Beyond the pristine protocol: Real-world evidence in emerging modalities

Key takeaways

Bridging the gap: Real-world evidence (RWE) is no longer just for post-market safety; it is increasingly vital for supporting regulatory decisions and label expansions in cell and gene therapies (CGTs) where randomized control trials (RCTs) are ethically or logistically difficult.
The "real" patient profile: Commercial data reveals that patients receiving CAR-T and radiopharmaceuticals in the clinic often differ significantly from trial populations—frequently older, with more comorbidities—yet often achieve comparable outcomes.
Radiopharma specifics: In radioligand therapy, RWE is proving critical for validating treatment benefits in patients who may not meet strict imaging or biomarker criteria (e.g., PSMA expression levels) required for pivotal trials like VISION.
Data infrastructure: The success of RWE relies heavily on breaking down silos between electronic health records (EHRs), claims data, and patient registries to create "regulatory-grade" datasets that satisfy FDA and EMA standards.

The randomized controlled trial (RCT) has long been the "gold standard" of clinical translation—a pristine, controlled environment designed to isolate variables and prove efficacy. But in the messy, high-stakes world of emerging modalities, the gold standard is occasionally losing its luster.

For developers of cell and gene therapies (CGTs), bispecific antibodies, and radiopharmaceuticals, the clean lines of an RCT often fail to capture the complex reality of clinical practice. Patient populations are small, historical controls are often nonexistent for rare diseases, and the "vein-to-vein" logistics of autologous therapies introduce variables that a protocol simply cannot predict.

Enter Real-World Evidence (RWE). Once relegated to the back office of pharmacovigilance, RWE has graduated to the C-suite of clinical strategy. It is now a primary tool for bridging the "efficacy-effectiveness gap"—the chasm between how a drug performs in a trial and how it works in the clinic.

The efficacy-effectiveness gap in CGTs

In the realm of Cell and Gene Therapy, the traditional placebo-controlled trial is often unethical or impossible. You cannot give a placebo gene therapy to an infant with Spinal Muscular Atrophy (SMA). This has forced the industry to lean heavily on natural history studies and RWE to build synthetic control arms [2].

Zolgensma serves as the industry's "hero story" here. The approval of this gene therapy relied partly on comparing single-arm trial data against historical cohorts—essentially, a "real-world" control group that demonstrated the devastating natural trajectory of SMA type 1.

However, the more recent challenge lies in CAR-T therapies. In pivotal trials like KarMMa (for Abecma) or CARTITUDE (for Carvykti), patients were hand-picked for fitness. But commercial data tells a different story. Real-world analyses presented at recent industry meetings (including ASH) consistently show that patients receiving commercial CAR-T products are older, have poorer performance status, and higher rates of organ dysfunction than their trial counterparts.

Despite this "fitness penalty," real-world outcomes have often mirrored trial efficacy. For example, real-world data on ide-cel (Abecma) and cilta-cel (Carvykti) in multiple myeloma has demonstrated that while "vein-to-vein" times may fluctuate and cytokine release syndrome (CRS) management varies by center, the durable responses seen in trials are largely reproducible in the broader patient population [3]. This validation is crucial for payers who are increasingly skeptical of high-price tags attached to single-dose cures.

Radiopharmaceuticals: defining the "real" patient

Radioligand therapy (RLT) presents a unique set of RWE challenges. Unlike a pill, RLTs like Pluvicto (¹⁷⁷Lu-PSMA-617) require a complex interplay of theranostics—diagnostic imaging paired with therapeutic isotopes.

The pivotal VISION trial for Pluvicto had strict inclusion criteria regarding PSMA expression levels on PET scans. Patients with "low" expression were excluded. However, in the clinic, oncologists face a dilemma: do you treat the patient who almost meets the criteria but has no other options?

Recent RWE studies have begun to answer this. Data from centers treating patients post-approval suggest that even those who might have been excluded from VISION (due to lower SUVmax values or discordant lesions) can still derive significant clinical benefit, albeit potentially at lower response rates [4]. This type of evidence is invaluable for clinicians making "off-protocol" decisions and for manufacturers looking to expand labels to "all-comer" populations.

Feature	Randomized Clinical Trial (RCT)	Real-World Practice (RWE)
Patient selection	Homogenous, strict inclusion/exclusion	Heterogenous, includes co-morbidities
Adherence	Strictly monitored and enforced	Variable, patient-dependent
Comparator	Placebo or active control	Standard of care (often variable)
Cost/Complexity	High cost, artificial infrastructure	Lower marginal cost, standard workflow
Primary utility	Regulatory approval (efficacy)	Label expansion, payer value (effectiveness)

Overcoming the data deficit

The transition from "anecdote" to "evidence" requires rigorous data architecture. For emerging modalities, standard claims data (billing codes) is rarely enough. A claim code tells you a patient received a CAR-T infusion; it does not tell you their T-cell expansion kinetics, their specific grade of CRS, or their genomic tumor mutational burden.

To make RWE "regulatory-grade" (fit for FDA submission), developers are increasingly turning to:

High-fidelity registries: Disease-specific registries (like those for interactions in hemophilia or rare cancers) that collect deep clinical phenotypic data.
AI-driven abstraction: Using Natural Language Processing (NLP) to scrape unstructured notes from EHRs to find critical data points like "progression-free survival" that aren't captured in billing codes.
Tokenization: connecting disparate datasets (e.g., linking a patient's genomic profile from a specialty lab to their long-term survival data in an insurance claims database) without violating privacy (HIPAA/GDPR).

As the FDA's 2023 guidance on RWE suggests [1], the agency is open to this data—but only if the provenance is clean and the methodology transparent. For the pioneers of emerging modalities, the message is clear: The trial gets you to the starting line, but the real-world evidence gets you to the finish.