Precision medicine’s patent landscape is fundamentally different from traditional pharma because the most valuable assets — diagnostic algorithms, biomarker panels, companion diagnostic correlations, AI-derived drug targets — sit in the categories most exposed to Section 101 subject matter rejections under the Alice-Mayo framework.
The FDA’s companion diagnostic (CDx) pathway has created a de facto patent-pairing strategy: companies that own both the therapeutic and the CDx biomarker patent cluster control the entire treatment paradigm, from patient selection through prescribing.
CAR-T cell therapy IP illustrates the complexity of precision medicine claim architecture — a single commercial CAR-T product can implicate 400 to 1,000+ patents across vector design, manufacturing process, leukapheresis protocols, and patient selection biomarkers.
Post-Myriad, isolated natural DNA is unpatentable; cDNA, engineered constructs, and non-naturally occurring synthetic analogs remain protectable. The practical implication is a systematic shift toward synthetic and engineered biology in precision medicine IP strategy.
Recentive Analytics v. Fox Corp. (2025) extended Alice to machine learning explicitly — applying a known ML method to a new medical dataset is not patentable without technical improvement to the ML system itself.
Effective exclusivity in precision medicine rarely comes from a single compound patent. It comes from layering primary compound protection, method-of-treatment patents, formulation patents, CDx patents, manufacturing process patents, and regulatory exclusivity stacking (Orphan Drug Designation, NCE exclusivity, pediatric exclusivity).
Section I: What Precision Medicine Actually Is — and Why Its IP Complexity Is Unmatched
Precision Medicine Defined: Beyond the Marketing Slogan
Precision medicine is the clinical application of molecular stratification — using a patient’s genetic, proteomic, metabolomic, and phenotypic profile to select, dose, and sequence therapeutic interventions. The clinical target is what practitioners call the ‘five rights’: the right drug to the right patient at the right dose by the right route at the right time. That is not a marketing framework; it is a clinical optimization problem with enormous data requirements and, consequently, enormous IP surface area.
The ‘four Ps’ framing — predictive, preventive, personalized, participative — captures the therapeutic intent. What it obscures is the IP implication: each ‘P’ generates a distinct category of protectable innovation. A predictive model for disease progression is a different IP asset from a preventive pharmacogenomic dosing protocol, which is in turn distinct from a personalized cell therapy manufacturing workflow. Precision medicine does not produce a single blockbuster compound patent. It produces layered, interdependent IP estates.
Traditional pharma IP is linear: one molecule, one NDA, one Orange Book listing, one loss-of-exclusivity (LOE) date. Precision medicine IP is reticulate. A single precision oncology product — say, a CDK4/6 inhibitor paired with an ESR1 mutation companion diagnostic — involves composition-of-matter claims on the compound, method-of-treatment claims covering ESR1-mutant patient populations, CDx patent claims on the biomarker assay methodology, and potentially trade secret protection over the AI-trained model that reads the assay output. Each layer has a different expiry, a different enforcement mechanism, and a different valuation methodology.
Pharmacogenomics as the IP Engine
Pharmacogenomics — the intersection of pharmacology and genomics to optimize drug therapy based on genetic makeup — is where precision medicine generates its most defensible IP. The practical applications are specific: HLA-B*5701 testing before abacavir prescription (GSK’s HIV therapy, Ziagen) prevents hypersensitivity reactions in ~5-8% of patients who carry the allele. Thymidine synthase (TS) gene expression analysis predicts 5-fluorouracil toxicity. DPYD genotyping before capecitabine dosing identifies patients at risk of grade 4 myelosuppression.
Each of those examples represents an IP opportunity at the intersection of the diagnostic method and the method of treatment. Each also illustrates the post-Mayo challenge: the correlation between HLA-B*5701 and abacavir hypersensitivity is a natural relationship. The patent question is whether the method of testing for it and using the result to guide prescribing adds an ‘inventive concept’ sufficient to pass Section 101 scrutiny. The answer, under current USPTO guidance, is context-dependent and requires claim drafting that goes beyond simply ‘measuring X and concluding Y.’
Key Takeaways: Section I
Precision medicine IP is structurally layered, not linear. Single-compound patent thinking misses the bulk of protectable value.
Pharmacogenomics creates concurrent IP opportunities in diagnostics, method-of-treatment, and patient selection — each with distinct patentability challenges.
The ‘five rights’ framework maps directly to patent claim categories: composition (what drug), method of treatment (which patient), dosing regimen (what dose and route), and timing/sequencing (when to administer).
IP teams entering precision medicine must build claims across all four categories simultaneously, not sequentially.
Section II: The Multi-Omics Stack: Where IP Value Lives in the Data Layer
Genomics, Proteomics, Metabolomics: The Source Data for Precision IP
The analytical foundation of precision medicine is multi-omics integration. Genomics provides inherited and somatic variant data. Proteomics captures protein expression and post-translational modification states that genomics cannot predict. Metabolomics reflects real-time cellular function and drug metabolism. Transcriptomics (RNA-seq) reveals gene expression patterns that bridge genotype and phenotype. Epigenomics captures DNA methylation and histone modification patterns that alter gene expression without sequence changes.
The IP value in multi-omics is not in any single data type. It is in the validated, cross-modal predictive model — the specific combination of, say, somatic mutation burden, PD-L1 protein expression, and serum metabolite ratios that predicts response to a checkpoint inhibitor with clinically meaningful sensitivity and specificity. That combination, when novel and non-obvious, can anchor a method-of-treatment patent with real commercial teeth, provided the claim drafting clears the Mayo framework by incorporating an inventive concept beyond the natural correlation.
High-throughput next-generation sequencing (NGS) has made genomic data generation cheap — whole-exome sequencing now costs under $200 per sample in high-volume clinical settings. The economic moat is no longer in generating the data; it is in the validated, clinically actionable interpretation layer. That is where patents matter most and where the Alice-Mayo gauntlet is most treacherous.
Liquid Biopsy and Circulating Tumor DNA: An Emerging Patent Battleground
Circulating tumor DNA (ctDNA) analysis from liquid biopsy is a precision medicine subsegment with particularly intense IP competition. The core detection technologies — digital PCR (ddPCR), NGS-based targeted panels, and whole-genome sequencing of cell-free DNA — are contested across major players including Foundation Medicine (Roche), Guardant Health, Grail (now part of Illumina’s former plans, now independent), and Personal Genome Diagnostics (PGDx, acquired by Labcorp).
The IP architecture in liquid biopsy is instructive for the broader field. Guardant Health’s patent estate, for example, covers specific error-suppression algorithms for detecting low-frequency variants in cfDNA, targeted panel designs for specific cancer types, and methods for distinguishing clonal hematopoiesis of indeterminate potential (CHIP) variants from tumor-derived mutations. Each of those is a different claim type — algorithmic (facing Alice scrutiny), panel design (composition claims, facing Myriad analysis), and differential diagnostic method (facing Mayo scrutiny). A comprehensive liquid biopsy IP strategy must address all three simultaneously.
The key patent filing strategy in ctDNA is to anchor composition claims on the specific probe set or primer sequences (synthetic nucleic acid constructs, cDNA-analogous, post-Myriad protectable), and to anchor method claims on the error-suppression or variant-calling algorithm, drafted to demonstrate technical improvement to the sequencing system rather than mere application of a natural correlation.
Wearables, Biosensors, and Real-Time Phenotyping: Device IP
Beyond the laboratory, continuous physiological monitoring via wearables generates ‘small data’ phenotyping that complements genomic ‘big data.’ Glucose-monitoring continuous CGM systems (Abbott’s FreeStyle Libre, Dexcom G7), neurostimulation devices, cardiac arrhythmia monitors, and sweat-analysis biosensors all generate streams of real-time phenotypic data that can refine precision treatment decisions.
The IP in wearable biosensors spans device claims (sensor design, materials, electrode geometry), method claims (the specific algorithm that converts raw sensor signal to a clinically meaningful output), and software claims (the patient-facing interface and the clinical decision support output). The device claims are typically the strongest and most defensible. The method and software claims face Section 101 pressure under Alice but can be strengthened by emphasizing technical improvements to sensor accuracy or signal processing rather than the downstream clinical inference.
Key Takeaways: Section II
The IP value in multi-omics is in the validated, cross-modal predictive model, not in raw data generation.
Liquid biopsy is a high-stakes IP battleground with simultaneous exposure to Myriad (probe/primer sequences), Mayo (diagnostic correlations), and Alice (variant-calling algorithms) patent eligibility doctrines.
Wearable biosensor IP is strongest in device claims (sensor design, materials) and weakest in the software-inference layer, which requires technical-improvement framing to survive Alice scrutiny.
Companies that own both the data-generation hardware and the validated interpretation algorithm have a structural IP advantage — each component reinforces the defensibility of the other.
Section III: AI and Machine Learning Patents: The Alice/Mayo Gauntlet
The 2025 State of AI Patentability in Precision Medicine
The Federal Circuit’s decision in Recentive Analytics, Inc. v. Fox Corp. (2025) closed the door on a claim strategy that many precision medicine companies had quietly relied on: applying a well-known machine learning technique (gradient boosting, transformer architecture, convolutional neural networks) to a new medical dataset and claiming the combination as novel. The court held that such claims are directed to abstract ideas under Alice step one, and that substituting one data domain for another does not add the ‘inventive concept’ required by Alice step two.
The practical consequence for precision medicine AI is severe. A model trained to predict immunotherapy response from tumor microenvironment RNA-seq profiles, if it uses standard ML architecture applied to a new dataset, is likely ineligible under Recentive. The path to eligibility requires demonstrating either a technical improvement to the ML system itself — novel training methodology, novel loss function design, novel architecture that addresses a specific computational problem posed by the data type — or a non-conventional integration of the ML system with hardware that produces a technical effect beyond the mere output of a prediction.
IBM’s Watson Oncology (the system that reportedly diagnosed a rare leukemia subtype in 10 minutes by cross-referencing genetic data against oncology literature) illustrates the challenge. The clinical value of such a system is undeniable. Its patent eligibility under current doctrine is contested: if the AI simply queries a database and returns a correlation using standard NLP, the claim looks abstract. If the system implements a novel retrieval-augmented reasoning architecture that technically improves how medical literature is indexed or cross-referenced, the claim becomes defensible.
What AI Patent Claims in Precision Medicine Need to Look Like
For AI-enabled precision diagnostics to survive post-Recentive scrutiny, claim drafting must satisfy four criteria, each of which needs to be demonstrated in the specification with technical specificity.
First, the claim must describe a technical improvement to the AI or computational system, not just a new application of existing AI. A novel attention mechanism that handles high-dimensional sparse genomic data more efficiently than standard transformers qualifies. Simply feeding genomic data into GPT-4 does not.
Second, claims should describe non-conventional data processing steps — novel preprocessing pipelines for multi-modal omics data, novel feature engineering that captures biological structure unavailable to standard vectorization, or novel data augmentation strategies specific to low-n genomic cohorts.
Third, the specification must detail how the system architecture addresses a specific, identified technical problem. ‘Predicting drug response’ is not a technical problem in the patent sense. ‘Handling batch effects and confounding between RNA-seq libraries generated across multiple sequencing runs in a federated multi-site training environment’ is a technical problem with concrete solutions that can form the basis of a defensible claim.
Fourth, training methodology innovations — novel loss functions, novel active learning strategies for rare disease populations, novel federated learning protocols that preserve patient privacy while enabling cross-institutional model training — are among the strongest AI patent claims in precision medicine because they directly improve the AI system’s technical performance rather than simply directing it toward a new task.
Inventorship and AI: The Unanswered Question
The USPTO’s current position is that AI cannot be a named inventor on a patent application. Inventorship requires a human. But as AI systems move from assisting human researchers to autonomously identifying novel drug targets, designing novel molecules, and predicting previously unknown biomarker-disease correlations, the inventorship question becomes a practical problem, not just a philosophical one.
Companies using AI-assisted drug discovery — Insilico Medicine, Recursion Pharmaceuticals, AbSci, Exscientia (acquired by Recursion) — face a documentation challenge: they must demonstrate that human researchers made the ‘conception’ of the claimed invention, not the AI. This requires careful internal documentation of the human decision-making steps that guided AI outputs toward a specific claimed invention. Without that documentation, a competitor challenging the patent’s inventorship could argue that the human contribution was merely operating the AI, not inventing the claimed subject matter.
Key Takeaways: Section III
Recentive (2025) explicitly extended Alice to machine learning — applying standard ML to new medical data is not patentable without technical improvement to the ML system.
Valid AI patent claims in precision medicine focus on novel training methodology, novel data processing architecture, or non-conventional integration with hardware.
‘Predicting X from data Y’ is a claim structure that fails post-Recentive. ‘A system that improves computational efficiency of multi-modal omics processing by [specific technical mechanism]’ can succeed.
Companies using AI-assisted drug discovery need internal documentation protocols that clearly establish human inventive contribution to each claimed output.
Section IV: CAR-T, Gene Editing, and Advanced Delivery: Device and Biologic IP Architecture
CAR-T Cell Therapy: A 1,000-Patent IP Estate per Product
Chimeric antigen receptor T cell (CAR-T) therapy represents precision medicine at its most technically complex — and its most IP-intensive. A single approved CAR-T product operates within an IP landscape of extraordinary density. Novartis’s Kymriah (tisagenlecleucel), the first FDA-approved CAR-T, sits at the center of a web of patents held by the University of Pennsylvania (foundational CAR-T platform IP licensed to Novartis), Novartis itself (manufacturing process, patient selection, safety monitoring), and a range of third parties covering lentiviral vector manufacturing, IL-2 signaling pathway modifications, and persistence-enhancing co-stimulatory domain designs.
The CAR-T IP stack has at least six distinct layers. The CAR construct design layer covers the scFv antigen-binding domain, the transmembrane domain, the co-stimulatory domain (CD28 vs. 4-1BB, a distinction with major clinical and IP implications), and the intracellular signaling domain (CD3-zeta). The vector design layer covers lentiviral vs. retroviral vs. non-viral delivery, with the Sleeping Beauty transposon system representing a distinct non-viral alternative that sidesteps lentiviral manufacturing IP while creating its own patent claims on transposase enzyme constructs and minicircle DNA design. The manufacturing process layer covers leukapheresis protocols, T cell activation and expansion protocols, cryopreservation methods, and quality control release assays. The patient selection layer covers the companion diagnostic biomarkers that identify eligible patients. The safety monitoring layer covers cytokine release syndrome (CRS) management protocols, which are increasingly claimed as method-of-treatment patents. The ‘off-the-shelf’ (allogeneic) CAR-T layer covers CRISPR-mediated TCR knockout, HLA class I deletion, and genome integration site selection — all of which generate distinct, layered IP.
Bristol Myers Squibb’s Breyanzi (lisocabtagene maraleucel) and its ip cluster around 4-1BB co-stimulatory domain (licensed from St. Jude Children’s Research Hospital) illustrates how foundational platform IP, held by academic institutions and exclusively licensed to commercial developers, creates structural IP dependencies that shape competitive strategy for a decade or more.
CRISPR-Cas9: The Most Litigated Gene Editing IP in History
CRISPR-Cas9 patent ownership has been contested in one of the longest, most expensive interference and derivation proceedings in USPTO history. The Broad Institute (MIT and Harvard) and the University of California, Berkeley fought over inventorship of CRISPR-Cas9 in eukaryotic cells for years. The core dispute: Jennifer Doudna and Emmanuelle Charpentier (UC Berkeley) demonstrated CRISPR-Cas9 function in vitro and in prokaryotic cells; Feng Zhang (Broad Institute) demonstrated it in eukaryotic (mammalian) cells. The USPTO ultimately awarded Broad Institute patents on eukaryotic application — the commercially relevant scope for human therapeutics — while UC Berkeley retained patents on the underlying system.
The commercial consequence is a fractured licensing landscape. Any company developing CRISPR-based therapeutics for human use must navigate licenses from both estates, plus third-party foundational IP from Caribou Biosciences, Editas Medicine, CRISPR Therapeutics, and others. Vertex Pharmaceuticals and CRISPR Therapeutics’ exagamglogene autotemcel (Casgevy), approved in 2023 for sickle cell disease and beta-thalassemia, is the first approved CRISPR therapy and operates under a complex licensing structure that reflects this IP fragmentation.
IP teams evaluating CRISPR-based assets must map not just composition-of-matter claims on the guide RNA or Cas9 variant, but also the freedom-to-operate (FTO) landscape across both foundational CRISPR estates and any specific therapeutic indication patents held by competitors. A guide RNA targeting BCL11A enhancers for hemoglobin switching (the mechanism in Casgevy) may be subject to claims from multiple holders simultaneously.
3D Printing and Controlled-Release Drug Delivery
Personalized 3D-printed pharmaceuticals represent a nascent but growing precision medicine IP category. Aprecia Pharmaceuticals’ SPRITAM (levetiracetam) — the first 3D-printed drug approved by the FDA — was manufactured using ZipDose technology, with patent protection covering the specific layer-by-layer printing process, the resulting porous tablet structure, and the rapid disintegration properties that make the dosage form clinically differentiated. The case illustrates that manufacturing process IP, often treated as secondary to composition claims, can be the primary moat in precision drug delivery.
Biomaterial-mediated controlled release — hydrogel matrices, PLGA microspheres, lipid nanoparticle (LNP) formulations — adds another layer. Moderna and Pfizer-BioNTech’s mRNA COVID vaccines demonstrated the commercial value of LNP delivery IP. The patent disputes over LNP composition between Moderna and Arbutus Biopharma, settled in 2024 after years of litigation, illustrate how delivery system IP can threaten multi-billion dollar products independent of the therapeutic payload’s patent status.
Key Takeaways: Section IV
CAR-T IP has at least six distinct layers: construct design, vector design, manufacturing process, patient selection, safety monitoring, and allogeneic platform. Competitive analysis requires mapping all six.
The 4-1BB co-stimulatory domain (held by St. Jude, licensed to BMS/Juno), CD19 scFv sequences, and lentiviral manufacturing methods are among the highest-leverage foundational CAR-T IP clusters.
CRISPR FTO analysis is non-trivial — both the Broad and UC Berkeley estates, plus therapeutic indication-specific patents, must be cleared before entering any CRISPR program.
LNP delivery IP is commercially critical for mRNA and gene therapy payloads. Delivery system patents can independently block a therapeutic program regardless of payload patent status.
Section V: Market Size and LOE Exposure: The Numbers That Drive Patent Urgency
Precision Medicine Market Valuation
The global precision medicine market was valued at approximately $70.45 billion in 2023, with projections to reach $170.64 billion by 2030, at a CAGR of roughly 10.5%. The related precision diagnostics subsegment, at $132.46 billion in 2023 and growing to an estimated $145.53 billion in 2024, is projected to reach $246.66 billion by 2029 at an 11.1% CAGR. Oncology holds the largest share by indication; genetic testing leads by diagnostic type.
These are large numbers, but portfolio managers should focus on a more operationally useful metric: patent-protected revenue at risk from LOE events. The precision oncology segment, where companion diagnostic-linked therapeutics command price premiums of $150,000 to $400,000+ per treatment course for CAR-T, is where patent expiry has the sharpest financial cliff. When Novartis’s Kymriah base composition patents expire, the manufacturing complexity and the established CDx ecosystem may provide continued pricing power — but the legal exclusivity is gone, and biosimilar CAR-T programs will enter.
LOE Exposure in Precision Oncology: Selected Case Studies
Pembrolizumab (Keytruda, Merck) is the most commercially significant example of precision medicine patent strategy in execution. Keytruda is approved across 40+ indications, many of them defined by biomarker status (MSI-H/dMMR, TMB-H, PD-L1 TPS/CPS thresholds). The primary compound patent was anticipated to expire around 2028, but Merck has filed an extensive patent estate covering specific biomarker-defined patient populations, dosing regimens (flat dosing vs. weight-based dosing), combination regimens, and manufacturing processes. Each of those secondary patents extends effective exclusivity beyond the base compound LOE. This is precision medicine evergreening executed at scale.
Osimertinib (Tagrisso, AstraZeneca) for EGFR-mutant non-small cell lung cancer is another instructive case. Tagrisso was developed specifically for EGFR T790M resistance mutation patients — a biomarker-defined population. AstraZeneca holds patents on the compound, on the method of treating T790M-mutant NSCLC, and on the companion diagnostic methodology for T790M detection. The method-of-treatment claims covering a genetically defined patient population are, subject to Mayo analysis, potentially more durable exclusivity tools than the base compound patent. Generics manufacturers entering osimertinib face not just compound patent litigation but also method-of-treatment Paragraph IV challenges tied to the biomarker selection criteria.
Investment Strategy for Analysts: Market and LOE Signals
For portfolio managers and institutional analysts assessing precision medicine assets, the relevant due diligence questions differ from traditional small-molecule pharma. The compound patent LOE date is necessary but insufficient. The full patent estate needs mapping: How many Orange Book-listed patents cover the asset? Do any of them cover biomarker-defined patient populations? What is the patent term extension (PTE) status under 35 U.S.C. § 156? Has the company secured any Hatch-Waxman exclusivity beyond the base compound NCE exclusivity?
Orphan Drug Designation (ODD) grants seven years of market exclusivity from approval in the U.S. for drugs treating diseases affecting fewer than 200,000 patients. In precision medicine, where biomarker-defined patient populations are by definition smaller than the unselected population, ODD is a recurring feature of the regulatory exclusivity stack. A company that secures ODD for a biomarker-defined indication gets seven years from approval, independent of and potentially extending beyond the compound patent expiry. For rare disease precision medicine programs, the ODD exclusivity period is often the binding constraint on generic entry, not the patent.
Pediatric exclusivity, which grants six months of additional exclusivity attached to any existing patent or regulatory exclusivity period upon completion of FDA-requested pediatric studies, is another underappreciated exclusivity stacking mechanism in precision medicine, particularly for oncology indications where pediatric tumor genomics often differ materially from adult tumors.
Key Takeaways: Section V
The global precision medicine market exceeds $70 billion and is growing at ~10-11% CAGR, with precision oncology driving the highest-value IP assets.
For financial analysts, compound patent LOE date is necessary but not sufficient for modeling generic entry risk. Method-of-treatment, CDx, formulation, and process patents all contribute to effective exclusivity.
ODD is a particularly powerful exclusivity tool in precision medicine because biomarker selection naturally narrows patient populations toward rare-disease thresholds.
Pembrolizumab (Keytruda) and osimertinib (Tagrisso) are the canonical case studies in precision medicine evergreening — both demonstrate how biomarker-specific method-of-treatment patents extend effective exclusivity beyond compound patent expiry.
Section VI: The Four Pillars of Patentability — Applied to Precision Medicine
Novelty: The First-to-File Race and the Disclosure Trap
An invention is novel if it has not been publicly known or disclosed before the filing date. In practice, the ‘first-to-file’ system creates a race to the USPTO, and in precision medicine — where genomic discoveries are published in preprint servers within days of completion — the gap between discovery and disclosure is dangerously narrow.
The U.S. provides a one-year grace period: a patent application filed within 12 months of the inventor’s own public disclosure remains valid for U.S. prosecution. But that grace period does not apply to international filings under PCT. Any public disclosure before PCT filing destroys foreign novelty. A genomics researcher who presents unpublished data at a conference poster session has started a clock: file the PCT within 12 months or lose international rights. File the U.S. application before the presentation or rely on the grace period with its limitations.
Non-Obviousness: The Interdisciplinary Challenge
Non-obviousness requires that the invention would not have been apparent to a person of ordinary skill in the art (POSITA) at the time of filing. In precision medicine, defining the POSITA is itself contested. Is a POSITA in a CAR-T patent a molecular biologist? An immunologist? A cell therapy manufacturing engineer? A physician-scientist specializing in adoptive cell therapy? The answer determines what prior art is combined and whether the combination is obvious.
The typical examiner challenge in multi-biomarker precision medicine patents: each individual biomarker may appear in the prior art in connection with the disease, so the combination of biomarkers appears obvious. The response requires evidence of unexpected results — demonstrating that the specific combination achieves clinical performance (sensitivity, specificity, AUC) materially superior to what any single biomarker or smaller subset achieves. That evidence should be built into the specification at filing, not manufactured during prosecution.
Utility: Clinical Validation Timing vs. Filing Urgency
Utility requires a specific, substantial, and credible real-world use. In early-stage precision medicine development, particularly for biomarkers identified through retrospective analysis of banked tumor samples, the utility of a novel predictive biomarker may be established in vitro or in early-phase clinical correlations without full Phase 3 validation.
The USPTO’s utility standard for medical diagnostics requires ‘definitive, substantial, and believable’ utility — not proof of Phase 3 efficacy. Credible early-phase clinical correlation data, properly documented in the specification, can satisfy this threshold. The strategic tension is filing early enough to preserve novelty against rapid publication, while filing late enough to have sufficient clinical data to demonstrate credible utility. Provisional applications can bridge this gap: file a provisional with available data to establish priority date, then build the complete clinical dataset before converting to non-provisional within the 12-month window.
Eligible Subject Matter: The 101 Minefield
Section 101 is where precision medicine patents die most often. The exclusions for natural phenomena, abstract ideas, and laws of nature are the primary obstacles. Gene sequences tied to natural biological function are natural phenomena post-Myriad. Correlations between a biomarker level and a clinical outcome are natural relationships post-Mayo. Algorithms that process medical data are abstract ideas post-Alice.
The path through 101 in precision medicine runs through the ‘significantly more’ analysis: the claim must add an inventive concept that transforms the natural law or abstract idea into patent-eligible subject matter. For diagnostics, this typically means integrating the biomarker correlation into a technical measurement method that is itself non-conventional. For AI, it means improving the computational system, not just directing it to a new dataset. For gene editing, it means the human-made modification — the engineered guide RNA, the synthetic Cas9 variant — rather than the natural sequence it targets.
Patentability Criteria Applied to Precision Medicine: Reference Table
Criterion
Precision Medicine Challenge
Claim Drafting Strategy
Novelty
Rapid preprint publication; conference disclosure before filing
File provisional before any public presentation; PCT before any disclosure
Non-Obviousness
Individual biomarkers in prior art; examiner combines references
Pre-load specification with unexpected results data for multi-biomarker combinations
Utility
Early-phase data before full clinical validation
Provisional filing with Phase 1/2 correlation data; convert after Phase 3
Claim the engineered construct, non-conventional method, or technical improvement
Key Takeaways: Section VI
Non-obviousness in multi-biomarker precision medicine patents requires pre-loaded ‘unexpected results’ evidence in the specification — not just argument during prosecution.
The utility threshold for diagnostic biomarker patents does not require Phase 3 evidence. Credible Phase 1/2 clinical correlation with sufficient sample size can satisfy the standard.
Section 101 eligibility is the primary kill zone for precision medicine patents. Claims must be drafted to claim the technical application of a natural phenomenon, not the phenomenon itself.
Provisional applications are the correct mechanism for bridging the filing urgency vs. clinical data maturity tension.
Section VII: The USPTO Process, Provisional Filings, and the Public Disclosure Trap
The Five-Stage USPTO Process: Where Precision Medicine Cases Get Stuck
The USPTO’s formal patent process moves through five stages: preparation, filing, examination (prosecution), grant, and maintenance. For precision medicine applicants, the examination stage is where the majority of delays and rejections originate — particularly in Technology Center 1600 (biotechnology and organic chemistry) and TC 2600 (communications and AI).
Average patent prosecution pendency for biotech applications at the USPTO currently runs 28 to 36 months from filing to first office action, with total pendency from filing to grant or abandonment averaging 36 to 48 months for contested cases. In a field where the clinical development timeline for a precision medicine product can run 8 to 12 years, the patent prosecution clock is running concurrently with clinical development. A composition-of-matter patent filed at IND-enabling studies may not issue until Phase 2 or Phase 3 completion, at which point the clock on the 20-year patent term from filing date has already consumed 4 to 6 years.
Patent Term Adjustment (PTA) under 35 U.S.C. § 154(b) provides day-for-day extension of patent term for USPTO delays beyond statutory deadlines. Patent Term Extension (PTE) under 35 U.S.C. § 156 provides up to five additional years of patent term to compensate for regulatory review time, subject to a maximum of 14 years of remaining patent term post-approval. For precision medicine products with long development timelines, PTE applications should be filed within 60 days of FDA approval to preserve the full extension.
Provisional Applications: Strategic Use, Not Just a Filing Placeholder
A provisional patent application establishes a priority date without beginning the 20-year patent term clock. It provides 12 months for the applicant to develop additional data, refine claims, and assess commercial potential before committing to the non-provisional filing costs. In precision medicine, provisionals are not administrative placeholders — they are strategic clinical-development synchronization tools.
The provisional must satisfy the written description and enablement requirements of 35 U.S.C. § 112 for the subject matter it is intended to protect. A provisional that describes a biomarker discovery but omits the specific assay methodology, threshold values, and patient population characteristics will not support claims in the non-provisional that rely on those omitted elements. Inadequate provisionals are a common and costly error: the priority date is lost for the subject matter not adequately disclosed, creating a prior art problem if that subject matter is published in the interval.
Public Disclosure: Every Form That Destroys Priority
The forms of disclosure that trigger the novelty bar are broader than most scientists realize. Academic conference presentations, poster abstracts posted on conference websites before the event, preprint manuscripts on bioRxiv or medRxiv, NIH grant applications (which become public after award), thesis dissertation defenses, and informal technical discussions with collaborators outside confidential disclosure agreement (CDA) coverage can all constitute public disclosure.
For international prosecution: any of the above occurring before PCT filing eliminates foreign patent rights permanently. The one-year U.S. grace period does not apply internationally. European Patent Convention (EPC) Article 54 defines prior art as ‘everything made available to the public before the date of filing.’ No grace period. A single conference abstract, published before the PCT is filed, destroys European novelty for the disclosed subject matter.
Key Takeaways: Section VII
PTE applications for approved precision medicine products must be filed within 60 days of approval. Missing this window can cost years of exclusivity.
Provisional applications must adequately disclose all subject matter intended for later claiming. Thin provisionals are a recurring, expensive error in academic-origin biotech IP.
Any public disclosure before PCT filing destroys international novelty. The European Patent Convention has no grace period.
Internal IP disclosure policies that require scientists to notify the IP department before any external communication — including informal collaborator discussions — are essential in precision medicine research organizations.
Section VIII: Landmark Cases That Redrew the Map: Mayo, Myriad, Alice, Recentive
Mayo Collaborative Services v. Prometheus Laboratories (2012): Diagnostic Methods Under Scrutiny
Mayo remains the controlling authority on diagnostic method patent eligibility. The patent at issue covered methods for determining the proper dosage of thiopurine drugs by measuring metabolite levels (6-thioguanine and 6-methylmercaptopurine) in the blood and using those levels to assess whether drug doses needed adjustment. The underlying correlation — between metabolite concentration and clinical efficacy/toxicity — was a natural relationship. The USPTO granted the patent; the Supreme Court reversed unanimously.
The Court’s reasoning: the correlation between thiopurine metabolite levels and drug effect is a law of nature. The additional steps in the claim — administering the drug, measuring metabolite levels, ‘indicating’ that doses need adjustment — were routine, conventional steps that added nothing inventive beyond applying the natural law. The claims ‘effectively grant a monopoly over the natural law itself,’ the Court held.
The practical aftermath was severe. The USPTO issued revised Section 101 examination guidelines that applied Mayo to all diagnostic and personalized medicine method claims. Between 2012 and 2019, hundreds of diagnostic patents were invalidated under the Alice-Mayo framework, including patents covering Myriad Genetics’ non-BRCA diagnostic methods, Cleveland Clinic’s cardiovascular biomarker patents, and Johns Hopkins’ sepsis diagnostic claims.
The path through Mayo for a diagnostic method requires demonstrating that the claim’s additional steps amount to ‘significantly more’ than applying the natural law. Courts have found ‘significantly more’ where claims require a novel, non-routine measurement technique (not conventional immunoassay or PCR), an unconventional combination of biomarkers that produces an unexpected synergistic diagnostic accuracy, or integration of the natural correlation into a technically improved diagnostic device or system.
Association for Molecular Pathology v. Myriad Genetics (2013): Gene Patents and the Product of Nature Doctrine
Myriad Genetics obtained patents on isolated BRCA1 and BRCA2 gene sequences and used those patents to maintain a monopoly on BRCA diagnostic testing, charging approximately $3,000 to $4,000 per test. Competing laboratories that sought to offer lower-cost BRCA testing received cease-and-desist letters from Myriad. The ACLU and Breast Cancer Action challenged the patents; the case reached the Supreme Court.
The Court held unanimously that isolated naturally occurring DNA segments are products of nature and cannot be patented, regardless of the effort required to isolate them. The isolation process does not create something new — the genetic information encoded in BRCA1 and BRCA2 is the same in the isolated sequence as in the chromosome. Myriad ‘did not create anything,’ the Court stated.
The Court drew a clear line at cDNA: complementary DNA, synthesized from mRNA reverse transcription, is not naturally occurring because it lacks the introns present in the genomic sequence. The lab technician creates something new when synthesizing cDNA, making it patent-eligible.
The commercial aftermath: multiple laboratories began offering BRCA testing within months of the decision. GeneDx, Ambry Genetics, LabCorp, and Quest Diagnostics all launched competing BRCA tests. The price of BRCA testing fell substantially. Myriad’s stock declined. The European Patent Office had already narrowed Myriad’s European patents through opposition proceedings; the U.S. decision aligned with a global trend toward restricting gene patent scope.
For precision medicine IP practitioners, Myriad’s holding creates a clear strategic directive: draft composition claims on engineered constructs (cDNA, synthetic oligonucleotides, non-natural base analogs, modified nucleic acid probes) rather than naturally occurring sequences. The natural sequence itself is not patentable. The engineered tool for detecting or manipulating it is.
Alice Corp. v. CLS Bank International (2014): Abstract Ideas and the Two-Step Test
Alice established the two-step Mayo framework’s application to computer-implemented inventions. Step one: is the claim directed to a patent-ineligible concept (abstract idea, law of nature, natural phenomenon)? Step two: if so, does the claim add ‘significantly more’ that transforms the claim into a patent-eligible application?
The Alice decision substantially narrowed software patent eligibility, with cascading effects on computational biology, bioinformatics, and AI-driven drug discovery platforms. Claims that simply apply a mathematical algorithm or conventional data analysis technique to medical data are abstract ideas under Alice step one. A claim that describes ‘applying linear regression to genomic expression data to predict drug response’ is directed to the abstract mathematical operation of regression, applied to a new dataset. That fails step two unless the claim incorporates a technical improvement to the computational system.
Recentive Analytics v. Fox Corp. (2025): Machine Learning Meets Alice
The Federal Circuit in Recentive extended Alice explicitly to machine learning models. Fox Corporation challenged Recentive’s patents on ML-based TV viewership prediction models. The Federal Circuit held that claims directed to applying generic ML techniques to a new data domain — even if the application is novel — are abstract ideas without patent-eligible application.
The court was specific: the claims did not improve ML technology. They did not describe novel training methods, novel architectural improvements, or any technical advancement to the ML system. They simply applied well-understood ML to sports broadcast data. Applying standard ML to medical imaging data, genomic sequencing data, or drug response datasets would be analyzed identically.
The practical implication for precision medicine AI companies: product-line patents on ML-based diagnostic or drug discovery tools built on standard architectures (standard CNNs for medical image analysis, standard transformers for genomic sequence analysis) are highly vulnerable to § 101 invalidity challenges. Companies should audit their AI patent portfolios for Recentive vulnerability and prioritize continuation applications that refocus claims on technical improvements.
Landmark Precision Medicine Patent Cases: Reference Table
Case
Year
Ruling
Precision Medicine Implication
Diamond v. Chakrabarty
1980
Genetically modified bacteria patentable as non-naturally occurring manufacture
Established that engineered biological organisms and constructs are patentable
Mayo v. Prometheus
2012
Diagnostic methods based on natural correlations using routine steps not patent-eligible
Isolated natural DNA unpatentable; cDNA patentable
Prohibits natural gene sequence patents; strategy shifts to engineered constructs and synthetic biology
Alice Corp. v. CLS Bank
2014
Computer-implemented abstract ideas not patent-eligible without ‘significantly more’
Extended Mayo to algorithms; computational precision medicine tools must improve the AI, not just apply it
Recentive Analytics v. Fox Corp.
2025
Applying standard ML to a new data domain is not patent-eligible
Explicitly closes the ‘new application of standard ML’ claim strategy for AI-based diagnostics
Key Takeaways: Section VIII
Mayo did not just affect diagnostics; it created a § 101 framework applied broadly to any claim incorporating a natural relationship, including pharmacogenomic correlations.
Myriad’s practical consequence: precision medicine IP strategy now systematically favors synthetic biology, engineered constructs, and non-natural analogs over naturally occurring sequences.
Recentive (2025) is the current high-water mark for AI patent rejection. Any ML-based precision medicine patent filed after Recentive must be audited against its standard.
EPO and USPTO continue to diverge on gene patent eligibility; EPO generally allows isolated gene patents with demonstrated industrial application, while the USPTO categorically follows Myriad.
Section IX: IP Valuation as a Core Portfolio Asset
How Patent Estates Are Valued in Precision Medicine M&A
Patent portfolios are core financial assets in precision medicine M&A, and their valuation is more complex than standard DCF-adjusted NPV models applied to clinical-stage programs. IP valuation in this context has three primary methodologies: income approach (risk-adjusted NPV of projected royalties or protected revenues), market approach (comparable transaction multiples for similar IP clusters), and cost approach (reproduction cost of the knowledge embedded in the patent estate).
For precision medicine assets, the income approach is dominant but requires careful adjustment for patent strength, remaining patent life, FTO clarity, and competitive moat durability. A compound patent with 15 years of remaining term, clear FTO, and no pending Paragraph IV challenges carries dramatically different value from a patent with 6 years of term, three Paragraph IV certifications filed against it, and pending inter partes review (IPR) petitions at the USPTO.
In biotech M&A, acquirers frequently commission ‘patent landscape analyses’ that go well beyond simple expiry date mapping. These analyses assess: claim scope and enforceability against likely competitive products, FTO across planned therapeutic indications and geographies, Paragraph IV exposure and litigation history, patent term adjustment and PTE status, and the competitive density of the surrounding IP environment (how many third-party patents would need to be licensed or challenged).
Drug-Specific IP Valuation: Keytruda as a Case Study
Merck’s pembrolizumab (Keytruda) is the highest-revenue drug in the world by 2023 sales (~$25 billion annually). Its IP valuation is instructive for understanding precision medicine patent estate architecture.
The Keytruda IP estate covers the PD-1 antibody composition (primary compound patent, originally from Organon, acquired through Schering-Plough merger), method-of-treatment patents covering specific biomarker-defined populations (MSI-H, TMB-H, PD-L1 CPS/TPS thresholds across 40+ indications), dosing regimen patents (flat dosing 200mg Q3W, which replaced weight-based dosing in 2016 and was separately patented), combination regimen patents (pembrolizumab + chemotherapy backbones, pembrolizumab + lenvatinib, pembrolizumab + axitinib), and manufacturing process patents on the antibody production and purification methodology.
The compound patent anticipated LOE is approximately 2028 in the U.S. Method-of-treatment patents on specific biomarker-defined indications extend effective exclusivity for those patient populations beyond 2028. Merck has pursued Paragraph IV litigation against each generic challenger, which typically triggers the 30-month statutory stay, effectively blocking generic entry while litigation proceeds. Pediatric exclusivity on at least one indication adds six months to expiry dates of listed patents on which pediatric studies were requested. The aggregate effective exclusivity picture, accounting for all layers, extends well into the 2030s for certain indications.
Companion Diagnostic IP as a Separate and Valuable Asset
The companion diagnostic (CDx) linked to a precision medicine therapeutic is often an undervalued IP asset. FDA’s CDx approval pathway, formalized under 21 CFR Part 809, requires that the CDx be co-developed and co-approved with the therapeutic for biomarker-defined indications. This regulatory linkage creates a structural IP pairing: the therapeutic cannot be prescribed on-label for the biomarker-defined indication without the CDx.
Companies that own both the therapeutic and the CDx IP (or exclusively license the CDx IP) control the entire treatment paradigm. Roche’s position in precision oncology illustrates this: Roche holds foundational NGS platform IP through Foundation Medicine (acquired 2018), tissue and liquid biopsy CDx IP, and therapeutic IP across multiple oncology products. Any company entering oncology with a biomarker-selected therapy that relies on Foundation Medicine’s assay enters into a dependency relationship with Roche that has both commercial and IP dimensions.
The CDx IP estate is a separate valuation line item in precision medicine M&A. Acquirers should assess: who holds the CDx IP, what is the regulatory approval status of the CDx across relevant geographies, and whether the CDx IP is owned or licensed (and if licensed, on what terms, with what exclusivity provisions).
Key Takeaways: Section IX
Precision medicine patent estate valuation requires analysis across compound patents, method-of-treatment claims, CDx patents, formulation and process patents, and regulatory exclusivity periods — each with distinct valuation inputs.
Keytruda’s IP estate demonstrates that effective exclusivity extends well beyond compound patent LOE through layered method-of-treatment and dosing regimen patents.
CDx IP is a separate, independently valuable asset class that should be assessed as a distinct line item in M&A due diligence.
IPR petitions at the USPTO, Paragraph IV certifications, and pending reexamination are material IP risk factors that must be modeled into patent estate valuations.
Markush Groups for Multi-Biomarker Precision Medicine Patents
Markush group claims allow a single claim to encompass multiple specific compounds, sequences, or structural variants within a defined class. In precision medicine, Markush claiming is the primary strategy for protecting multi-biomarker diagnostic panels, where the commercial product may use a specific subset of markers but the full protective scope should cover functionally equivalent alternative subsets.
The specific application in biomarker panel patents: if a multi-gene predictive model uses 15 genes selected from a larger co-varying cluster of 40 genes, and any subset of 10 to 15 from that cluster produces statistically equivalent clinical performance, the Markush approach allows claiming the functional class of equivalent subsets rather than just the specific 15-gene commercial product. This prevents competitors from developing an ‘almost identical’ test using 13 of the same 15 genes plus 2 alternatives.
The USPTO examines Markush groups for ‘unity of invention’ — the members of the group must share a common structural feature and a common property or activity arising from that shared structure. For gene-based diagnostic panel Markush groups, the common structural feature is the co-variation relationship (the biological mechanism linking the genes to the diagnostic endpoint), and the common property is the equivalent diagnostic performance. Establishing this in the specification with statistical evidence of co-variation and equivalent performance data is essential for Markush claims to survive prosecution.
Evergreening: The Precision Medicine Technology Roadmap
Evergreening — extending effective market exclusivity through secondary patents on improvements, formulations, and new indications — takes specific forms in precision medicine that differ from traditional small-molecule pharma.
Traditional small-molecule evergreening follows a well-mapped playbook: polymorph patents, extended-release formulation patents, metabolite patents, enantiomer patents, new indication patents (often secured under Hatch-Waxman’s three-year new clinical investigation exclusivity), and combination product patents. The Orange Book listing of each secondary patent triggers a 30-month stay upon Paragraph IV certification, providing additional de facto exclusivity.
Precision medicine evergreening has additional layers:
Biomarker expansion patents: After initial approval in a narrow biomarker-defined population, clinical development expands into broader or adjacent biomarker-defined populations. Each expansion is a new method-of-treatment claim, potentially with its own Hatch-Waxman three-year exclusivity period for the new indication and its own Orange Book listing. EGFR mutation-positive NSCLC approval generates patents on treating EGFR exon 19 deletion patients. EGFR exon 21 L858R patients generate separate claims. EGFR-amplified patients generate further claims.
Companion diagnostic upgrade patents: As the CDx technology improves — from IHC-based PD-L1 testing to NGS-based tumor mutational burden quantification to liquid biopsy-based ctDNA mutation profiling — each new CDx methodology generates new patent claims on the improved assay.
Manufacturing process patents: For complex biologics and cell therapies, process improvements (higher yield fermentation conditions, improved purification steps, novel formulation excipients) generate process patents that are listed in the Orange Book if the product is a biologic BLA with reference product status.
Resistance mechanism patents: When patients relapse, the resistance mutations that emerge are often predictable and can be claimed in anticipation — a patent on a method of treating EGFR T790M-mediated resistance to first-generation EGFR inhibitors is the template Astra Zeneca used to protect osimertinib’s position as the dominant third-generation agent.
Paragraph IV Strategy for Precision Medicine Generics
Generic manufacturers entering precision medicine markets face a more complex Paragraph IV filing strategy than traditional small-molecule Hatch-Waxman. In addition to certifying against compound composition-of-matter patents, generics must address Orange Book-listed method-of-treatment patents covering biomarker-defined patient populations — claims that, while often Section 101-vulnerable under Mayo, are still Orange Book-listed and trigger the 30-month stay.
Generics frequently attempt to design around method-of-treatment biomarker patents by obtaining an approval for a narrower indication not covered by the originator’s method claims, then relying on off-label prescribing for the biomarker-selected population. This strategy — ‘skinny labeling’ — has been the subject of significant litigation. GlaxoSmithKline v. Teva (carvedilol, although not precision medicine) established that inducement liability can attach to skinny-label generics if the labeling directs physicians toward a patented use. The skinny-label strategy in precision medicine is therefore higher-risk than in traditional generics, because biomarker-defined uses are often the primary commercial indication, making it difficult to argue that a generic with skinny labeling is not inducing infringement of the biomarker-method patent.
Key Takeaways: Section X
Markush claims for biomarker panels require specification support demonstrating co-variation and equivalent diagnostic performance across the claimed gene/biomarker subsets.
Precision medicine evergreening operates across biomarker expansion, CDx methodology upgrades, manufacturing process improvements, and resistance mechanism patents — each with its own exclusivity period stacking potential.
Skinny-label strategies for precision medicine generics carry substantial inducement liability risk when the biomarker-defined use is the primary commercial indication.
Resistance mechanism patents — filed in anticipation of clinically predictable resistance pathways — are among the highest-value evergreening tools in precision oncology.
A biotech or pharma company’s IP position in precision medicine is not defined by patents alone. The full architecture has four components, each providing distinct protection with distinct duration, enforcement mechanism, and strategic role.
Patents provide the primary legal exclusivity, with the 20-year term from filing and enforcement through district court litigation or ITC proceedings. Trade secrets provide perpetual protection for information that cannot be observed through product analysis or disclosed in regulatory filings — manufacturing process details, cell culture media compositions, proprietary algorithm training datasets, and internal quality control specifications are common precision medicine trade secrets. Trade secret protection depends entirely on reasonable confidentiality measures: NDAs with all employees and contractors, documented access controls, and systematic monitoring of departing employee IP obligations.
Regulatory exclusivity runs parallel to and often outlasts compound patent protection in precision medicine. The U.S. FDA’s exclusivity framework for biologics includes 12 years of reference product exclusivity for approved biologics under the Biologics Price Competition and Innovation Act (BPCIA), during which the FDA cannot approve a biosimilar based on the reference product’s safety and efficacy data. For precision medicine small molecules, NCE exclusivity (5 years for new chemical entities) and new clinical investigation exclusivity (3 years for products requiring new clinical studies for the new indication) layer on top of compound patent protection.
The strategic insight: for many precision medicine biologics, the 12-year BPCIA reference product exclusivity, running from the date of BLA approval, is the binding constraint on biosimilar entry — not the compound patent. Humira (adalimumab) illustrated this at scale: AbbVie’s extensive secondary patent estate (over 100 Orange Book-listed patents) supplemented the base compound exclusivity, but the BPCIA exclusivity period was the structural floor. A similar dynamic will govern future biosimilar CAR-T competition, where manufacturing complexity and BPCIA exclusivity may provide meaningful protection well beyond base patent expiry.
Key Takeaways: Section XI
Trade secret protection for manufacturing processes, training datasets, and quality control specifications provides perpetual protection that patent prosecution cannot replicate.
BPCIA 12-year reference product exclusivity is often the binding constraint on biosimilar entry for precision medicine biologics, not compound patent expiry.
ODD seven-year exclusivity, pediatric six-month extension, and NCE/new clinical investigation exclusivity should be modeled as part of every precision medicine asset’s effective exclusivity calculation.
Regulatory exclusivity and patent protection are cumulative, not mutually exclusive. Building a precision medicine IP strategy requires simultaneous optimization across both dimensions.
Section XII: International Filing Strategy: PCT, EPO Divergence, and Jurisdiction Selection
PCT Filing: The 30-Month Window and Its Strategic Use
The Patent Cooperation Treaty (PCT) allows a single international application to establish a priority date in 150+ contracting states simultaneously. The PCT filing buys 30 months from the priority date before national phase entry is required — 30 months to conduct clinical development, assess commercial potential in each jurisdiction, and decide which national filings justify the significant per-country prosecution costs.
For precision medicine applicants, PCT strategy should sequence as follows: file provisional with initial data to establish U.S. priority; file PCT at or before 12 months from provisional filing (preserving foreign novelty if no earlier public disclosure occurred); use the 30-month national phase entry window to complete Phase 2 clinical data, secure regulatory pre-submission meetings, and refine the commercial geographic priority list; then enter national phase in the highest-value jurisdictions (U.S., EU, Japan, China, South Korea, Australia, Canada, Brazil at minimum for a major precision oncology product).
The PCT process also includes an International Preliminary Examination (IPE) option, which generates a non-binding assessment of patentability from the International Searching Authority. A favorable IPE report can significantly accelerate national phase prosecution by providing examiners in each country with an authoritative prior art search and patentability opinion, reducing prosecution time and cost.
EPO vs. USPTO: Diverging Standards for Precision Medicine
The European Patent Office applies different patentability standards that matter materially for precision medicine applicants. On gene sequences: the EPO, operating under the EU Biotechnology Directive (98/44/EC), permits patents on isolated human gene sequences with demonstrated industrial application, provided the sequence is not claimed in its natural chromosomal environment. This diverges from the post-Myriad USPTO position.
On diagnostic method exclusions: EPC Article 53(c) excludes from patentability ‘methods for treatment of the human or animal body by surgery or therapy and diagnostic methods practised on the human or animal body.’ This exclusion is broader in its literal scope than the U.S. § 101 exclusions, though the EPO’s ‘practised on the human body’ interpretation has been narrowed to require direct technical interaction with the body. A diagnostic method that involves only in vitro analysis of patient samples can avoid the Article 53(c) exclusion, making ex vivo companion diagnostic assays generally patentable at the EPO.
On software and AI: the EPO applies a ‘technical character’ requirement — claims must solve a technical problem by technical means. Pure business methods and mathematical algorithms without technical character are not patentable. But AI systems that produce a technical output (an improved image processing result, a more accurate pattern recognition output) in a technical context can satisfy EPO requirements even if they would fail under the USPTO’s Alice framework. The EPO’s approach to AI patentability is currently more permissive than the post-Recentive USPTO standard for certain claim types.
China: A Distinct and Growing Priority Jurisdiction
China has rapidly become a critical jurisdiction for precision medicine IP. China National IP Administration (CNIPA) has substantially reformed patent examination procedures over the past decade, with increasingly rigorous examination standards and a growing sophistication in pharmaceutical patent prosecution. The Chinese market for precision oncology, liquid biopsy, and cell therapy is substantial and growing, making Chinese national phase entry essential for any globally significant precision medicine program.
Key considerations for China: Chinese patent law historically excluded diagnostic methods and treatment methods from patentability, similar to EPC Article 53(c). Composition-of-matter claims and apparatus claims on diagnostic devices are protectable; pure diagnostic method claims require careful drafting to avoid the prohibition. The Chinese pharmaceutical patent linkage system, introduced in 2021, created a mechanism for pharma patent holders to register and enforce patents against biosimilar and generic applicants in a process analogous to the U.S. Hatch-Waxman framework.
Key Takeaways: Section XII
PCT filing must occur before any public disclosure to preserve all international novelty. The 30-month national phase window is a strategic commercialization assessment period, not a deferral of decision.
EPO standards are more permissive than post-Recentive USPTO standards for AI patent eligibility; consider EPO-first prosecution strategies for AI-heavy precision medicine claims.
Chinese national phase entry is essential for major precision medicine programs; Chinese patent linkage reform (2021) creates Hatch-Waxman analogous enforcement opportunities.
Jurisdiction selection should be driven by market size, IP enforcement quality, healthcare system reimbursement capacity, and disease prevalence in the target indication.
Section XIII: Collaborative R&D, Joint Inventorship, and Co-Development IP Agreements
Academic-Industry Precision Medicine Partnerships: Bayh-Dole and Beyond
The majority of foundational precision medicine IP originates in academic research institutions and is commercialized through university technology transfer offices (TTOs) under the framework established by the Bayh-Dole Act (1980). Bayh-Dole permits universities, small businesses, and nonprofits to retain title to inventions made with federal funding, provided they report inventions promptly, file patent applications, and ensure the U.S. government receives a royalty-free license.
The CAR-T ecosystem illustrates the downstream commercial implications of academic IP origination. The University of Pennsylvania’s foundational CAR-T patents, developed with federal NIH funding, were exclusively licensed to Novartis in a 2012 deal that ultimately contributed to the development of Kymriah. The University of Pennsylvania retained an equity stake in the resulting venture and received milestone and royalty payments. When Kymriah was approved and launched at $475,000 per infusion, the licensing terms embedded in the original Bayh-Dole-compliant license agreement became enormously consequential for both parties.
Any precision medicine company licensing from an academic institution should diligently examine the Bayh-Dole compliance record of the licensed patents. Failure by the academic institution to comply with reporting obligations, filing requirements, or march-in right procedures can affect patent title and create unexpected government license rights in the technology.
Co-Inventorship: The Most Common Source of Precision Medicine Patent Disputes
Joint inventorship is a legally precise concept: each named inventor must have contributed to the conception of at least one claimed invention in the patent. In precision medicine collaborations, where data scientists, physicians, molecular biologists, and biostatisticians all contribute to a predictive model or diagnostic tool, determining who is a co-inventor is genuinely difficult and genuinely consequential.
Named co-inventors automatically hold undivided interest in the patent — meaning each co-inventor can independently license the patent to a third party without permission from or compensation to the other co-inventors, unless a written assignment or employment agreement transfers those rights. A university researcher who is a named co-inventor on a patent and whose assignment to the university TTO is incomplete or legally defective can independently license the same patent to a competing company.
The practical resolution: establish inventorship early, document conception history thoroughly (lab notebooks, timestamped computational analysis records, meeting minutes), and ensure employment agreements and consulting agreements include clear IP assignment clauses that cover all relevant inventions. Companies entering collaborations with academic researchers should require executed IP assignment agreements before any collaborative technical work begins.
Key Takeaways: Section XIII
Bayh-Dole compliance by academic licensors should be verified as a standard due diligence step when licensing foundational precision medicine IP from universities.
Each named co-inventor holds an undivided interest that can be independently licensed without co-inventor consent, absent a written IP assignment. Defective assignment chains create major downstream IP risk.
Inventorship documentation through timestamped lab records, computational analysis logs, and meeting minutes is an operational requirement in collaborative precision medicine research, not an administrative formality.
Consulting agreements with academic collaborators must include comprehensive IP assignment clauses covering all work product and inventions arising from the collaboration.
Section XIV: Investment Strategy for Analysts: Reading Precision Medicine Patent Portfolios {#section-14}
The Analyst’s Precision Medicine IP Due Diligence Framework
For portfolio managers and sell-side analysts, precision medicine patent estate quality is a first-order valuation input, not a legal footnote. The financial impact of patent cliff timing, Paragraph IV exposure, IPR petition outcomes, and regulatory exclusivity stacking can move NPV models by billions of dollars for major precision oncology products.
A rigorous IP due diligence framework for precision medicine assets proceeds through six analytical layers.
The first layer is compound patent mapping: identify all Orange Book-listed patents for the product, map expiry dates with PTA and PTE adjustments, and note any pending PTE applications. For biologics, map the 12-year BPCIA exclusivity window from the BLA approval date.
The second layer is Paragraph IV certification history: identify all Paragraph IV filers, their filing dates, and the current status of resulting litigation. A 30-month stay triggered by a Paragraph IV certification provides de facto exclusivity for the stay duration, but the underlying patent must ultimately survive litigation. Assess the patent’s litigation track record and any IPR petitions at the USPTO (IPR petitions from generics are a leading indicator of planned commercial entry and patent strength concerns).
The third layer is method-of-treatment and CDx patent analysis: identify all Orange Book-listed secondary patents covering biomarker-defined patient populations or dosing regimens, assess their Section 101 vulnerability under the Mayo/Alice framework, and evaluate whether a generic could plausibly use a skinny label to avoid those claims.
The fourth layer is regulatory exclusivity stacking: calculate the sum of all applicable exclusivity periods (NCE, new clinical investigation, ODD, pediatric, BPCIA reference product), identify which exclusivity periods are the binding constraint on generic or biosimilar entry, and model the financial impact of each exclusivity layer’s expiry.
The fifth layer is competitive IP landscape: identify all Paragraph IV filers’ patent estates to assess their 180-day first-to-file generic exclusivity eligibility, and map the competitive product IP landscape to assess substitution risk when the originator’s exclusivity expires.
The sixth layer is IP ownership chain integrity: verify clean assignment chains from inventors to company, identify any pending inventorship disputes or derivation proceedings, and assess Bayh-Dole compliance for any federally funded foundational IP.
Signals of Strong vs. Weak Precision Medicine IP
Strong precision medicine IP has: broad composition-of-matter claims on novel engineered constructs (not natural sequences), multiple Orange Book-listed secondary patents with varied expiry dates, CDx IP ownership or exclusive CDx license, clean FTO across planned indications and geographies, no pending IPR petitions, and a demonstrated track record of successful Paragraph IV litigation against challengers.
Weak precision medicine IP has: narrow compound claims with near-term expiry and no secondary patents, heavy reliance on diagnostic method claims with high § 101 vulnerability under Mayo, CDx dependency on a third-party-owned diagnostic platform without exclusivity provisions, active IPR proceedings challenging key Orange Book patents, and pending inventorship disputes or assignment chain defects.
Key Takeaways: Section XIV
IP due diligence for precision medicine assets requires six distinct analytical layers: compound patents, Paragraph IV history, method-of-treatment/CDx patents, regulatory exclusivity stacking, competitive IP landscape, and ownership chain integrity.
Active IPR petitions against Orange Book-listed patents are a leading indicator of generic entry planning and patent strength concern — material risk factors for financial models.
CDx IP ownership or exclusive licensing is a structural competitive advantage that should be valued separately from therapeutic compound IP.
Analysts who model only compound patent LOE dates, ignoring method-of-treatment, CDx, and regulatory exclusivity stacking, will systematically underestimate effective exclusivity duration for precision medicine leaders.
Section XV: Technology Roadmap: Where Precision Medicine IP Is Heading Through 2032 {#section-15}
Spatial Transcriptomics and Single-Cell Multi-Omics: The Next IP Wave
The next generation of precision medicine IP is forming around technologies that were research-stage in 2020 and are entering clinical translation now. Spatial transcriptomics — which maps gene expression across tissue architecture at single-cell resolution — produces data that neither bulk RNA-seq nor single-cell sequencing captures: the physical location of each cell’s expression profile within the tumor microenvironment. 10x Genomics’ Visium platform and NanoString’s CosMx are the current commercial leaders, with dense IP estates around capture probe designs, spatial barcoding chemistry, and computational deconvolution algorithms.
Clinical applications of spatial transcriptomics in precision medicine will generate method patents on using spatial gene expression patterns to predict treatment response — and those method patents will face the Mayo gauntlet. The natural correlation between spatial tumor heterogeneity and immunotherapy response is a law of nature; the specific spatial transcriptomic assay methodology and the computational pipeline that converts spatial data into a clinically actionable prediction will need to add the ‘significantly more’ inventive step to survive § 101 scrutiny.
Large Language Models in Drug Discovery: IP at the Frontier
Large language models (LLMs) trained on protein sequence databases (ESM-2, AlphaFold2, RoseTTAFold2) and chemical structure databases are rapidly becoming primary drug discovery tools. Generative AI systems like Insilico Medicine’s Chemistry42, AbSci’s protein design platform, and Exscientia’s (now Recursion’s) design platform are generating novel molecular candidates at speeds impossible with traditional high-throughput screening.
The IP implications are acute. An AI-generated molecule, if novel and non-obvious, may be patentable as a composition of matter — but only if a human inventor can claim conception. If the AI system autonomously generated the molecular structure without meaningful human creative direction, the invention may be unpatentable under current USPTO guidance (following the DABUS decisions, which held that AI cannot be a named inventor). Companies using generative AI in drug discovery must design human-AI workflows that preserve demonstrable human inventive contribution at each critical decision point.
Polygenic Risk Scores: Diagnostic Method Patents in a High-Risk § 101 Environment
Polygenic risk scores (PRS) — weighted aggregates of thousands of common genomic variants that together predict disease risk or treatment response — are emerging as a major precision medicine tool. Companies like Genomics PLC, Allelica, and Myriad RBM are developing clinically validated PRS for cardiovascular disease, breast cancer risk, psychiatric disorder susceptibility, and pharmacogenomic response prediction.
PRS-based clinical decision tools face a particularly challenging § 101 landscape. Each individual variant in a PRS is a naturally occurring polymorphism. The aggregate correlation between the weighted sum of variants and clinical outcome is a complex natural relationship. The computational model that generates the PRS is an algorithm. Under Mayo, Myriad, and Alice, all three layers are § 101-vulnerable. The viable patent strategy focuses on: the specific weighting algorithm and its technical improvements over existing statistical methods, the non-conventional data processing pipeline that handles LD pruning, ancestry correction, and cross-population calibration, and the integrated clinical decision support system that translates PRS output into a specific clinical recommendation.
Cell and Gene Therapy Manufacturing: The Next Process Patent Frontier
As CAR-T, TCR-T, and allogeneic NK cell therapies scale from academic centers to commercial manufacturing, process IP is becoming the dominant value driver. The commercial differentiation between autologous CAR-T manufacturers is increasingly in manufacturing efficiency — turnaround time from leukapheresis to infusion, cell expansion protocols, cryopreservation conditions, and final product consistency metrics.
Automated closed-system manufacturing platforms (Lonza’s Cocoon, Miltenyi’s CliniMACS Prodigy, Cytovance’s bioreactor systems) are building IP estates around the specific engineering that enables automated, scalable cell therapy manufacturing. Those process patents, if listed on a BLA for an approved biologic CAR-T product, will be enforceable through the BPCIA litigation pathway against biosimilar applicants who reference the same manufacturing process.
Key Takeaways: Section XV
Spatial transcriptomics clinical applications will generate the next wave of Mayo-challenged diagnostic method patents; claim drafting must front-load the non-conventional assay methodology and computational pipeline as the inventive concept.
Generative AI drug discovery creates a human inventorship documentation challenge: companies must establish workflow protocols that preserve demonstrable human inventive contribution at each AI-assisted design step.
PRS-based clinical tools face layered § 101 exposure (natural variants + natural correlation + algorithm); viable claims focus on technical improvements to the weighting algorithm and clinical decision support system integration.
Cell therapy manufacturing process IP, listed as biologic product patents on BLAs, is the next major BPCIA enforcement frontier as the allogeneic CAR-T market develops.
Comprehensive Summary: The Precision Medicine Patent Playbook
Precision medicine’s commercial value is inseparable from its IP architecture. The field produces not one patent per product but estates of dozens to hundreds of interdependent claims spanning compound composition, biomarker-defined method of treatment, companion diagnostic methodology, manufacturing process, and AI-driven analytical tools. Each layer has distinct patentability challenges, distinct exclusivity duration, and distinct enforcement strategy.
The post-Mayo, post-Myriad, post-Recentive legal environment demands claim drafting that consistently emphasizes engineered constructs over natural sequences, technical improvement to AI systems over new application of standard methods, non-conventional measurement or analytical methodologies over natural correlations applied with routine techniques, and inventive combinations with unexpected synergistic results over predictable aggregations of prior art elements.
For pharmaceutical IP teams, portfolio managers, and R&D leads, the operational imperative is integration: IP strategy must be embedded in clinical development planning (when to file provisionals, how to time publications), in business development (how to structure CDx co-development agreements), in M&A (how to value layered patent estates), and in regulatory strategy (how to stack ODD, NCE, pediatric, and BPCIA exclusivities for maximum effective protection).
The market will reach approximately $247 billion by 2029. The companies that capture the largest share of that market will be those that understand, from program inception, that the IP estate is not a legal overhead cost. It is the primary commercial asset.
