent quantitative benefit-risk framework and the comprehensive signal evaluation cross-referencing in multiple assessment reports.
"The PSUR process had become our organization's single largest operational bottleneck. With 150 reports due annually and a portfolio that keeps growing, we were running out of headroom. ArcaScience didn't just automate the paperwork — they gave us a fundamentally different operating model. Our pharmacovigilance scientists are now spending their time on the science: evaluating signals, interpreting benefit-risk data, and making decisions that matter for patient safety. The $12 million in annual savings is significant, but the real value is that we can now scale our portfolio without scaling our headcount proportionally. That changes the economics of pharmacovigilance entirely."
Dr. Marcus Ehrenberg
Senior Vice President, Global Drug Safety Operations
Novartis Pharma AG — Basel, Switzerland