Oncology care is fragmenting — PV impact

Oncologists are now choosing drugs by the molecule and the moment, not by a one‑size‑fits‑all protocol. Patients move through a rapid, branching sequence of targeted agents, immunotherapies, and brief combination regimens guided by next‑generation sequencing and repeat liquid biopsies. That clinical choreography produces many short exposures, frequent switches, and overlapping toxicities that live in different parts of the health system. Those scattered exposures make it harder to say which treatment caused which harm. Oncology teams used to attribute an adverse event to the single trial drug; today, a patient may have three prior targeted agents, a checkpoint inhibitor from six months ago, and a supportive care drug started last week—any of which could plausibly produce the same symptom. A multi‑stakeholder review concluded that current attribution practices in oncology are unreliable and need new methods to parse cause when treatments accumulate and interact. (cancertreatmentreviews.com)30062-3/fulltext) The data that would let safety teams untangle that history are themselves fragmented. Genomic profiling is often sent to external labs, dispensing happens through limited distribution networks or specialty pharmacies, and some infusions occur at community clinics while others happen at academic centers. Those split records break the longitudinal patient timeline that pharmacovigilance depends on, so exposure windows are missing or incomplete in claims and EHR extracts. Industry groups and pharmacy coalitions document how oncology’s limited distribution models and medically integrated dispensing raise capture‑rate problems for prescriptions and clinical follow‑up. (ncoda.org) Regulators are noticing and pushing sponsors toward using real‑world data that can reconnect those timelines. The U.S. FDA’s final guidance on real‑world data and real‑world evidence set expectations for when EHRs, claims, and registries can support regulatory decisions and post‑market study obligations. (fda.gov) The European Medicines Agency adopted a reflection paper last year that lays out methodological standards for non‑interventional RWD studies intended for regulatory use. (ema.europa.eu) Those documents point to a practical pivot for safety operations. Spontaneous reporting and disproportionality screens remain necessary but no longer sufficient. Signal detection needs to be sequence‑aware: time‑to‑event and self‑controlled designs, target‑trial emulation, and causal‑inference methods that explicitly model prior exposures offer a more defensible path to attribution in patients with complex treatment histories. To make those methods work, companies must stitch together clinicogenomic metadata with dispensing and outcome data. Researchers have begun defining a core variable set for clinicogenomic RWD—standard fields for test type, panel content, sampling time relative to therapy, and response metrics—that would let safety and evidence teams align biomarkers, drugs, and adverse events across data sources. (esmorwd.org)00006-2/fulltext) For a safety leader at Genentech this means three concrete priorities. First, build data agreements with major CGP and liquid‑biopsy labs and with specialty pharmacies to obtain timestamped test, dispense, and administration records. Second, deploy sequence‑aware analytics in pharmacovigilance: incorporate self‑controlled case series, target‑trial emulation, and interaction scans into routine signal detection. Third, use regulatory touchpoints to align on fit‑for‑purpose RWE: pre‑submission meetings for post‑market studies and early scientific advice can lock in what data standards and methods the FDA or EMA will accept. Regulators are already increasing their use of RWD studies and asking for better quality and governance; sponsors who move now to centralize clinicogenomic and dispensing timelines will preserve their ability to detect and defend safety signals as oncology care fragments. (encepp.europa.eu)