The Easy Hurdle That Kills Programs
A drug candidate can clear the standard off-target panel and still be a bad clinical bet. That is the uncomfortable truth many development teams, boards, and investors prefer not to say out loud. A 44-target or 47-target screen may show that a molecule avoided the obvious historical liabilities, but it does not prove that the molecule is clean, well-behaved, or tolerable in real human use. The standard panel is a useful early tool, but you must remember that it’s a rather narrow window. It checks only the usual suspects, not the whole crime scene. Eurofins’ SafetyScreen44, for example, was designed around 44 selected targets recommended by four major pharmaceutical companies and includes important liabilities such as hERG, 5-HT2B, transporters, ion channels, nuclear receptors, and selected enzymes. That is valuable, but it is not a full biological interrogation.
This is where management teams can accidentally create a murderous form of false confidence. The molecule “passed,” the slide turns green, and the project moves forward. But in Phase 3, any number of green slides mean nothing. In a JAMA Internal Medicine analysis of 640 novel therapeutics entering phase 3 or other pivotal testing, 54% failed in clinical development. Among those failures, 57% were attributed to inadequate efficacy and 17% to safety concerns. That does not mean every failure was caused by off-target biology, but it does prove the broader point: late-stage development punishes assumptions that should have been challenged earlier. The paper’s exact count is: 344 failed agents, of which 59 failed because of safety concerns. That is why modern off-target profiling belongs in the boardroom vocabulary, not just the discovery team’s methods list. An estimate means that a whopping $2.832 billion was wasted by not investigating off-target effects better before the clinical trials even began…
Tolerability Is Not Drug Safety’s Softer Cousin
Safety is often discussed as if it means “does it trigger a dramatic adverse event?” Tolerability is subtler and, in many commercial contexts, deadlier. A medicine taken once in a hospital setting has a different risk equation from a medicine taken every day for years by patients who have jobs, families, comorbidities, other prescriptions, and very little patience for nausea, fatigue, dizziness, sleep disruption, cognitive fog, gastrointestinal irritation, or quiet laboratory abnormalities. For chronic or preventive therapies, small off-target sins become recurring invoices. Biology collects those with interest that keeps compounding.
Regulators already understand that duration matters. ICH M3(R2) states that the recommended duration of repeated-dose toxicity studies is usually related to the duration, therapeutic indication, and scope of the proposed clinical trial, and that animal toxicity studies should generally equal or exceed the duration of human clinical trials up to recommended limits. ICH S7A also makes clear that cardiovascular, respiratory, and central nervous systems are core life-supporting priorities, while other organ systems may become especially important depending on patient population and intended use. That is exactly the point when making go/no-go decisions: the tolerability bar rises as administration becomes more frequent, longer, broader, or more elective. A tolerability nuisance in late-stage oncology may be manageable. The same nuisance in a long-term cardiometabolic, CNS, autoimmune, dermatology, or women’s health therapy can destroy adherence, differentiation, pricing power, and ultimately the asset’s value.
Reality did Several Billion Years Ago – The Industry Is Finally Moving Beyond 44 Targets
The classic panel is not useless, but insufficient. The industry’s thinking has moved on. In 2024, a Nature Reviews Drug Discovery paper based on a survey across 18 companies described secondary pharmacology screening as standard practice, but also highlighted variation in screening strategies and the need for optimization, including safety-associated targets not covered by many current panels. The follow-up discussion in 2025 points toward a broader Safety-77 approach: an expanded core safety panel of 77 diverse targets intended to improve the detection of potential human safety liabilities while remaining feasible. Can be interpreted as the industry admitting that the old “usual suspects” model was never enough. And, in all honesty, that is the truth also for the Safety-77 approach.
Modern off-target profiling therefore means more than buying a bigger menu of assays. The new toolbox includes expanded functional secondary pharmacology, kinome-wide profiling for kinase-active chemistry, activity-based protein profiling for enzyme and covalent-mechanism questions, affinity-based chemical proteomics such as Kinobeads, label-free CETSA or thermal proteome profiling, and limited-proteolysis mass spectrometry. Each has strengths and blind spots. Activity-based protein profiling is powerful for interrogating protein function and enzyme families in complex proteomes. Kinobeads can expose kinase polypharmacology at scale. A Nature Chemical Biology study profiled 1,183 compounds and generated roughly 500,000 compound-target interactions. CETSA-MS and thermal proteome profiling are attractive because they can detect compound-induced protein stability changes without modifying the compound, target, or cell system. Reviews describe mass-spectrometry-based CETSA/TPP as capable of probing many thousands of proteins and identifying both desired target engagement and unwanted off-target interactions.
Crowning the New Gold Standard
The crown for modern off-target profiling should go to the integrated decision stack, not a single vendor’s brochure. The best current practice is to start with a rational, expanded secondary pharmacology strategy, preferably functional where mechanism matters. Add modality-appropriate deep profiling such as kinome, ABPP, affinity chemoproteomics, CETSA-MS, thermal proteome profiling, or LiP-MS; and then follow suspicious signals with orthogonal functional assays at clinically relevant exposures. That is the gold standard because it respects the actual question: “what is this compound likely to do in a human system, at relevant exposure, over relevant time?” Not just ticking a simplified box.
Within that stack, Pelago deserves serious attention. (Disclosure: Pelago is currently a client of Biosector, but I have assessed this area seriously and as objectively as I can). Pelago’s CETSA-MS Selectivity Profiling assesses compound interactions across more than 5,000 proteins in live cells or tissues, using CETSA with quantitative mass spectrometry in physiologically relevant systems. That is a strong proposition because it addresses the exact weakness of narrow panels: unexpected biology. But CETSA-MS still measures binding-associated thermal-stability changes, not automatically clinical toxicity, functional agonism, antagonism, pathway disruption, or tolerability. LiP-MS (Commercial reading / Scientific reading) offers another important direction because it can detect structural changes in thousands of proteins and, in some workflows, provide peptide-level information useful for approximating binding sites. The objective conclusion is simple: Pelago may be a leading current commercial embodiment of label-free, proteome-wide selectivity profiling, but the gold standard is the workflow that turns those findings into decisions.
The Next Gold Standard Will Be Human, Contextual, and Longitudinal
The next generation will not be a larger panel with more boxes. It will be a human-context risk engine. The future gold standard will probably integrate dose-response and time-course off-target binding, tissue-relevant exposure, patient-relevant human cells, organoids, organ-on-chip systems, functional omics, causal AI models, and longitudinal tolerability hypotheses before the asset becomes too expensive to challenge. This is where off-target biology finally meets the real world: not isolated proteins in a tube, but human biology under repeated exposure, in disease-relevant contexts, with enough functional readout to separate harmless binding from strategic danger.
Regulatory momentum is also moving in that direction. The FDA’s New Approach Methodologies page, updated in 2026, describes NAMs as including in vitro human-based systems, in silico modeling, and other platforms that can evaluate immunogenicity, toxicity, and pharmacodynamics while improving predictive relevance. It specifically highlights AI-powered models, human organ-on-chip systems, and real-world data as part of the shift toward more human-relevant safety evaluation.
Nature Biotechnology has also reported that the FDA is encouraging organ-on-chip, organoid, in silico, and other in vitro approaches, while cautioning that these technologies are not yet ready to fully replace animal testing. That is the right level of ambition and humility.
The future is not: throw out the old methods.
The future is: stop pretending the old methods are anywhere near enough. I mean, how much would you be ready to invest to avoid losing $2.832 billion?
At Biosector, we help life-science companies, investors, and strategic partners challenge assumptions before Phase 3, regulators, patients, or the market do it for them.
If you are evaluating a drug discovery program, preparing a Japan-facing strategy, assessing a biotech investment, or pressure-testing a preclinical asset, we can help you separate reassuring evidence from decision-grade evidence.
Before you move the next asset forward, ask the harder questions.
E-mail info@biosector.jp
Or book a meeting here!
Biosector’s Objective Gold Standard Verdict
The current new gold standard is not one panel and not one vendor. It is a tiered evidence stack: expanded functional secondary pharmacology, preferably moving toward Safety-77 or fit-for-purpose broader panels; proteome-wide or near-proteome-wide chemoproteomic selectivity profiling; and functional, exposure-aware follow-up in relevant human biology. Pelago’s CETSA-MS selectivity profiling is a strong current contender in the proteome-wide, label-free target-engagement layer, but it is not a complete replacement for functional assays, classical safety pharmacology, or indication-specific tolerability work. Nature Reviews Drug Discovery reported in 2024 that secondary pharmacology screening is now standard practice and that regulators increasingly request activity data against targets with recognised adverse-effect relationships; a 2025 follow-up specifically discusses an expanded 77-target core safety panel, Safety-77.
Comments are closed