FUNCTIONAL PRECISION MEDICINE
Published evidence.
The field's evidence shows strong signals in small evaluable cohorts. Every figure below comes from the source it cites, with its evidence tier labeled, peer-reviewed journal, preprint, or conference abstract, and its design and limitations stated.
Clinical Studies of FPM Platforms
Study
Platform
Design
Population
Cohort
Key finding, as published
Image-based single-cell functional screening (scFPM).
Prospective, single-arm, open-label.
Advanced aggressive hematologic cancers.
143 analyzed; 56 (39%) treated per assay.
54% of treated patients reached a PFS ratio of at least 1.3 versus their own prior therapy, meeting the primary endpoint; 12 patients (40% of responders) were exceptional responders; overall-survival benefit versus a physician's-choice cohort.
Patient-derived organoids.
Prospective living biobank, matched to patients' responses in phase I and II trials.
Metastatic gastrointestinal cancers.
Biobank cohort; per-patient matched comparisons (see paper).
Organoid drug response forecast the patients' actual clinical responses with 100% sensitivity, 93% specificity, positive predictive value 88%, and negative predictive value 100%.
Mass response study Stevens et al., JCO Precision Oncology, 2024 †
Source:
DOI
10.1200/PO.23.00349.
Cellular mass response (MRT).
Prospective specimen collection; assay result compared against response to physician-selected therapy.
Solid (69) and hematologic (35) malignancies.
104 collected; 81 (78%) qualified; 41 (51%) evaluable.
Assay prediction and clinical response were concordant: odds ratio 14.80 (95% CI, 2.83 to 102.9; P = .0003); positive predictive value 0.83; overall accuracy 0.80.
Ex vivo drug sensitivity testing with genomics.
Prospective observational (NCT03860376).
Relapsed or refractory pediatric cancers.
25 enrolled; 21 (84%) tested.
Functional testing was completed even in heavily pretreated pediatric specimens, and 100% of tested patients had at least one effective agent identified ex vivo (drug sensitivity score above 10).
† Related-party disclosure: the mass response study was conducted by Travera, which NCPO CEO Clifford Reid founded and on which he is a co-author. NCPO is not a party to any study in this table.
The Genomics Benchmark
FPM complements genomic testing rather than replacing it. The landmark genomic trial sets the scale of the gap functional testing addresses: for the majority of patients whose panel finds nothing actionable, a functional test asks a different question of the same tumor.
Study
Platform
Design
Population
Cohort
Key finding, as published
Next-generation sequencing, tumor profiling.
Prospective master protocol.
Refractory solid tumors, lymphomas, myeloma.
5,954 specimens; profiling successful in 93.0%; 5,540 sequenced.
An actionable alteration was found in 37.6% of profiled patients; 17.8% (985 of 5,540) were assigned to a treatment arm. Seven of 27 reported arms met their primary endpoint (objective response 14% to 50%); across all reported arms, 10.3% of evaluable treated patients responded (79 of 765).
‡ Related-party disclosure: Keith Flaherty is first author of NCI-MATCH and chairs the NCPO Oncology Advisory Board.
Platform-Specific Validation, by Company
NCPO is test-agnostic. The table below lists validation evidence by company for Treatment Selection Tests in the landscape NCPO orchestrates, in alphabetical order. Listing is descriptive, not an endorsement. Evidence tiers are labeled: peer-reviewed journal articles, preprints, and conference abstracts; preprints and abstracts have not undergone full peer review.
Company & platform
Modality
Cohorts & key figures
Evidence tier
Ex vivo drug testing with genomics.
Prospective pediatric study: 25 enrolled, 21 tested (84%); at least one effective agent identified ex vivo in all tested patients.
Combinatorial ex vivo drug testing.
Completed prospective NHL study: 126 tested cases, 105 evaluable; overall test accuracy 74.5%; 59% overall response rate with QPOP-guided off-label combinations; 59.3% of guided patients exceeded their prior response duration; longer two-year PFS than salvage therapy. Earlier prospective NHL cohort: 71 enrolled; reports for 67 of 75 samples; six-day median turnaround; of 17 patients treated per report, five complete responses. AML: reports in 55 of 63 samples (87.3%); five-day median turnaround; sensitivity 83.3%, specificity 90.9%, accuracy 86.2% against clinical outcomes.
Patient-derived organoid pharmacotyping.
Peer-reviewed case reports of organoid-guided durable responses, and an organoid drug-screening cohort in colorectal peritoneal metastases. Cohort-level clinical correlations reported to date as conference abstracts: 403 patients across 42 tumor types with a potentially effective drug identified ex vivo in 94%; real-world overall-survival observations; a prospective ovarian correlation study (PROSPERITY).
Transcriptomic machine-learning prediction (computational, not ex vivo).
Retrospective multi-cohort external validation of a transcriptomics-guided model for treatment-response prediction in breast cancer; model trained on 11,414 patients across 15 cancer types and 150 regimens. Pancreatic application described in a 2026 conference abstract.
3D microtumor ex vivo functional profiling.
17 ovarian cancer tissues exposed ex vivo to NCCN-recommended drugs and combinations; statistically significant interpatient heterogeneity in drug efficacy (P < .001). Ex vivo characterization; no clinical-outcome correlation reported yet.
Cell-mass response, ex vivo.
41 evaluable patients; odds ratio 14.80; positive predictive value 0.83; overall accuracy 0.80. Malignancy-agnostic assay pipeline described in a peer-reviewed methods paper; HIPEC agent selection and immuno-oncology applications reported as conference abstracts.
† Related-party disclosure: NCPO CEO Clifford Reid founded Travera, is a co-author of the Stevens paper (JCO Precision Oncology. 2024;8:e2300349), and is a co-author on the ASCO 2024 abstract (abstract 3132). NCPO is not a party to any study in this table, and inclusion of any company is not an endorsement.
The studies above are single-arm, observational, retrospective, or
ex vivo only. Their endpoints differ and their evaluable cohorts are small. Assay platforms are not interchangeable, and a result from one does not validate another.
Every NCPO-supported case is enrolled in
CCCS001 (NCT07343024), a registered, site-less observational study sponsored by Cancer Commons. The predictive accuracy of each treatment selection test is tracked and will be published as evidence accumulates.
Last update: August 2026
