PubMed Daily Triage Digest — 2026-06-25

Window: 2026-06-24 09:00 UTC → 2026-06-25 09:00 UTC
Topics scanned: 9 + sentinel
Candidates reviewed: ~152 PMIDs
Articles emitted: 25 (5 HIGH, 20 STANDARD)
HIGH alerts sent: 5 (Telegram)

🔴 HIGH PRIORITY (Score ≥ 8)

1. ⭐⭐⭐ [SCORE 10] Ten-Year Outcomes after CAR T-Cell Therapy for B-Cell Lymphomas

PMID: 42341302 | Journal: N Engl J Med (2026 Jun 25) | Topics: T1 (Hematologic malignancies), T6 (CAR-T)
DOI: 10.1056/NEJMoa2518035
Authors: Ruella M, Paruzzo L, Chong ER, Chong EA et al. (Schuster SJ; UPenn)

Key Finding:
At median 10.1-year follow-up (n=38; 24 DLBCL + 14 FL), tisagenlecleucel (anti-CD19, 4-1BB-costimulated) achieves:

• No relapses beyond 5.4 years — establishes a "cure threshold"
• 10-yr lymphoma-free survival: 32% (DLBCL), 47% (FL)
• 10-yr OS: 17% (DLBCL), 50% (FL)
• 21% cumulative incidence of second primary cancers; 18% non-relapse mortality
• Higher CAR-transgene persistence associated with long-term response
• Persistent B-cell aplasia in 44% long-term responders

Why it matters: First 10-year outcomes data for CAR-T in B-cell lymphoma. Definitively establishes curative potential — ~1 in 3 DLBCL, ~1 in 2 FL patients achieve decade-long remission from a single infusion. Landmark for the field; will inform eligibility criteria and long-term survivorship protocols.

2. ⭐⭐⭐ [SCORE 9] Disparate Privacy Risks from Medical AI

PMID: 42343130 | Journal: Nature (2026 Jun 24) | Topics: T4 (AI/ML diagnostics)
DOI: 10.1038/s41586-026-10688-0
Authors: Knolle MA, Menten MJ, Jungmann F, Meissen F, Glocker B, Rueckert D, Kaissis G (TUM / Imperial College)

Key Finding:
First patient-level privacy audit of medical AI. Membership inference attacks (MIAs) against diagnostic AI models:

• Achieve near-perfect success rates for individual patients even when aggregate metrics appear random
• Privacy risk increases substantially with model capacity
• Underrepresented groups (by disease status, race, insurance, sex, imaging protocol) face disproportionately high attack success rates
• Aggregate privacy metrics severely underestimate individual privacy risk

Why it matters: Fundamental safety concern for deployed medical AI systems. Shows that standard aggregate privacy metrics provide false assurance. Directly relevant to any institution using AI trained on patient imaging or clinical data. Will drive regulatory and technical development of patient-level privacy protections.

3. ⭐⭐⭐ [SCORE 9] Automated Reanalysis of Genomic Data for Rare Disease Diagnostics at Scale

PMID: 42343115 | Journal: Nat Med (2026 Jun 24) | Topics: T9 (Rare diseases), T5 (Precision genomics)
DOI: 10.1038/s41591-026-04477-5
Authors: Welland MJ, Ahlquist KD, De Fazio P et al. (Stark Z, MacArthur DG; Garvan/Broad/VCGS)

Key Finding:
Talos (open-source automated variant prioritization tool):

• Integrates dynamically updated gene-disease + variant-level evidence with inheritance-aware filtering
• Validation in 1,089 individuals: 90% sensitivity for known diagnoses at 1.3 variants/case average
• Iterative monthly reanalysis: 1 variant per 200 cases burden
• Applied to 4,735 undiagnosed patients: 241 new diagnoses (5.1% yield)
  • 32% from new gene-disease relationships; 22% new variant evidence; 45% improved analysis strategies

Why it matters: Solves the scalability bottleneck for rare disease genomic reanalysis. Automated, open-source, iterative — demonstrated 5.1% additional diagnostic yield in previously undiagnosed patients. Will accelerate rare disease diagnosis globally. Microsoft Research co-authorship. Immediate clinical translation potential.

4. ⭐⭐ [SCORE 8] Multimodal Blood-Based Proteomic Profiling Reveals Insights into Mechanisms of Immunotherapy Resistance

PMID: 42342704 | Journal: Nat Commun (2026 Jun 24) | Topics: T6 (Immunotherapy), T3 (Liquid biopsy)
DOI: 10.1038/s41467-026-74790-7
Authors: Wright SJ, Zamora I, Parikh M et al. (Hacohen N, Mehta A, Boland GM; Broad Institute / MGH / Harvard)

Key Finding:
Plasma proteomic profiling (>2,900 proteins) + mass cytometry + scRNA/bulk RNA in 250 metastatic melanoma patients on ICB:

• Proteins upregulated post-ICB → effector immune pathway activation
• Non-responder proteins → suppressive myeloid subsets and malignant cell expression
• Spatial transcriptomics confirm immunosuppressive/pro-tumor interactions
• Epithelial-specific proteins in responders post-ICB → correlate with toxicity (healthy tissue degradation)

Why it matters: Largest multimodal blood-based ICB biomarker study. Identifies actionable peripheral biomarker signatures for primary resistance prediction, early progression detection, and toxicity anticipation. Paired blood + tumor data with serial sampling. Platform translatable across cancer types.

5. ⭐⭐ [SCORE 8] Effects of Exercise and Liraglutide on Vascular Health during Weight Loss Maintenance: S-LiTE Trial

PMID: 42342869 | Journal: Nat Metab (2026 Jun 24) | Topics: T7 (GLP-1/cardiometabolic)
DOI: 10.1038/s42255-026-01554-4
Authors: Sandsdal RM, Holt J et al. (Torekov SS; University of Copenhagen)

Key Finding:
Prespecified secondary analysis of S-LiTE RCT (n=130; exercise ±liraglutide for 52 weeks weight maintenance):

• Exercise (alone or combined) → ↓ carotid intima-media thickness, ↓ IL-6, ↓ IFN-γ
• Exercise + liraglutide → also improves endothelial biomarkers (↓ sICAM-1, sVCAM-1, tPA)
• Liraglutide alone: no significant vascular improvements
• Conclusion: physical activity is essential for vascular benefits; GLP-1R agonists alone insufficient

Why it matters: Directly challenges the assumption that GLP-1R agonists provide vascular protection independent of exercise. Nature Metabolism publication; RCT-level evidence. Practical implication: patients on semaglutide/liraglutide need concurrent exercise prescription for cardiovascular benefit. Will influence clinical guidelines.

Additional HIGH-Scoring Findings (Score 8, Alert-threshold)

[Score 8] TCR-mimic Bispecific Nanobody T Cell Engager (Signal Transduct Target Ther)
PMID: 42342658 | DOI: 10.1038/s41392-026-02745-x | Topics: T6
First-in-class TCRm Bi-NbTE (bispecific VHH-VHH) targets intracellular tumor antigens (HLA-A2/WT1, HLA-A2/GPC3) via pMHC I complexes. Overcomes limitation of conventional bispecific T cell engagers restricted to extracellular antigens. Strong preclinical efficacy in CDX and PDX models; clinically translatable platform.

[Score 8] Phase II FT14 Conditioning for AML Allo-HSCT in CR1 (Bone Marrow Transplant)
PMID: 42342968 | DOI: 10.1038/s41409-026-02937-7 | Topics: T1
82 patients, multicenter prospective; treosulfan+fludarabine (FT14) myeloablative conditioning for AML allo-HSCT in CR1 (ages 40-65). 1-yr LFS 81.7%, CIR 14.9%, near-zero NRM. Supports FT14 as well-tolerated alternative to busulfan-based conditioning.

[Score 8] LUNA25 AI vs 65 Radiologists for Lung Nodule Malignancy (Radiol Artif Intell)
PMID: 42340186 | DOI: 10.1148/ryai.260179 | Topics: T4, T3
AI AUC 0.78 vs radiologist average 0.70 (P=0.001) for 5-15mm indeterminate nodules on screening LDCT. AI: 12% more malignancies detected at matched specificity, 20% fewer false positives at matched sensitivity. External validation on European trial data.

[Score 8] Transferrin + Chemotherapy Improves AML Outcomes (Sci Transl Med)
PMID: 42341081 | DOI: 10.1126/scitranslmed.adu0167 | Topics: T1
ApoTF (iron-free apotransferrin) redistributes toxic NTBI, enhances bone marrow erythropoiesis, improves adaptive immunity. In AML mouse models: apoTF + chemo reduces leukemic burden and improves survival (immune-dependent). Also reduces sepsis mortality in leukemic mice receiving chemo. Mechanism: suppressed CCL2/IL-6 via macrophage polarization.

🟡 STANDARD PRIORITY (Score 6-7)

Run Statistics

Generated by pubmed-triage cron · Friday, 2026-06-25 09:00 UTC