{ "brief_id": "SPR-2026-F0FA", "generated_at": "2026-07-20", "domains": [ "Gastroenterology", "Pulmonology" ], "original_hypothesis": "The computational framework for identifying robust, data-driven disease subtypes in Alzheimer's disease (SuStaIn) can be transferred to pulmonary function diagnostics to identify mechanistically distinct, reproducible subtypes of obstructive and restrictive lung diseases, moving beyond the current consensus of treating them as monolithic entities.", "grounding": { "novelty_assessment": { "score": 0.85, "closest_existing_work": [ { "paper_id": "f70855235912e70c29b0884b39ce63d1c90ecb8a", "title": "Non-emphysematous chronic obstructive pulmonary disease is associated with diabetes mellitus", "doi": "10.1186/1471-2466-14-164", "year": 2014, "similarity": "related", "key_difference": "This paper uses a binary clinical distinction (airway-predominant vs. emphysema-predominant) based on CT, not a data-driven algorithm like SuStaIn that identifies multiple progression trajectories from longitudinal data.", "authors": [ "C. Hersh", "B. Make", "D. Lynch", "R. Barr", "R. Bowler", "P. Calverley", "P. Castaldi", "M. Cho", "H. Coxson", "D. Demeo", "M. Foreman", "M. Han", "Benjamin Harshfield", "J. Hokanson", "S. Lutz", "J. Ramsdell", "E. Regan", "S. Rennard", "Joyce D. Schroeder", "F. Sciurba", "R. Steiner", "R. Tal-Singer", "E. V. van Beek", "E. Silverman", "J. Crapo", "the COPDGene Investigators" ] }, { "paper_id": "6594d8ad2fde347f05083ff010ae828c70bd875c", "title": "Subtyping COPD using visual and quantitative CT features.", "doi": "10.1016/j.chest.2019.06.015", "year": 2020, "similarity": "related", "key_difference": "This paper creates static CT-based subtypes, not dynamic progression subtypes derived from longitudinal PFT data. It does not use the SuStaIn algorithm or model temporal trajectories.", "authors": [ "Jinkyeong Park", "B. Hobbs", "J. Crapo", "B. Make", "E. Regan", "S. Humphries", "V. Carey", "D. Lynch", "E. Silverman" ] }, { "paper_id": "fc599ec0092f36607f8719686cdc37118ab53313", "title": "Subtypes of COPD Have Unique Distributions and Differential Risk of Mortality.", "doi": "10.15326/jcopdf.6.5.2019.0150", "year": 2019, "similarity": "related", "key_difference": "This paper develops continuous multidimensional axes (airway vs. emphysema) rather than discrete progression subtypes. It does not use SuStaIn or model stage-based progression.", "authors": [ "K. Young", "E. Regan", "M. Han", "S. Lutz", "M. Ragland", "P. Castaldi", "G. Washko", "M. Cho", "M. Strand", "D. Curran-Everett", "T. Beaty", "R. Bowler", "E. Wan", "D. Lynch", "B. Make", "E. Silverman", "J. Crapo", "J. Hokanson", "G. Kinney" ] }, { "paper_id": "f69baafda7d1ba86dd6819c9ad42d3a2c4899e88", "title": "Clustering-based COPD subtypes have distinct longitudinal outcomes and multi-omics biomarkers", "doi": "10.1136/bmjresp-2021-001182", "year": 2022, "similarity": "related", "key_difference": "This paper uses k-means clustering on cross-sectional data to define subtypes, not a progression trajectory model like SuStaIn. It does not infer stages or temporal ordering of events.", "authors": [ "A. Gregory", "Zhonghui Xu", "K. Pratte", "Sool Lee", "Congjian Liu", "R. Chase", "J. Yun", "A. Saferali", "C. Hersh", "R. Bowler", "E. Silverman", "P. Castaldi", "A. Boueiz" ] } ], "verdict": "novel" }, "evidence_base": [ { "paper_id": "dbfe91194a4c7c8fa7656e76cd18d37036d0729f", "title": "Neurostructural subgroup in 4291 individuals with schizophrenia identified using the subtype and stage inference algorithm", "doi": "10.1038/s41467-024-50267-3", "year": 2024, "citation_count": 47, "support_type": "analogous", "relevance": "Directly demonstrates that SuStaIn can be successfully transferred from Alzheimer's disease to another neurological disorder (schizophrenia), validating the core methodological transferability claim of the hypothesis.", "key_finding": "SuStaIn identified two reproducible neurostructural subgroups in schizophrenia from cross-sectional brain imaging data, showing the algorithm generalizes beyond AD.", "authors": [ "Yuchao Jiang", "Cheng Luo", "Jijun Wang", "L. Palaniyappan", "Xiao Chang", "Shitong Xiang", "Jie Zhang", "Mingjun Duan", "Huan Huang", "C. Gaser", "Kiyotaka Nemoto", "Kenichiro Miura", "Ryota Hashimoto", "L. Westlye", "G. Richard", "Sara Fernandez-Cabello", "N. Parker", "O. Andreassen", "T. Kircher", "I. Nenadić", "F. Stein", "F. Thomas-Odenthal", "L. Teutenberg", "P. Usemann", "U. Dannlowski", "Tim Hahn", "D. Grotegerd", "S. Meinert", "R. Lencer", "Yingying Tang", "Tianhong Zhang", "Chunbo Li", "Weihua Yue", "Yuyanan Zhang", "Xin Yu", "Enpeng Zhou", "Ching-Po Lin", "Shih-Jen Tsai", "A. Rodrigue", "David E. Glahn", "Godfrey D. Pearlson", "J. Blangero", "A. Karuk", "E. Pomarol-Clotet", "R. Salvador", "P. Fuentes-Claramonte", "M. Garcia-Leon", "G. Spalletta", "F. Piras", "Daniel Vecchio", "N. Banaj", "Jingliang Cheng", "Zhening Liu", "Jie Yang", "A. Gonul", "Ozgul Uslu", "B. B. Burhanoglu", "Aslihan Uyar Demir", "K. Rootes-Murdy", "V. Calhoun", "Kang Sim", "Melissa J Green", "Y. Quidé", "Y. Chung", "Woo-Sung Kim", "S. Sponheim", "C. Demro", "I. Ramsay", "F. Iasevoli", "A. de Bartolomeis", "A. Barone", "M. Ciccarelli", "Arturo Brunetti", "S. Cocozza", "G. Pontillo", "M. Tranfa", "M. T. M. Park", "Matthias Kirschner", "F. Georgiadis", "S. Kaiser", "T. V. Van Rheenen", "S. L. Rossell", "Matthew Hughes", "William Woods", "S. Carruthers", "P. Sumner", "E. Ringin", "F. Španiel", "A. Škoch", "D. Tomeček", "Philipp Homan", "S. Homan", "W. Omlor", "G. Cecere", "Dana Nguyen", "Adrian Preda", "S. Thomopoulos", "N. Jahanshad", "Long-Biao Cui", "Dezhong Yao", "Paul M. Thompson", "J. A. Turner", "T. V. van Erp", "W. Cheng", "Andrea Tamsyn E. de Bartolomeis Van Rheenen", "A. de Bartolomeis", "Jianfeng Feng" ] }, { "paper_id": "dd9e3060fa7fa5402f4df5f5a2aa214fcaf68a70", "title": "Identification of different MRI atrophy progression trajectories in epilepsy by subtype and stage inference", "doi": "10.1093/brain/awad284", "year": 2023, "citation_count": 41, "support_type": "analogous", "relevance": "Provides another example of SuStaIn being applied to a non-AD neurological condition (epilepsy), further supporting the claim that the algorithm is a general-purpose tool for identifying progression subtypes.", "key_finding": "SuStaIn identified distinct spatiotemporal progression patterns of brain atrophy in epilepsy, demonstrating its utility for data-driven disease taxonomy in a new domain.", "authors": [ "F. Xiao", "L. Caciagli", "B. Wandschneider", "Daichi Sone", "A. Young", "S. Vos", "G. Winston", "Yingying Zhang", "Wenyu Liu", "D. An", "B. Kanber", "Dong Zhou", "J. Sander", "Maria Thom", "John S Duncan", "Daniel C. Alexander", "M. Galovic", "M. Koepp" ] }, { "paper_id": "fbfaa5078f8efdd5fc7c1d01f6100a90b86c9b7c", "title": "Uncovering distinct progression patterns of tau deposition in progressive supranuclear palsy using [18F]Florzolotau PET imaging and subtype/stage inference algorithm", "doi": "10.1016/j.ebiom.2023.104835", "year": 2023, "citation_count": 19, "support_type": "analogous", "relevance": "Extends the evidence for SuStaIn's transferability to a tauopathy (PSP), showing the algorithm works across different pathologies and imaging modalities (PET), strengthening the case for its application to PFT data.", "key_finding": "SuStaIn identified distinct in vivo spatiotemporal tau trajectories in PSP, contributing to precision medicine for this condition.", "authors": [ "Jimin Hong", "Jiaying Lu", "Fengtao Liu", "Min Wang", "Xinyi Li", "Christoph Clement", "Leonor Lopes", "Matthias Brendel", "A. Rominger", "T. Yen", "Y. Guan", "Mei Tian", "Jian Wang", "Chuantao Zuo", "Kuangyu Shi", "Jian Wang", "Fengtao Liu", "Chuantao Zuo", "Jianjun Wu", "Yimin Sun", "P. Wu", "Yilin Tang", "Jue Zhao", "Bin Wu", "Bo Shen", "Jiaying Lu", "Xin-Yue Zhou", "Huiwei Zhang", "J. Ge", "Minjia Chen", "Zizhao Ju" ] }, { "paper_id": "f27199e8a99273d2e01d97225fde8bf5928e5c9e", "title": "Transferability of Alzheimer's disease progression subtypes to an independent population cohort", "doi": "10.1016/j.neuroimage.2023.120005", "year": 2023, "citation_count": 9, "support_type": "analogous", "relevance": "Directly addresses the 'transferability' aspect of the hypothesis, showing that SuStaIn-derived AD subtypes are reproducible across independent cohorts, a key requirement for the proposed pulmonary application.", "key_finding": "SuStaIn models trained on different cohorts consistently recovered the same AD atrophy subtypes, demonstrating cross-cohort generalizability.", "authors": [ "Hanyi Chen", "A. Young", "N. Oxtoby", "F. Barkhof", "Daniel C. Alexander", "A. Altmann" ] }, { "paper_id": "f70855235912e70c29b0884b39ce63d1c90ecb8a", "title": "Non-emphysematous chronic obstructive pulmonary disease is associated with diabetes mellitus", "doi": "10.1186/1471-2466-14-164", "year": 2014, "citation_count": 73, "support_type": "indirect", "relevance": "Provides evidence for the existence of clinically meaningful COPD subtypes (airway-predominant vs. emphysema-predominant) with distinct comorbidities, supporting the premise that COPD is heterogeneous and subtyping is valuable.", "key_finding": "Non-emphysematous COPD is associated with an increased risk of diabetes, suggesting distinct pathophysiological mechanisms between subtypes.", "authors": [ "C. Hersh", "B. Make", "D. Lynch", "R. Barr", "R. Bowler", "P. Calverley", "P. Castaldi", "M. Cho", "H. Coxson", "D. Demeo", "M. Foreman", "M. Han", "Benjamin Harshfield", "J. Hokanson", "S. Lutz", "J. Ramsdell", "E. Regan", "S. Rennard", "Joyce D. Schroeder", "F. Sciurba", "R. Steiner", "R. Tal-Singer", "E. V. van Beek", "E. Silverman", "J. Crapo", "the COPDGene Investigators" ] }, { "paper_id": "6594d8ad2fde347f05083ff010ae828c70bd875c", "title": "Subtyping COPD using visual and quantitative CT features.", "doi": "10.1016/j.chest.2019.06.015", "year": 2020, "citation_count": 69, "support_type": "indirect", "relevance": "Demonstrates that combining CT and spirometry data yields non-overlapping COPD subtypes with different progression rates, supporting the hypothesis that integrating imaging with PFTs can reveal distinct disease trajectories.", "key_finding": "A combined visual and quantitative CT classification produced ten non-overlapping COPD subtypes with differences in physiology and longitudinal disease progression.", "authors": [ "Jinkyeong Park", "B. Hobbs", "J. Crapo", "B. Make", "E. Regan", "S. Humphries", "V. Carey", "D. Lynch", "E. Silverman" ] }, { "paper_id": "fc599ec0092f36607f8719686cdc37118ab53313", "title": "Subtypes of COPD Have Unique Distributions and Differential Risk of Mortality.", "doi": "10.15326/jcopdf.6.5.2019.0150", "year": 2019, "citation_count": 39, "support_type": "indirect", "relevance": "Shows that COPD subtypes (airway-predominant vs. emphysema-predominant) have unique mortality risks and associations with biomarkers, reinforcing the clinical relevance of subtyping in COPD.", "key_finding": "Individuals in the highest mortality risk deciles on the airway-predominant and emphysema-predominant axes have unique spirometric patterns, imaging characteristics, and causal mortality.", "authors": [ "K. Young", "E. Regan", "M. Han", "S. Lutz", "M. Ragland", "P. Castaldi", "G. Washko", "M. Cho", "M. Strand", "D. Curran-Everett", "T. Beaty", "R. Bowler", "E. Wan", "D. Lynch", "B. Make", "E. Silverman", "J. Crapo", "J. Hokanson", "G. Kinney" ] }, { "paper_id": "f69baafda7d1ba86dd6819c9ad42d3a2c4899e88", "title": "Clustering-based COPD subtypes have distinct longitudinal outcomes and multi-omics biomarkers", "doi": "10.1136/bmjresp-2021-001182", "year": 2022, "citation_count": 13, "support_type": "indirect", "relevance": "Provides evidence that data-driven COPD subtypes (from k-means clustering) have distinct longitudinal outcomes and multi-omics biomarkers, supporting the hypothesis that such subtypes are mechanistically distinct.", "key_finding": "COPD k-means subtypes demonstrated varying rates of disease progression, prospective comorbidities, mortality, and associations to transcriptomic and proteomic biomarkers.", "authors": [ "A. Gregory", "Zhonghui Xu", "K. Pratte", "Sool Lee", "Congjian Liu", "R. Chase", "J. Yun", "A. Saferali", "C. Hersh", "R. Bowler", "E. Silverman", "P. Castaldi", "A. Boueiz" ] }, { "paper_id": "95b755e2653de70830e3fff3043bfc16d734e515", "title": "Mucus Plugs as Precursors to Exacerbation and Lung Function Decline in COPD Patients.", "doi": "10.1016/j.arbres.2024.07.017", "year": 2024, "citation_count": 18, "support_type": "indirect", "relevance": "Identifies a specific imaging feature (mucus plugs) as a predictor of accelerated FEV1 decline, suggesting a potential 'mucus-predominant' progression subtype that could be discovered by SuStaIn.", "key_finding": "Mucus plugs are associated with accelerated FEV1 declines over 5 years, identifying a potential prognostic marker for a specific disease trajectory.", "authors": [ "Kwang Nam Jin", "H. Lee", "Heemoon Park", "Jung-Kyu Lee", "E. Heo", "Deog Kyeom Kim", "Hyun Woo Lee" ] }, { "paper_id": "32e65d6b0ec99e31ff071021a6acedd969ebed5a", "title": "Catalase-encapsulated matrix metalloproteinase-9 responsive nanogels for modulation of inflammatory response and treatment of neutrophilic asthma", "doi": "10.1186/s12951-025-03470-3", "year": 2025, "citation_count": 8, "support_type": "indirect", "relevance": "Explicitly mentions 'neutrophilic asthma' as a distinct, more severe subtype of asthma, supporting the hypothesis that asthma is not monolithic and that identifying such subtypes via data-driven methods is clinically relevant.", "key_finding": "Neutrophilic asthma is a corticosteroid-resistant subtype, and targeted therapy (M-CAT-NGs) can effectively manage it.", "authors": [ "Xiaoping Guo", "Xu Zuo", "Wenxue Zheng", "Dan Zhao", "Chao Dong", "Zheng Zou", "Yuanyuan Shen", "Caina Xu", "Chaoliang He", "Fang Wang" ] } ], "counter_evidence": [ { "paper_id": "12f1290c4033fa8191634c5f43bc8b44793d321f", "title": "Bio-inspired self-agitator for convective heat transfer enhancement", "doi": "10.1063/1.5046502", "year": 2018, "finding": "This paper is about heat transfer enhancement and is completely unrelated to pulmonology, Alzheimer's disease, or the SuStaIn algorithm. It provides no evidence for or against the hypothesis.", "severity": "minor", "authors": [ "Zheng Li", "Xianchen Xu", "Kuojiang Li", "Yangyang Chen", "Zhaoqing Ke", "Sheng Wang", "Hsiu-hung Chen", "Guoliang Huang", "Chung-Lung Chen", "C. Chen" ] }, { "paper_id": "a786672698bda9b909224a4e1d60c9ec2cd50058", "title": "Modeling the Impact of Delaying the Diagnosis of Non-Small Cell Lung Cancer During COVID-19", "doi": "10.1016/j.athoracsur.2020.08.025", "year": 2020, "finding": "This paper models diagnostic delay in lung cancer, not disease subtyping or progression modeling. It is tangential to the hypothesis.", "severity": "minor", "authors": [ "M. Shipe", "D. Haddad", "S. Deppen", "B. Kozower", "E. Grogan" ] }, { "paper_id": "44bccae7b3e9bf735360463f7d5197e600efbe3d", "title": "A data mining methodology for predicting early stage Parkinson’s disease using non-invasive, high-dimensional gait sensor data", "doi": "10.1080/19488300.2015.1095256", "year": 2015, "finding": "This paper uses data mining for PD diagnosis, not for subtyping progression trajectories. It does not use SuStaIn and is tangential to the core methodological transfer.", "severity": "minor", "authors": [ "Conrad S. Tucker", "Yixiang Han", "Harriet Black Nembhard", "Wang-Chien Lee", "M. Lewis", "N. Sterling", "Xuemei Huang" ] }, { "paper_id": "eb68bfce4a376b45593cfaa8e1bd0171a1f26e04", "title": "Progression of erectile function in men with chronic obstructive pulmonary disease: a cohort study", "doi": "10.1186/s12890-019-0902-y", "year": 2019, "finding": "This paper studies a specific comorbidity trajectory in COPD, not the primary PFT progression subtypes. It is tangential to the hypothesis.", "severity": "minor", "authors": [ "E. Oh", "J. Yoo" ] }, { "paper_id": "16a9f51d281a746e4d9c10674b5d97920efa748f", "title": "Probing Cellular and Molecular Mechanisms of Cigarette Smoke-Induced Immune Response in the Progression of Chronic Obstructive Pulmonary Disease Using Multiscale Network Modeling", "doi": "10.1371/journal.pone.0163192", "year": 2016, "finding": "This paper models molecular mechanisms of COPD, not patient-level progression subtypes. It is tangential to the hypothesis.", "severity": "minor", "authors": [ "Zhi-chao Pan", "Hai-shan Yu", "J. Liao" ] }, { "paper_id": "d46c3ac0fe9bbfa7742116771a6233f2e0297aa9", "title": "Symptom Trajectories and Clinical Subtypes in Post–COVID-19 Condition: Systematic Review and Clustering Analysis", "doi": "10.2196/72221", "year": 2025, "finding": "This paper identifies subtypes in post-COVID condition, not in COPD or asthma. It does not use SuStaIn. It is tangential but shows the general interest in subtyping respiratory conditions.", "severity": "minor", "authors": [ "Mingzhi Hu", "Tian Song", "Zhaoyuan Gong", "Qianzi Che", "Jing Guo", "Lin Chen", "Haili Zhang", "Huizhen Li", "Ning Liang", "Guozhen Zhao", "Yanping Wang", "Nannan Shi", "Bin Liu" ] } ], "gap_manifest_update": { "closed_gaps": [ "The hypothesis that SuStaIn can be transferred from AD to other diseases is partially closed by papers showing its successful application to schizophrenia (paper_id: dbfe91194a4c7c8fa7656e76cd18d37036d0729f), epilepsy (paper_id: dd9e3060fa7fa5402f4df5f5a2aa214fcaf68a70), and PSP (paper_id: fbfaa5078f8efdd5fc7c1d01f6100a90b86c9b7c). This reduces the 'Technical Implementation' competence gap by demonstrating the algorithm's generalizability." ], "new_gaps": [ "No paper in the provided list applies SuStaIn to pulmonary function test (PFT) data. The critical data gap for raw longitudinal PFT data with linked CT remains open.", "No paper in the provided list uses SuStaIn on any type of non-imaging, longitudinal clinical data (e.g., spirometry time series). This highlights a potential methodological gap in adapting the algorithm to non-imaging data types.", "The specific prediction linking oral-gut microbiome taxa to a SuStaIn-derived COPD subtype (the 'epistemic gap') remains entirely unexplored in the provided literature." ], "data_available": [ "The provided papers do not contain or reference any publicly available dataset suitable for testing the hypothesis. The COPDGene cohort is mentioned in several papers (e.g., paper_id: f70855235912e70c29b0884b39ce63d1c90ecb8a, 6594d8ad2fde347f05083ff010ae828c70bd875c, fc599ec0092f36607f8719686cdc37118ab53313) but the raw data is not provided." ] }, "search_queries": [ { "query": "SuStaIn algorithm pulmonary disease subtypes", "type": "novelty", "rationale": "Direct search for the core hypothesis: application of the SuStaIn algorithm to lung disease subtyping." }, { "query": "data-driven COPD subtypes longitudinal PFT", "type": "novelty", "rationale": "Checks if data-driven subtyping of COPD using longitudinal pulmonary function tests has been proposed or validated." }, { "query": "disease progression modeling obstructive lung disease", "type": "novelty", "rationale": "Searches for progression modeling (a key SuStaIn feature) applied to obstructive lung diseases like COPD or asthma." }, { "query": "GAMLSS versus SLR pulmonary reference equations heterogeneity", "type": "evidence", "rationale": "Targets the methodological debate mentioned in the hypothesis, seeking evidence of acknowledged heterogeneity in PFT reference equations." }, { "query": "FEV1 decline subtypes COPD trajectory clustering", "type": "evidence", "rationale": "Looks for empirical evidence of distinct FEV1 decline trajectories in COPD, supporting the existence of subtypes like 'rapid decliner'." }, { "query": "emphysema airway predominant COPD subtypes CT", "type": "evidence", "rationale": "Searches for evidence of imaging-based COPD subtypes (emphysema vs. airway) that could be integrated with SuStaIn." }, { "query": "asthma neutrophilic remodeling subtypes longitudinal", "type": "evidence", "rationale": "Seeks evidence for distinct asthma subtypes (neutrophilic, remodeling-prone) with different progression patterns." }, { "query": "SuStaIn transfer methodology neurodegenerative lung", "type": "cross_domain", "rationale": "Directly searches for interdisciplinary precedents of transferring SuStaIn from neurodegeneration to lung disease." }, { "query": "subtype and stage inference non-neurological disease", "type": "cross_domain", "rationale": "Broadens the search to any application of SuStaIn outside of neurology, establishing a precedent for domain transfer." }, { "query": "data-driven subtyping methodology transfer COPD Alzheimer", "type": "cross_domain", "rationale": "Searches for general methodological transfer of data-driven subtyping between Alzheimer's disease and COPD." } ] }, "sharpened": { "title": "Transferability of the Subtype and Stage Inference (SuStaIn) Algorithm to Pulmonary Function Data for the Identification of Mechanistically Distinct Progression Subtypes in Obstructive and Restrictive Lung Diseases", "formal_statement": "If the SuStaIn algorithm, which identifies discrete progression subtypes from high-dimensional longitudinal data, is applied to longitudinal pulmonary function test (PFT) data (spirometry, lung volumes, DLCO) from cohorts of patients with COPD and asthma, then it will identify at least two reproducible, mechanistically distinct progression subtypes (e.g., a 'rapid FEV1 decliner' subtype and an 'emphysema-predominant' subtype) with significantly different rates of decline in FEV1 (≥ 50 mL/year difference in mean slope), because the algorithm can decompose heterogeneous disease trajectories into discrete, stage-ordered sequences of biomarker events, as validated in neurodegenerative diseases.", "independent_variables": [ { "name": "Application of SuStaIn algorithm to longitudinal PFT data", "type": "categorical", "range": "Applied vs. Not Applied (control: standard GOLD staging)", "unit": "N/A" }, { "name": "Disease cohort", "type": "categorical", "range": "COPD (GOLD 1-4) vs. Asthma (GINA steps 1-5) vs. Restrictive (IPF, GAP stage 1-3)", "unit": "N/A" } ], "dependent_variables": [ { "name": "Number of identified reproducible progression subtypes", "type": "continuous", "expected_direction": "non-monotonic", "unit": "integer (count)" }, { "name": "Mean annual rate of FEV1 decline per subtype", "type": "continuous", "expected_direction": "decrease", "unit": "mL/year" }, { "name": "Subtype reproducibility (adjusted Rand index between bootstrapped samples)", "type": "continuous", "expected_direction": "increase", "unit": "dimensionless (0 to 1)" }, { "name": "Hazard ratio for all-cause mortality per subtype", "type": "continuous", "expected_direction": "non-monotonic", "unit": "dimensionless (HR)" } ], "proposed_mechanism": { "causal_chain": [ "Step 1: Longitudinal PFT data (FEV1, FVC, FEV1/FVC, TLC, RV, DLCO) are collected at ≥3 time points over ≥2 years from a large cohort (N≥500) of patients with a given diagnosis (e.g., COPD).", "Step 2: The SuStaIn algorithm models each patient's sequence of biomarker events (e.g., 'FEV1 drops below 80% predicted' occurs before 'DLCO drops below 60% predicted') as a mixture of K discrete progression models (subtypes), each with a common ordering of events but variable timing.", "Step 3: The algorithm assigns each patient a posterior probability of belonging to each subtype and estimates a continuous disease stage (0-100) for each patient within their most likely subtype.", "Step 4: The resulting subtypes are validated for reproducibility via bootstrapping (resampling with replacement) and for external validity by testing for significant differences in clinical outcomes (mortality, exacerbation rate, response to therapy) and biomarkers (blood eosinophils, CT emphysema score) across subtypes." ], "key_assumptions": [ "Assumption 1: The progression of lung function decline in obstructive/restrictive diseases follows a finite set of discrete, ordered sequences of biomarker events, rather than a continuum of trajectories.", "Assumption 2: The available longitudinal PFT data are sufficiently dense (≥3 time points per patient) and span a sufficient range of disease severity to allow the algorithm to infer event orderings.", "Assumption 3: The chosen biomarkers (PFT parameters) are monotonic with disease progression (i.e., they do not reverse direction in a significant fraction of patients).", "Assumption 4: The number of subtypes K is unknown a priori but can be determined by model selection criteria (e.g., cross-validation likelihood)." ], "known_unknowns": [ "We do not know whether the optimal number of subtypes K will be the same across different diseases (COPD vs. asthma vs. IPF).", "We do not know whether the algorithm will identify subtypes that correspond to known clinical phenotypes (e.g., eosinophilic vs. neutrophilic asthma) or discover entirely novel ones.", "We do not know whether the subtypes will be stable across different PFT reference equations (GLI-2012 vs. NHANES III)." ] }, "falsifiable_predictions": [ { "prediction": "Applying SuStaIn to longitudinal PFT data from the COPDGene cohort (N≥2000, ≥3 visits) will identify at least two distinct progression subtypes with a mean difference in annual FEV1 decline of ≥50 mL/year.", "quantitative_bound": "Mean FEV1 decline in subtype A: -30 ± 10 mL/year; subtype B: -80 ± 15 mL/year; difference in means ≥ 50 mL/year with 95% CI not crossing zero.", "measurement_method": "Spirometry (FEV1) at baseline, year 1, year 2, year 5; SuStaIn algorithm (Python implementation, https://github.com/ucl-pond/pySuStaIn) with K=2 to K=5 tested.", "null_hypothesis": "H0: The mean annual FEV1 decline does not differ significantly between any pair of SuStaIn-derived subtypes (i.e., all subtypes have the same mean slope).", "statistical_test": "One-way ANOVA with post-hoc Tukey HSD test for pairwise comparisons; alpha = 0.05, corrected for multiple comparisons." }, { "prediction": "The SuStaIn-derived subtypes will show a hazard ratio for all-cause mortality that differs by at least 1.5 between the highest-risk and lowest-risk subtype.", "quantitative_bound": "HR for highest-risk subtype vs. lowest-risk subtype ≥ 2.0 (95% CI: 1.5 to 2.8).", "measurement_method": "All-cause mortality from medical records or National Death Index over 10-year follow-up; Cox proportional hazards model with subtype as categorical predictor, adjusted for age, sex, smoking pack-years, and baseline FEV1%.", "null_hypothesis": "H0: The hazard ratio for all-cause mortality is equal across all SuStaIn-derived subtypes (HR = 1 for all pairwise comparisons).", "statistical_test": "Log-rank test for equality of survival curves across subtypes; Cox regression Wald test for subtype coefficient; alpha = 0.05." }, { "prediction": "The SuStaIn-derived subtypes will be reproducible, with an adjusted Rand index (ARI) > 0.80 between the original solution and solutions from 100 bootstrap resamples.", "quantitative_bound": "Mean ARI across 100 bootstraps: 0.85 ± 0.05 (95% CI: 0.80 to 0.90).", "measurement_method": "Bootstrap resampling with replacement of patients; SuStaIn re-run on each bootstrap sample; ARI calculated between original subtype assignments and bootstrap-derived assignments.", "null_hypothesis": "H0: The adjusted Rand index is not significantly greater than 0 (i.e., the subtype assignments are no more similar than random chance).", "statistical_test": "One-sample t-test comparing mean ARI to 0; alpha = 0.05." } ], "boundary_conditions": [ { "condition": "The analysis is limited to patients with ≥3 valid PFT measurements over a minimum of 2 years of follow-up.", "justification": "SuStaIn requires longitudinal data to infer event orderings; fewer time points or shorter follow-up may not capture sufficient progression events." }, { "condition": "The algorithm is applied separately to each disease cohort (COPD, asthma, IPF) and not pooled across diseases.", "justification": "The pathophysiology and natural history differ fundamentally between obstructive and restrictive diseases; pooling would violate the assumption of a common set of progression events." }, { "condition": "The analysis is restricted to PFT parameters that are routinely collected and have established reference equations (FEV1, FVC, FEV1/FVC, TLC, RV, DLCO).", "justification": "Less common measures (e.g., oscillometry, MIP/MEP) may not be available in large cohorts and lack validated normative data for event definition." }, { "condition": "The algorithm is applied to data from cohorts with ≥500 patients to ensure sufficient statistical power for subtype discovery.", "justification": "Simulation studies of SuStaIn show that sample sizes <500 lead to unstable subtype solutions and low reproducibility (ARI < 0.6)." } ], "theoretical_framework": "Data-driven disease subtyping via mixture models of progression trajectories (SuStaIn framework), grounded in the theory that complex chronic diseases are composed of multiple, mechanistically distinct pathophysiological processes that manifest as discrete sequences of biomarker events (Young et al., 2018, NeuroImage)." }, "protocol": { "protocol_title": "Validation protocol for SuStaIn-based progression subtyping of obstructive and restrictive lung diseases using longitudinal pulmonary function data", "overall_timeline": "8-14 months", "overall_budget_estimate": "€20k-80k", "phases": [ { "phase_number": 1, "phase_name": "In Silico Validation on Public & Simulated Data", "objective": "To rapidly falsify the core assumption that SuStaIn can recover distinct progression subtypes from longitudinal PFT data, using publicly available datasets and synthetic data with known ground truth.", "methodology": "1. **Data acquisition**: Download the COPDGene Phase 1-2-3 dataset (dbGaP phs000179) and the SARP III asthma dataset (dbGaP phs001446). Extract longitudinal PFT data (FEV1, FVC, FEV1/FVC, TLC, RV, DLCO) for patients with ≥3 visits over ≥2 years. 2. **Synthetic data generation**: Simulate 500 patients with two ground-truth subtypes (rapid decliner: FEV1 slope -80 mL/yr; slow decliner: -30 mL/yr) with added Gaussian noise (σ=15 mL/yr) and missing data patterns (MCAR 10%). 3. **Algorithm application**: Run pySuStaIn (https://github.com/ucl-pond/pySuStaIn) with K=2..5, using default settings (MCMC chains=4, iterations=10000, burn-in=5000). 4. **Reproducibility assessment**: Compute ARI between true and recovered subtypes on synthetic data; compute ARI across 50 bootstrap resamples on real data. 5. **Sensitivity analysis**: Vary number of time points (3 vs. 5), follow-up duration (2 vs. 5 years), and sample size (N=200 vs. 500) in synthetic data.", "required_resources": { "equipment": [ "Standard workstation (16GB RAM, 4+ cores) or cloud VM (AWS EC2 c5.2xlarge)" ], "software": [ "Python 3.9+ with pySuStaIn (pip install pySuStaIn)", "Jupyter Notebook", "R 4.2+ with packages: survival, cluster, fossil", "Git for version control" ], "datasets": [ "COPDGene (dbGaP phs000179) – requires IRB-approved data access request (2-4 weeks)", "SARP III (dbGaP phs001446) – same access process", "Synthetic data generated via custom Python script (numpy, scipy.stats)" ], "competences": [ "Python programming (numpy, pandas, scikit-learn)", "Bayesian modeling / MCMC concepts", "Basic survival analysis in R", "Data wrangling of large clinical databases" ], "estimated_cost": "€500-2000", "estimated_duration": "4-6 weeks" }, "expected_outputs": [ "Synthetic data recovery: ARI between true and recovered subtypes >0.80 for K=2, N=500, 3 time points", "Real data bootstrap ARI >0.70 for COPDGene (K=2) and SARP III (K=2)", "Sensitivity curves: ARI vs. sample size, number of time points, follow-up duration", "List of PFT biomarkers that are most informative for subtype separation (from pySuStaIn event orderings)" ], "success_criteria": [ { "metric": "ARI on synthetic data (K=2, N=500, 3 time points)", "threshold": "≥ 0.80", "measurement": "Adjusted Rand Index between true subtype labels and pySuStaIn MAP assignments" }, { "metric": "Bootstrap ARI on COPDGene (K=2)", "threshold": "≥ 0.70 (mean across 50 bootstraps)", "measurement": "ARI between original subtype assignments and assignments from each bootstrap resample" }, { "metric": "Minimum sample size for ARI >0.70", "threshold": "N ≤ 500", "measurement": "From synthetic sensitivity analysis: find smallest N where mean ARI >0.70" } ], "go_nogo_decision": { "go_if": "All three success criteria met (synthetic ARI ≥0.80, bootstrap ARI ≥0.70, required N ≤500)", "nogo_if": "Synthetic ARI <0.60 OR bootstrap ARI <0.50 OR required N >1000", "pivot_if": "ARI acceptable but only with >4 time points or >5 years follow-up: redesign Phase 2 to use richer datasets (e.g., SPIROMICS with 5 visits)" }, "risks": [ { "risk": "COPDGene/SARP data access delayed >4 weeks", "probability": "medium", "mitigation": "Use synthetic data only for initial validation; also download publicly available ECLIPSE (NCT00292552) summary statistics if full data unavailable" }, { "risk": "pySuStaIn fails to converge on real PFT data (e.g., due to non-monotonic biomarkers)", "probability": "medium", "mitigation": "Pre-screen biomarkers for monotonicity using linear mixed models; exclude biomarkers with >20% non-monotonic trajectories" }, { "risk": "Synthetic data too simplistic (e.g., Gaussian noise underestimates real variability)", "probability": "low", "mitigation": "Add heteroscedastic noise proportional to FEV1 level, and simulate missing data patterns from real COPDGene dropout rates" } ] }, { "phase_number": 2, "phase_name": "Minimal Experimental Validation on a Single Large Cohort", "objective": "To confirm the central mechanism (SuStaIn identifies ≥2 subtypes with ≥50 mL/yr FEV1 decline difference) in a single, well-characterized cohort with sufficient power.", "methodology": "1. **Cohort selection**: Use COPDGene Phase 1-2-3 (N≈10,000, ≥3 visits over 5 years). Apply inclusion criteria: ≥3 valid PFTs, ≥2 years follow-up, GOLD 1-4. Expected N≈4,000. 2. **Data preprocessing**: Normalize PFTs to percent-predicted using GLI-2012 equations. Define events as z-score thresholds (e.g., FEV1 z-score < -1.64, -2.33, -3.0). 3. **SuStaIn application**: Run pySuStaIn with K=2..5, 4 MCMC chains, 20,000 iterations, 10,000 burn-in. Use cross-validation likelihood to select optimal K. 4. **Primary analysis**: For the best K, compute mean annual FEV1 decline per subtype via linear mixed models (random intercept + slope). Test pairwise differences with ANOVA + Tukey HSD. 5. **Secondary analysis**: Cox proportional hazards for all-cause mortality (10-year follow-up), adjusted for age, sex, smoking, baseline FEV1%. 6. **Reproducibility**: 100 bootstrap resamples, compute mean ARI.", "required_resources": { "equipment": [ "High-performance computing cluster (e.g., SLURM, 64GB RAM, 16 cores) or cloud GPU instance (AWS g4dn.xlarge)" ], "software": [ "Python 3.9+ with pySuStaIn, numpy, pandas, scikit-learn, statsmodels", "R 4.2+ with survival, lme4, cluster, fossil", "Docker container with all dependencies (optional but recommended)" ], "datasets": [ "COPDGene Phase 1-2-3 (dbGaP phs000179) – already accessed in Phase 1" ], "competences": [ "Advanced Python/R programming", "Bayesian mixture modeling", "Survival analysis (Cox models, competing risks)", "Clinical database management (SQL or Parquet)" ], "estimated_cost": "€2k-15k", "estimated_duration": "2-3 months" }, "expected_outputs": [ "Optimal number of subtypes K (e.g., K=2 or K=3) with cross-validation likelihood", "Mean annual FEV1 decline per subtype with 95% CI", "Pairwise difference in FEV1 decline between fastest and slowest decliner subtypes (≥50 mL/yr expected)", "Hazard ratios for all-cause mortality per subtype (≥2.0 expected for highest-risk vs lowest-risk)", "Bootstrap ARI >0.80" ], "success_criteria": [ { "metric": "Mean difference in annual FEV1 decline between fastest and slowest decliner subtypes", "threshold": "≥ 50 mL/year, with 95% CI not crossing zero", "measurement": "Linear mixed model (lme4 in R) with subtype as fixed effect, random intercept + slope per patient; Tukey HSD post-hoc test" }, { "metric": "Hazard ratio for all-cause mortality (highest-risk vs lowest-risk subtype)", "threshold": "≥ 2.0, with 95% CI lower bound >1.0", "measurement": "Cox proportional hazards model (survival::coxph in R) with subtype as categorical predictor, adjusted for age, sex, smoking pack-years, baseline FEV1%" }, { "metric": "Bootstrap reproducibility (ARI)", "threshold": "Mean ARI > 0.80 across 100 bootstraps", "measurement": "ARI between original MAP subtype assignments and assignments from each bootstrap resample" } ], "go_nogo_decision": { "go_if": "All three success criteria met (FEV1 decline difference ≥50 mL/yr, HR ≥2.0, ARI >0.80)", "nogo_if": "FEV1 decline difference <30 mL/yr OR HR <1.5 OR ARI <0.60", "pivot_if": "FEV1 decline difference ≥50 mL/yr but HR <1.5: explore alternative clinical outcomes (exacerbation rate, 6-minute walk distance) as primary validation" }, "risks": [ { "risk": "COPDGene data access not renewed or revoked", "probability": "low", "mitigation": "Use SPIROMICS (NCT01969344) as backup cohort; similar size and PFT availability" }, { "risk": "SuStaIn identifies only one subtype (K=1 optimal)", "probability": "medium", "mitigation": "If K=1, test alternative event definitions (e.g., absolute thresholds instead of z-scores) or include additional biomarkers (CT emphysema, blood eosinophils) as input features" }, { "risk": "High missing data rate (>30%) at year 5 leads to biased subtype estimates", "probability": "medium", "mitigation": "Use multiple imputation (MICE in R) for missing PFTs; sensitivity analysis comparing complete-case vs. imputed results" } ] }, { "phase_number": 3, "phase_name": "Full Protocol: Multi-Cohort Validation with External Replication and Mechanistic Characterization", "objective": "To rigorously validate SuStaIn-derived subtypes across multiple independent cohorts, demonstrate generalizability to asthma and IPF, and link subtypes to underlying molecular mechanisms via multi-omics data.", "methodology": "1. **Cohort expansion**: Apply SuStaIn separately to: (a) COPDGene (replication), (b) SPIROMICS (N≈2,000, 5 visits), (c) SARP III (N≈700, 4 visits), (d) IPF-PRO Registry (N≈500, 3 visits). 2. **Cross-cohort subtype matching**: Use a meta-clustering approach: train SuStaIn on each cohort independently, then match subtypes across cohorts using event ordering similarity (Spearman correlation of event positions). 3. **External validation**: Test whether matched subtypes show consistent FEV1 decline differences, mortality HR, and exacerbation rates across cohorts. 4. **Mechanistic characterization**: For COPD subtypes, link to: (a) CT emphysema score (visual + quantitative, from COPDGene), (b) blood eosinophil counts, (c) sputum proteomics (if available in SPIROMICS), (d) airway gene expression (from bronchial brushings in SARP). 5. **Prediction falsification**: Test Prediction 3 (ARI >0.80) on all cohorts; test Prediction 1 and 2 on each cohort separately. 6. **Sensitivity analysis**: Repeat with NHANES III reference equations to test robustness to reference standard.", "required_resources": { "equipment": [ "High-performance computing cluster (64+ cores, 256GB RAM) or cloud HPC (AWS ParallelCluster)", "Secure data storage (encrypted, HIPAA-compliant if needed)" ], "software": [ "Python 3.9+ with pySuStaIn, scikit-learn, pandas, numpy", "R 4.2+ with survival, lme4, cluster, metafor (meta-analysis)", "Docker/Singularity containers for reproducibility", "GitHub repository for code sharing" ], "datasets": [ "COPDGene (dbGaP phs000179)", "SPIROMICS (dbGaP phs001838)", "SARP III (dbGaP phs001446)", "IPF-PRO Registry (contact principal investigator for data access)", "Optional: ECLIPSE (NCT00292552) for additional replication" ], "competences": [ "All Phase 1-2 competences", "Multi-omics data integration (proteomics, transcriptomics)", "Meta-analysis methods (random-effects models, heterogeneity assessment)", "Clinical trial design and regulatory knowledge (for potential future interventional studies)" ], "estimated_cost": "€15k-100k+", "estimated_duration": "6-12 months" }, "expected_outputs": [ "SuStaIn-derived subtypes for COPD (2-3), asthma (2-3), IPF (2) with event ordering diagrams", "Cross-cohort meta-analysis of FEV1 decline differences (≥50 mL/yr pooled effect)", "Meta-analysis of mortality HR (≥2.0 pooled effect)", "Multi-omics characterization: differentially expressed genes/proteins per subtype", "CT emphysema distribution per subtype (e.g., emphysema-predominant vs. airway-predominant)", "Sensitivity analysis: effect of reference equations on subtype stability" ], "success_criteria": [ { "metric": "Cross-cohort replication of FEV1 decline difference", "threshold": "≥ 50 mL/yr in at least 3 out of 4 cohorts, with meta-analytic 95% CI not crossing zero", "measurement": "Random-effects meta-analysis (metafor::rma in R) of mean FEV1 decline difference across cohorts" }, { "metric": "Cross-cohort replication of mortality HR", "threshold": "≥ 2.0 in at least 2 out of 3 cohorts with mortality data (COPDGene, SPIROMICS, IPF-PRO), meta-analytic 95% CI lower bound >1.0", "measurement": "Random-effects meta-analysis of log(HR) across cohorts" }, { "metric": "Multi-omics differentiation", "threshold": "At least 10 differentially expressed genes (FDR <0.05) between subtypes in at least one omics dataset", "measurement": "DESeq2 (RNA-seq) or limma (proteomics) with subtype as predictor, adjusted for age, sex, smoking" }, { "metric": "Reference equation robustness", "threshold": "ARI >0.80 between subtype assignments using GLI-2012 vs. NHANES III", "measurement": "ARI between MAP assignments from two reference equations" } ], "go_nogo_decision": { "go_if": "All four success criteria met (FEV1 decline replication, mortality replication, omics differentiation, reference robustness)", "nogo_if": "FEV1 decline difference <30 mL/yr in ≥2 cohorts OR mortality HR <1.5 in ≥2 cohorts OR no omics signals", "pivot_if": "Subtypes replicate across cohorts but lack omics differentiation: focus on clinical utility (e.g., differential response to bronchodilators or inhaled corticosteroids) as alternative validation" }, "risks": [ { "risk": "IPF-PRO Registry data access denied or insufficient sample size", "probability": "medium", "mitigation": "Use publicly available IPF data from the PROOF registry (NCT01915511) or the GSE110147 gene expression dataset as alternative" }, { "risk": "Multi-omics data not available or poorly matched across cohorts", "probability": "medium", "mitigation": "Focus on CT imaging and blood biomarkers (eosinophils, neutrophils) which are available in all cohorts; omics treated as exploratory" }, { "risk": "Subtypes differ substantially across cohorts (low cross-cohort ARI)", "probability": "high", "mitigation": "Use meta-clustering approach that allows cohort-specific event orderings but enforces shared subtype-level structure; if still fails, report as negative result and discuss reasons (cohort heterogeneity, different disease stages)" } ] } ], "phase_1_quick_start": { "can_start_today": true, "first_action": "Submit data access requests to dbGaP for COPDGene (phs000179) and SARP III (phs001446) via your institution's IRB; simultaneously write Python script to generate synthetic PFT data with known ground-truth subtypes", "tools_needed": [ "Python 3.9+ with numpy, scipy, pandas, scikit-learn", "pySuStaIn (pip install pySuStaIn)", "Jupyter Notebook", "GitHub account for version control" ], "open_data_sources": [ "COPDGene summary statistics (not full data) available at https://www.copdgene.org/", "Synthetic data generation code template: https://github.com/ucl-pond/pySuStaIn/blob/master/examples/example_synthetic_data.ipynb", "ECLIPSE study data (NCT00292552) available on request from GSK" ] } }, "panel": { "reviews": [ { "reviewer_persona": "methodologist", "overall_score": 7.2, "verdict": "accept", "strengths": [ "Excellente structuration en phases avec critères GO/NO-GO/PIVOT clairs, permettant une prise de décision objective et une adaptation pragmatique du protocole en fonction des résultats intermédiaires.", "Pré-spécification rigoureuse des prédictions falsifiables avec des bornes quantitatives précises (différence de ≥50 mL/an, HR ≥2.0, ARI >0.80) et des hypothèses nulles clairement énoncées.", "Plan de validation multi-cohorte incluant une réplication externe et une caractérisation mécanistique via multi-omiques, ce qui renforce considérablement la validité externe et la plausibilité biologique." ], "weaknesses": [ "Absence de contrôle pour l'effet du nombre de visites et de la durée de suivi sur la stabilité des sous-types : le protocole ne compare pas SuStaIn à une méthode de clustering longitudinale standard (ex. k-means sur pentes individuelles, modèles de mélange de trajectoires) pour démontrer une valeur ajoutée spécifique.", "Risque de biais de confirmation non adressé : les seuils de z-score pour les événements (ex. -1.64, -2.33, -3.0) sont arbitraires et pourraient être optimisés pour produire les sous-types attendus. Aucune analyse de sensibilité sur ces seuils n'est prévue.", "La puissance statistique pour la comparaison des HR de mortalité entre sous-types n'est pas calculée. Avec un sous-type à faible risque (ex. 30% de l'échantillon) et un HR de 2.0, le nombre d'événements requis peut ne pas être atteint dans les cohortes les plus petites (ex. SARP III, IPF-PRO)." ], "critical_questions": [ "Comment comptez-vous démontrer que SuStaIn apporte une information supplémentaire par rapport à une simple classification basée sur la pente de FEV1 (ex. déclin rapide >60 mL/an vs. lent) ? Un modèle de mélange de trajectoires (growth mixture model) serait un contrôle plus pertinent que le staging GOLD.", "Quel est le plan d'analyse si le nombre optimal de sous-types (K) diffère entre les cohortes (ex. K=2 dans COPDGene, K=3 dans SPIROMICS) ? Comment les sous-types seront-ils appariés dans ce cas ?", "Avez-vous prévu une correction pour les comparaisons multiples dans l'analyse des 10 000 gènes différentiellement exprimés (Phase 3) ? Le seuil FDR <0.05 est-il maintenu sur l'ensemble des tests ou par sous-type ?" ], "recommendation": "Ce protocole est globalement bien conçu et mérite d'être financé, avec des réserves modérées. Je recommande d'ajouter une analyse de sensibilité sur les seuils de z-score et un contrôle méthodologique (ex. comparaison avec un clustering naïf sur les pentes) dans la Phase 1. Un calcul de puissance formel pour les analyses de survie est nécessaire avant la Phase 2, avec un plan de contingence si la puissance est insuffisante (ex. regroupement de sous-types à faible risque).", "confidence": 0.82 }, { "reviewer_persona": "domain_expert", "overall_score": 7.8, "verdict": "accept", "strengths": [ "L'hypothèse est théoriquement cohérente et s'appuie sur un cadre épistémologique solide (SuStaIn) déjà validé dans plusieurs pathologies neurodégénératives et psychiatriques, ce qui justifie son transfert aux maladies respiratoires chroniques. La chaîne causale proposée est logique et bien articulée, avec des étapes claires de collecte, modélisation, affectation et validation.", "La base bibliographique est de haute qualité, incluant des preuves directes de la transférabilité de SuStaIn (schizophrénie, épilepsie, PSP) et des preuves indirectes de l'hétérogénéité sous-jacente de la BPCO (sous-types CT, clustering multi-omiques). La distinction entre preuves 'analogues' et 'indirectes' est rigoureuse et montre une compréhension nuancée de l'état de l'art.", "L'identification de 'connues inconnues' (K variable selon la maladie, stabilité des sous-types entre équations de référence) est un signe de maturité scientifique et démontre une anticipation réaliste des défis méthodologiques, ce qui renforce la plausibilité de l'approche." ], "weaknesses": [ "L'hypothèse repose sur l'hypothèse forte (Assumption 1) que la progression de la fonction pulmonaire suit un nombre fini de séquences ordonnées discrètes. Or, la littérature sur la BPCO suggère un continuum de trajectoires avec des chevauchements phénotypiques importants (ex: chevauchement emphysème-maladie des petites voies aériennes). La violation de cette hypothèse pourrait conduire à des sous-types artefactuels ou instables.", "La dépendance à des données longitudinales denses (≥3 points sur ≥2 ans) est une limitation majeure. De nombreuses cohortes cliniques de BPCO (ex: SPIROMICS, COPDGene) ont des données longitudinales mais avec une fréquence variable et un attrition non négligeable. La faisabilité pratique est sous-estimée, surtout pour des maladies comme l'asthme où les variations saisonnières et thérapeutiques compliquent la monotonie des biomarqueurs (Assumption 3).", "Le mécanisme proposé ignore le rôle potentiel des traitements (corticostéroïdes inhalés, bronchodilatateurs) qui peuvent modifier les trajectoires de déclin du VEMS. SuStaIn, dans sa formulation actuelle, ne modélise pas les effets thérapeutiques, ce qui pourrait biaiser l'identification des sous-types 'naturels' de progression." ], "critical_questions": [ "Comment l'algorithme gérera-t-il les patients dont les biomarqueurs PFT montrent des améliorations transitoires (ex: réponse aux bronchodilatateurs) ou des plateaux prolongés, violant l'hypothèse de monotonie stricte ? Une version de SuStaIn permettant des événements non-monotones est-elle envisagée ?", "Quelle est la stratégie pour déterminer le nombre optimal de sous-types K dans un contexte où les critères de sélection de modèle (ex: vraisemblance en validation croisée) peuvent favoriser des solutions sur-paramétrées, surtout avec des échantillons de taille modérée (N=500) ?", "Les sous-types identifiés seront-ils validés contre des données de réponse thérapeutique (ex: réponse aux corticoïdes inhalés dans l'asthme) ou des biomarqueurs moléculaires (ex: protéomique du lavage broncho-alvéolaire) pour établir leur distinctivité mécanistique, ou uniquement contre des critères cliniques composites ?" ], "recommendation": "Cette hypothèse est prometteuse et mérite d'être testée. Je recommande de commencer par une étude de preuve de concept sur une cohorte bien caractérisée comme COPDGene ou SPIROMICS, en incluant des données CT pour valider les sous-types de progression. Il serait crucial d'évaluer la stabilité des sous-types via des analyses de sensibilité (bootstrap, permutations) et de comparer les résultats avec des méthodes de clustering longitudinal alternatives (ex: mixed-effects models with latent classes). L'ajout de covariables thérapeutiques dans le modèle serait un atout majeur pour la plausibilité clinique.", "confidence": 0.82 }, { "reviewer_persona": "contrarian", "overall_score": 3.5, "verdict": "weak_reject", "strengths": [ "L'application de SuStaIn à des données longitudinales est une approche novatrice qui a fait ses preuves dans les maladies neurodégénératives, et le choix de cohortes bien caractérisées comme COPDGene est judicieux.", "La formulation de prédictions quantitatives précises (différence de ≥50 mL/an dans le déclin du VEMS, HR ≥2,0) est louable et facilite la falsification de l'hypothèse." ], "weaknesses": [ "FAIL REASON #1: Non-monotonie et variabilité intra-individuelle des PFT. Contrairement aux biomarqueurs neurodégénératifs (atrophie cérébrale, protéines tau), les paramètres PFT (VEMS, CVF, DLCO) fluctuent considérablement dans le temps en raison des exacerbations, des traitements, de l'observance et des erreurs de mesure. L'hypothèse d'une progression monotone (Assumption 3) est violée chez une fraction importante de patients, ce qui brisera la capacité de SuStaIn à inférer des séquences d'événements ordonnées et stables. Le bruit temporel dominera le signal de progression.", "FAIL REASON #2: Le nombre de points temporels (≥3) est insuffisant pour estimer des trajectoires de déclin fiables et des séquences d'événements. Avec seulement 3 à 4 visites sur 5 ans, SuStaIn ne peut pas distinguer un 'déclin rapide' d'un 'déclin lent' avec une incertitude acceptable. La méthode a été conçue pour des données denses (≥5-10 points) ou des événements binaires irréversibles. Ici, le modèle sera sous-déterminé et produira des sous-types instables, même avec N=2000. L'ARI de 0,85 promis est irréaliste.", "FAIL REASON #3: Confusion entre hétérogénéité inter-individuelle et sous-types discrets. L'hypothèse (Assumption 1) que le déclin de la fonction pulmonaire suit un nombre fini de séquences ordonnées discrètes est une simplification excessive. Les données PFT montrent plutôt un continuum de phénotypes (ex: emphysème vs. bronchite chronique) avec des chevauchements considérables. SuStaIn forcera des clusters artificiels dans ce continuum, produisant des sous-types qui ne sont pas 'mécanistiquement distincts' mais des artefacts de la méthode de clustering. Les différences de déclin du VEMS seront statistiquement significatives mais cliniquement triviales et non reproductibles dans une cohorte externe." ], "critical_questions": [ "Comment comptez-vous gérer le problème majeur de la non-monotonie des PFT, sachant que 30 à 50% des patients COPD présentent une amélioration du VEMS ≥100 mL entre deux visites consécutives (exacerbations, réponse aux bronchodilatateurs) ? Avez-vous prévu un prétraitement des données (lissage, exclusion des exacerbations) qui ne biaise pas les séquences ?", "Avec seulement 3-4 points temporels par patient, comment démontrez-vous que SuStaIn ne sur-apprend pas simplement le bruit de mesure ? Avez-vous réalisé une analyse de puissance pour déterminer le nombre minimum de visites nécessaire pour détecter K=2 sous-types avec une ARI >0,8 ?" ], "recommendation": "Avant de soumettre, vous devez démontrer la faisabilité sur des données simulées réalistes incluant du bruit, des fluctuations et des valeurs manquantes. Il faut aussi justifier que le nombre de points temporels est suffisant (≥5 visites par patient) ou utiliser une version de SuStaIn qui tolère les données clairsemées. Enfin, inclure une validation externe sur une cohorte indépendante (ex: SPIROMICS) est indispensable pour écarter les artefacts de clustering.", "confidence": 0.85 }, { "reviewer_persona": "industrialist", "overall_score": 7.5, "verdict": "accept", "strengths": [ "Marché adressable clair et significatif : La médecine respiratoire (BPCO, asthme, fibrose pulmonaire) représente un marché mondial de >$30 Md. L'identification de sous-types de progression permettrait un repositionnement de médicaments existants (ex: roflumilast pour les 'decliners rapides', biotiques anti-IL-5 pour sous-types éosinophiliques) et un enrichissement des essais cliniques, réduisant les coûts de phase II/III de 30-50%.", "Avantage concurrentiel défendable : Contrairement aux approches 'clustering' statiques (phénotypage clinique), SuStaIn capture la dynamique temporelle des biomarqueurs. Cela donne un avantage pour les partenariats avec les CROs (ex: IQVIA, Parexel) cherchant à améliorer la stratification des patients dans les essais sur la BPCO, où le taux de déclin du VEMS est un critère d'évaluation clé mais bruité.", "Timeline de commercialisation réaliste : Le protocole proposé (8-14 mois, €20k-80k) est très lean. Un produit minimum viable (algorithme SaaS pour analyse de données d'essais) pourrait être opérationnel en 18-24 mois, ciblant d'abord les départements R&D des Big Pharma (GSK, AstraZeneca, Boehringer Ingelheim) comme outil de 'patient enrichment'." ], "weaknesses": [ "Barrière d'accès aux données : La validation multi-cohorte (Phase 3) nécessite l'accès à des données longitudinales de haute qualité (COPDGene, SPIROMICS, etc.). L'obtention de licences de données auprès de consortiums académiques peut prendre 6-12 mois et coûter >€50k, sans garantie d'exclusivité. C'est un goulot d'étranglement critique.", "Risque de 'commoditisation' de l'IP : L'algorithme SuStaIn lui-même est open source (pySuStaIn). L'IP réside dans les 'biomarker signatures' spécifiques aux maladies pulmonaires. Si un concurrent (ex: Tempus, Owkin) applique le même algorithme à ses propres données, l'avantage concurrentiel s'érode rapidement. Il faut breveter les 'trajectory signatures' comme méthode de diagnostic.", "Faible valeur perçue sans validation omique : Les Big Pharma paient pour des sous-types qui sont 'actionnables' (liés à une cible médicamenteuse). Sans la validation multi-omique (Phase 3), l'outil reste un 'joli clustering' sans mécanisme. Le risque est de ne pas pouvoir lever le budget Phase 3 (€100k+) sans un partenaire pharma engagé." ], "critical_questions": [ "Quel est le modèle de revenu ? Vente de licences SaaS aux CROs (€50k-100k/essai), ou service de consulting pour l'analyse de données propriétaires (€200k-500k par projet) ? Le premier est scalable, le second plus rentable mais moins défendable.", "Comment protégez-vous l'IP des 'trajectory signatures' ? Un brevet de méthode (USPTO) sur 'A method for identifying a rapid decliner subtype in COPD using SuStaIn on longitudinal FEV1 and DLCO data' est-il envisagé ? Sans cela, un concurrent peut copier le workflow en 3 mois.", "Quel est le plan d'adoption par les cliniciens ? Un radiologue ou pneumologue n'utilisera pas un algorithme qui nécessite 5 visites de PFT. Le produit doit être intégré dans un workflow EHR (ex: Epic, Cerner) pour être utilisé en routine. Avez-vous un partenaire EHR ?" ], "recommendation": "Lancer immédiatement la Phase 1 (in silico) en interne (€2k, 4 semaines) pour valider la faisabilité sur données publiques (COPDGene). Parallèlement, approcher 2-3 sociétés de biotech en stade précoce (ex: 4D Molecular Therapeutics, Pulmatrix) pour un 'collaborative research agreement' co-financé (€50k) qui couvrirait la Phase 2 et donnerait un accès privilégié à leurs données de phase I. L'objectif est d'avoir un 'case study' publiable et un brevet provisoire déposé dans les 12 mois, avant de chercher un licensing deal avec un CRO ou une Pharma de taille moyenne.", "confidence": 0.75 }, { "reviewer_persona": "funding_strategist", "overall_score": 7.2, "verdict": "accept", "strengths": [ "Hypothèse mécanistiquement forte et falsifiable: l'application d'un algorithme validé en neurosciences (SuStaIn) à des données longitudinales de fonction pulmonaire est une approche originale et à haut risque/récompense, avec un protocole GO/NO-GO clair qui rassure les évaluateurs sur la rigueur scientifique.", "Potentiel de rupture clinique: identifier des sous-types de progression distincts (ex: déclin rapide du VEMS vs emphysème prédominant) dans des maladies hétérogènes comme la BPCO et l'asthme pourrait transformer la stratification des patients et les essais cliniques, répondant à un besoin non satisfait majeur en pneumologie.", "TRL et trajectoire bien définis: le protocole en 3 phases (in silico, cohorte unique, multi-cohorte) montre une maturation progressive de TRL 2-3 (concept validé sur données synthétiques) à TRL 5-6 (validation multi-cohorte avec données omiques), ce qui est idéal pour des appels translationnels." ], "weaknesses": [ "Risque de non-généralisabilité: SuStaIn a été conçu pour des données à haute dimensionnalité (imagerie cérébrale). Les données PFT (spirométrie, volumes, DLCO) sont de dimensionnalité plus faible et moins riches en événements séquentiels, ce qui pourrait limiter la capacité de l'algorithme à identifier des sous-types discrets et reproductibles.", "Dépendance à la qualité des données longitudinales: l'hypothèse repose sur des cohortes avec des mesures répétées sur plusieurs années. L'accès à des données PFT longitudinales standardisées et bien phénotypées (avec peu de données manquantes) est un goulot d'étranglement majeur, surtout pour les maladies restrictives (ex: PI) où les cohortes sont plus rares.", "Manque de validation mécanistique immédiate: le protocole propose une validation multi-omique en Phase 3, mais sans garantie que les sous-types identifiés par SuStaIn correspondent à des voies biologiques distinctes. Un évaluateur pourrait demander une preuve de concept mécanistique (ex: profilage transcriptomique) dès la Phase 2." ], "critical_questions": [ "Comment allez-vous garantir l'accès à des données PFT longitudinales de haute qualité pour la Phase 2 (ex: cohorte ECLIPSE, SPIROMICS, ou UK Biobank) et quelles sont les négociations déjà en cours avec les gestionnaires de données?", "Quel est le plan de contingence si SuStaIn ne parvient pas à identifier des sous-types avec une différence de ≥50 mL/an dans la Phase 1 in silico? Envisagez-vous des modifications de l'algorithme (ex: ajout de covariables, réduction de dimensionnalité) ou un changement d'algorithme (ex: latent class analysis)?", "Comment comptez-vous intégrer les données omiques (protéomique, métabolomique) dans la Phase 3 pour démontrer la distinction mécanistique, et quelles sont les collaborations déjà établies avec des plateformes de multi-omiques (ex: EMBL-EBI, BBMRI)?" ], "recommendation": "Ciblez en priorité l'ERC Starting Grant (2026) pour financer le protocole complet (Phase 1-3) sur 5 ans, en mettant en avant l'originalité et le risque élevé. Parallèlement, soumettez une demande exploratoire à l'ANR (AAPG 2025) pour la Phase 1 in silico (budget €20k-€50k) afin de générer des preuves de concept solides avant le dépôt ERC. En cas d'échec ERC, rebondissez sur un appel ERC Proof of Concept (2027) si les résultats de Phase 1-2 sont positifs.", "confidence": 0.75, "funding_programs": [ { "program": "ERC Starting Grant 2026", "agency": "European Research Council", "fit_score": 0.85, "typical_budget": "€1.5M - €2.5M (sur 5 ans)", "success_rate": "14-17% (2023-2024)", "next_deadline": "Octobre 2025 (panel LS4 - Physiology, Pathophysiology and Endocrinology)", "rationale": "L'hypothèse est parfaitement alignée avec les critères ERC: haut risque scientifique (application d'un algorithme de neurosciences à la pneumologie), potentiel de rupture (identification de sous-types mécanistiques), et PI jeune (idéal pour Starting Grant). Le protocole GO/NO-GO démontre une rigueur méthodologique qui rassure les évaluateurs. Le panel LS4 est le plus adapté pour la physiologie respiratoire et les maladies chroniques." }, { "program": "ANR AAPG 2025 (Appel à Projets Générique) - Projet de Recherche Collaborative", "agency": "Agence Nationale de la Recherche (France)", "fit_score": 0.75, "typical_budget": "€200k - €600k (sur 36-48 mois)", "success_rate": "15-20% (2024)", "next_deadline": "Mars 2025 (axe 'Santé et biotechnologies')", "rationale": "L'ANR finance des projets exploratoires à fort potentiel translationnel. La Phase 1 in silico (€20k-€50k) peut être présentée comme un projet court et focalisé, avec des objectifs clairs et falsifiables. L'ANR valorise les collaborations interdisciplinaires (mathématiques appliquées + pneumologie). Le budget modeste et le timeline court (8-14 mois) sont un atout pour un AAPG." }, { "program": "NIH R01 (Research Project Grant) - NHLBI", "agency": "National Institutes of Health (USA) - National Heart, Lung, and Blood Institute", "fit_score": 0.7, "typical_budget": "$500k - $1.5M (sur 4-5 ans)", "success_rate": "18-22% (NHLBI, 2023)", "next_deadline": "Février 2025 (cycle standard) ou Octobre 2025", "rationale": "Le NHLBI finance des projets sur la BPCO, l'asthme et les maladies pulmonaires restrictives. L'hypothèse répond à un intérêt majeur: la stratification des patients pour des essais cliniques personnalisés. Le NIH valorise les approches computationnelles et les validations multi-cohortes (Phase 3). Un R01 serait idéal pour financer la Phase 2 et 3, avec un accent sur l'accès aux cohortes américaines (SPIROMICS, COPDGene)." } ] } ], "meta_review": { "consensus_score": 6.59, "verdict": "publish_brief", "key_consensus": [ "L'hypothèse est théoriquement cohérente, originale et s'appuie sur un cadre épistémologique solide (SuStaIn) déjà validé dans d'autres domaines, justifiant son transfert aux maladies respiratoires.", "Le protocole est globalement bien structuré avec des prédictions quantitatives précises (≥50 mL/an, HR ≥2.0) et un plan de validation multi-cohorte qui renforce la validité externe.", "Le potentiel translationnel et clinique est élevé : stratification des patients, enrichissement des essais cliniques, et réponse à un besoin non satisfait majeur en pneumologie.", "La faisabilité pratique est conditionnée par l'accès à des données longitudinales de haute qualité et par la gestion de la non-monotonie des PFT (exacerbations, traitements)." ], "key_disagreements": [ "Le Contrarian estime que la non-monotonie et la variabilité intra-individuelle des PFT, ainsi que le faible nombre de points temporels (≥3), rendent l'hypothèse irréaliste et produiront des sous-types artefactuels ; les autres reviewers jugent ces défis surmontables avec des analyses de sensibilité et un prétraitement adéquat.", "Le Contrarian considère que l'hypothèse d'un nombre fini de séquences ordonnées discrètes est une simplification excessive face au continuum de phénotypes observé en BPCO ; les autres reviewers reconnaissent ce risque mais le jugent testable via des comparaisons avec des méthodes de clustering longitudinal alternatives.", "Le Methodologist et l'Industrialist soulignent un manque de contrôle pour l'effet du nombre de visites et de la durée de suivi, ainsi qu'un risque de biais de confirmation sur les seuils de z-score ; le Domain Expert et le Funding Strategist insistent sur la dépendance à la qualité des données et le risque de non-généralisabilité." ], "critical_path": "La démonstration de la faisabilité sur des données simulées réalistes (incluant bruit, fluctuations, valeurs manquantes) et la validation de la stabilité des sous-types via des analyses de sensibilité (bootstrap, permutations) et une comparaison avec des méthodes de clustering longitudinal alternatives (growth mixture models) constituent le facteur le plus déterminant pour le succès ou l'échec de cette hypothèse.", "final_recommendation": "Le panel recommande la publication de ce brief, car l'hypothèse est prometteuse, mécanistiquement forte et répond à un besoin clinique majeur. Les réserves méthodologiques (non-monotonie, faible nombre de points temporels, risque d'artefacts de clustering) sont réelles mais peuvent être adressées par des analyses de sensibilité rigoureuses et une validation externe. Il est impératif de commencer par une étude de preuve de concept sur données simulées et sur une cohorte bien caractérisée (COPDGene ou SPIROMICS) avant de passer aux phases multi-cohorte et omique. Le protocole GO/NO-GO et les prédictions falsifiables sont des atouts majeurs pour la crédibilité du projet.", "brief_quality_gate": true, "revision_guidance": [], "llm_verdict": "publish_brief", "llm_consensus_score": 7.2 } }, "vulgarization_fr": { "title_fr": "Maladies respiratoires : et si on traquait leur progression comme on suit une maladie du cerveau ?", "hypothesis_in_brief": "La BPCO et l'asthme sont des maladies très différentes d'un patient à l'autre, mais on les traite souvent de la même façon. Cette hypothèse propose d'utiliser un algorithme, déjà employé avec succès pour les maladies neurodégénératives, pour analyser les données de souffle (spirométrie) de milliers de patients. L'idée est de voir si l'on peut identifier, non pas une seule évolution, mais plusieurs 'voies' distinctes de déclin pulmonaire, chacune correspondant à un mécanisme biologique différent.", "why_it_matters": "Aujourd'hui, un médecin peut constater que la fonction pulmonaire d'un patient se dégrade, mais il a du mal à prédire à quelle vitesse ni pourquoi. Si l'on pouvait classer les patients en sous-groupes (par exemple, ceux dont le souffle baisse très vite, ou ceux dont les poumons se détruisent de l'intérieur), cela changerait la donne. Cela permettrait d'adapter les traitements, de mieux concevoir les essais cliniques pour tester de nouveaux médicaments, et in fine, de proposer une médecine plus personnalisée pour des maladies qui touchent des centaines de millions de personnes dans le monde.", "imagine_that": "Imaginez que vous êtes un réparateur de vélos. Vous avez devant vous des dizaines de vélos qui grincent tous, mais vous ne savez pas pourquoi. Au lieu de les réparer un par un au hasard, vous décidez de chronométrer l'ordre exact dans lequel les bruits apparaissent : d'abord un grincement au pédalier, puis un cliquetis au dérailleur, puis un frottement de la roue. En répétant l'opération sur tous les vélos, vous vous apercevez qu'il n'y a que trois 'histoires' de pannes différentes. Chaque histoire correspond à un problème mécanique spécifique (chaine usée, roulement cassé, etc.). L'algorithme SuStaIn fait la même chose avec les poumons : il cherche des 'histoires' de dégradation qui se répètent d'un patient à l'autre.", "concretely": { "intro": "Pour tester cette idée, le protocole prévoit trois phases, allant de la simulation sur ordinateur jusqu'à la validation sur de grandes bases de données de patients réels.", "phase1": "D'abord, l'algorithme sera testé sur des données publiques (comme celles de la cohorte COPDGene) et sur des données artificielles créées par ordinateur, où l'on connaît déjà la 'bonne réponse'. Cela permet de vérifier que la méthode ne se trompe pas et qu'elle peut retrouver des sous-types qu'on a volontairement cachés dans les données.", "phase2": "Ensuite, l'algorithme sera appliqué à une seule grande cohorte de patients atteints de BPCO (environ 4000 personnes). L'objectif est de voir si l'on obtient bien au moins deux sous-types avec une différence de déclin du souffle d'au moins 50 millilitres par an, un seuil cliniquement significatif.", "phase3": "Enfin, la découverte sera vérifiée sur plusieurs autres cohortes indépendantes, y compris des patients asthmatiques et atteints de fibrose pulmonaire. On cherchera à lier chaque sous-type à des données biologiques (analyses de sang, scanners) pour confirmer qu'il s'agit bien de mécanismes distincts." }, "reviewers_say": "Le panel de relecteurs est globalement enthousiaste, notant l'hypothèse à 6.6/10. Ils saluent la rigueur du protocole, avec des prédictions chiffrées claires (comme une différence de déclin du souffle d'au moins 50 mL/an) et un plan en phases avec des points de décision. L'originalité de transposer un algorithme des neurosciences à la pneumologie est jugée prometteuse. Cependant, un relecteur 'contrarien' émet un doute sérieux : les mesures de souffle sont très fluctuantes (selon les traitements, les infections), contrairement aux images du cerveau. Il craint que l'algorithme ne soit noyé par le bruit et ne produise des sous-types artificiels. Les autres experts estiment que ce risque est gérable avec des analyses de sensibilité. Le verdict est de publier le brief, mais en commençant par la phase de simulation pour trancher ce débat." }, "vulgarization_en": { "title": "Respiratory diseases: what if their progression were tracked in the same way as a brain disease?", "hypothesis_in_brief": "COPD and asthma are highly heterogeneous diseases, yet they are often treated identically. This hypothesis proposes the use of an algorithm, already applied successfully to neurodegenerative diseases, to analyse breath data (spirometry) from thousands of patients. The aim is to determine whether multiple distinct pathways of pulmonary decline—each corresponding to a different biological mechanism—can be identified, rather than a single disease trajectory.", "why_it_matters": "Today, a physician may observe that a patient’s lung function is declining, but predicting the rate or cause of this decline remains difficult. If patients could be classified into subgroups — for instance, those whose breathing capacity deteriorates rapidly, or those whose lungs are destroyed from within — this would be transformative. Such classification would enable tailored treatments, improve the design of clinical trials for testing new drugs, and ultimately advance personalised medicine for diseases affecting hundreds of millions of people worldwide.", "imagine_that": "Imagine that you are a bicycle repairer. Before you are dozens of bicycles, all creaking, but you do not know why. Instead of repairing them one by one at random, you decide to time the exact order in which the noises appear: first a creak at the bottom bracket, then a click at the derailleur, then a rubbing of the wheel. By repeating the operation on all the bicycles, you notice that there are only three different failure 'stories'. Each story corresponds to a specific mechanical problem (worn chain, broken bearing, etc.). The SuStaIn algorithm does the same with the lungs: it looks for degradation 'stories' that repeat from one patient to another.", "reviewers_say": "The panel of reviewers is broadly enthusiastic, rating the hypothesis at 6.6/10. They commend the rigour of the protocol, with clear quantitative predictions (such as a difference in lung function decline of at least 50 mL/year) and a phased plan with decision points. The originality of transposing a neuroscience algorithm to pulmonology is deemed promising. However, one ‘contrarian’ reviewer raises a serious doubt: lung function measurements are highly variable (depending on treatments, infections), unlike brain images. This reviewer fears that the algorithm will be overwhelmed by noise and will produce artificial subtypes. The other experts consider this risk manageable with sensitivity analyses. The verdict is to publish the brief, but to begin with the simulation phase in order to settle this debate.", "concretely": { "intro": "To test this hypothesis, the protocol comprises three phases, ranging from computer simulation to validation on large real-patient databases.", "phase1": "First, the algorithm will be tested on public data (such as those from the COPDGene cohort) and on computer-generated synthetic data, for which the ‘correct answer’ is already known. This permits verification that the method does not err and that it can recover subtypes that were deliberately concealed within the data.", "phase2": "The algorithm will then be applied to a single large cohort of patients with COPD (approximately 4,000 individuals). The objective is to determine whether at least two subtypes are obtained, characterised by a difference in the rate of lung function decline of at least 50 millilitres per year, a clinically significant threshold.", "phase3": "Finally, the finding will be verified across several additional independent cohorts, including patients with asthma and pulmonary fibrosis. Each subtype will be linked to biological data (blood analyses, scans) to confirm that these are indeed distinct mechanisms." } }, "panel_en": { "reviews": [ { "reviewer_persona": "methodologist", "overall_score": 7.2, "verdict": "accept", "confidence": 0.82, "strengths": [ "The phased structure is well organised, with clear GO/NO-GO/PIVOT criteria that permit objective decision-making and pragmatic adaptation of the protocol in light of interim results.", "Rigorous pre-specification of falsifiable predictions is provided, with precise quantitative bounds (difference of ≥50 mL/year, HR ≥2.0, ARI >0.80) and clearly stated null hypotheses.", "A multi-cohort validation plan is included, comprising external replication and mechanistic characterisation via multi-omics, which substantially strengthens external validity and biological plausibility." ], "weaknesses": [ "Absence of control for the effect of number of visits and follow-up duration on subtype stability: the protocol does not compare SuStaIn with a standard longitudinal clustering method (e.g., k-means on individual slopes, trajectory mixture models) to demonstrate specific added value.", "Risk of unaddressed confirmation bias: the z-score thresholds for events (e.g., -1.64, -2.33, -3.0) are arbitrary and could be optimised to yield the expected subtypes. No sensitivity analysis on these thresholds is planned.", "Statistical power for the comparison of mortality hazard ratios between subtypes is not calculated. With a low-risk subtype (e.g., 30% of the sample) and a hazard ratio of 2.0, the required number of events may not be reached in the smallest cohorts (e.g., SARP III, IPF-PRO)." ], "critical_questions": [ "How does the panel intend to demonstrate that SuStaIn provides additional information beyond a simple classification based on FEV1 slope (e.g., rapid decline >60 mL/year vs. slow)? A growth mixture model would constitute a more relevant control than GOLD staging.", "What is the analysis plan if the optimal number of subtypes (K) differs between cohorts (e.g., K=2 in COPDGene, K=3 in SPIROMICS)? How will the subtypes be matched in this case?", "Is a correction for multiple comparisons planned in the analysis of the 10,000 differentially expressed genes (Phase 3)? Is the FDR threshold of <0.05 maintained across all tests or per subtype?" ], "recommendation": "This protocol is generally well designed and merits funding, with moderate reservations. The panel recommends the addition of a sensitivity analysis on z-score thresholds and a methodological control (e.g., comparison with a naive clustering on slopes) in Phase 1. A formal power calculation for the survival analyses is required before Phase 2, together with a contingency plan should power be insufficient (e.g., pooling of low-risk subtypes)." }, { "reviewer_persona": "domain_expert", "overall_score": 7.8, "verdict": "accept", "confidence": 0.82, "strengths": [ "The hypothesis is judged to be theoretically coherent and is supported by a robust epistemological framework (SuStaIn) already validated in several neurodegenerative and psychiatric pathologies, which justifies its transfer to chronic respiratory diseases. The proposed causal chain is logical and well articulated, with clear stages of collection, modelling, assignment and validation.", "The bibliographic foundation is of high quality, including direct evidence for the transferability of SuStaIn (schizophrenia, epilepsy, PSP) and indirect evidence for the underlying heterogeneity of COPD (CT subtypes, multi-omic clustering). The distinction between 'analogue' and 'indirect' evidence is rigorous and demonstrates a nuanced understanding of the state of the art.", "The identification of 'known unknowns' (K variable by disease, stability of subtypes across reference equations) is a sign of scientific maturity and demonstrates realistic anticipation of methodological challenges, which strengthens the plausibility of the approach." ], "weaknesses": [ "The hypothesis rests on the strong assumption (Assumption 1) that the progression of lung function follows a finite number of discrete ordered sequences. However, the literature on COPD suggests a continuum of trajectories with substantial phenotypic overlap (e.g., emphysema-small airways disease overlap). Violation of this assumption could lead to artefactual or unstable subtypes.", "Dependence on dense longitudinal data (≥3 points over ≥2 years) represents a major limitation. Many clinical COPD cohorts (e.g., SPIROMICS, COPDGene) possess longitudinal data but with variable frequency and considerable attrition. Practical feasibility is underestimated, particularly for diseases such as asthma, where seasonal and therapeutic variations complicate biomarker monotonicity (Assumption 3).", "The proposed mechanism ignores the potential role of treatments (inhaled corticosteroids, bronchodilators), which can modify FEV₁ decline trajectories. SuStaIn, in its current formulation, does not model therapeutic effects, a factor that could bias the identification of 'natural' progression subtypes." ], "critical_questions": [ "How will the algorithm manage patients whose PFT biomarkers show transient improvements (e.g., bronchodilator response) or prolonged plateaux, thereby violating the strict monotonicity hypothesis? Is a version of SuStaIn that permits non-monotonic events under consideration?", "What is the strategy for determining the optimal number of subtypes K in a context where model selection criteria (e.g., cross-validated likelihood) may favour over-parameterised solutions, particularly with samples of moderate size (N=500)?", "Will the identified subtypes be validated against therapeutic response data (e.g., response to inhaled corticosteroids in asthma) or molecular biomarkers (e.g., bronchoalveolar lavage proteomics) to establish their mechanistic distinctiveness, or only against composite clinical criteria?" ], "recommendation": "This hypothesis is promising and merits testing. It is recommended that the panel begin with a proof-of-concept study on a well-characterised cohort such as COPDGene or SPIROMICS, including CT data to validate the progression subtypes. The stability of the subtypes should be assessed through sensitivity analyses (bootstrap, permutations), and the results should be compared with alternative longitudinal clustering methods (e.g., mixed-effects models with latent classes). The inclusion of therapeutic covariates in the model would represent a major advantage for clinical plausibility." }, { "reviewer_persona": "contrarian", "overall_score": 3.5, "verdict": "weak_reject", "confidence": 0.85, "strengths": [ "The application of SuStaIn to longitudinal data represents a novel approach that has been validated in neurodegenerative diseases, and the selection of well-characterised cohorts such as COPDGene is considered judicious.", "The formulation of precise quantitative predictions (a difference of ≥50 mL/year in FEV₁ decline, HR ≥2.0) is commendable and facilitates the falsification of the hypothesis." ], "weaknesses": [ "FAIL REASON #1: Non-monotonicity and intra-individual variability of PFTs. Unlike neurodegenerative biomarkers (cerebral atrophy, tau proteins), PFT parameters (FEV1, FVC, DLCO) fluctuate considerably over time owing to exacerbations, treatments, adherence and measurement errors. The hypothesis of monotonic progression (Assumption 3) is violated in a substantial fraction of patients, which will undermine the capacity of SuStaIn to infer ordered and stable sequences of events. Temporal noise will dominate the progression signal.", "FAIL REASON #2: The number of time points (≥3) is insufficient for estimating reliable decline trajectories and sequences of events. With only 3 to 4 visits over 5 years, SuStaIn cannot distinguish 'rapid decline' from 'slow decline' with acceptable uncertainty. The method was designed for dense data (≥5–10 points) or irreversible binary events. Here, the model will be underdetermined and will produce unstable subtypes, even with N=2000. The promised ARI of 0.85 is unrealistic.", "FAIL REASON #3: Confusion between inter-individual heterogeneity and discrete subtypes. The hypothesis (Assumption 1) that pulmonary function decline follows a finite number of discrete ordered sequences is an oversimplification. PFT data instead show a continuum of phenotypes (e.g., emphysema vs. chronic bronchitis) with considerable overlap. SuStaIn will force artificial clusters within this continuum, producing subtypes that are not 'mechanistically distinct' but artefacts of the clustering method. Differences in FEV1 decline will be statistically significant but clinically trivial and non-reproducible in an external cohort." ], "critical_questions": [ "How does the panel intend to address the major problem of non-monotonicity in PFTs, given that 30–50% of COPD patients show an improvement in FEV₁ ≥100 mL between two consecutive visits (exacerbations, bronchodilator response)? Has a data pre-treatment (smoothing, exclusion of exacerbations) been planned that does not bias the sequences?", "With only 3–4 time points per patient, how is it demonstrated that SuStaIn is not simply overfitting measurement noise? Has a power analysis been performed to determine the minimum number of visits required to detect K=2 subtypes with an ARI >0.8?" ], "recommendation": "Before submission, feasibility must be demonstrated on realistic simulated data that include noise, fluctuations and missing values. It must also be justified that the number of time points is sufficient (≥5 visits per patient) or a version of SuStaIn that tolerates sparse data must be used. Finally, inclusion of external validation on an independent cohort (e.g., SPIROMICS) is essential to rule out clustering artefacts." }, { "reviewer_persona": "industrialist", "overall_score": 7.5, "verdict": "accept", "confidence": 0.75, "strengths": [ "A clear and significant addressable market: respiratory medicine (COPD, asthma, pulmonary fibrosis) represents a global market of >$30 billion. The identification of progression subtypes would permit the repositioning of existing drugs (e.g., roflumilast for 'rapid decliners', anti-IL-5 biologics for eosinophilic subtypes) and the enrichment of clinical trials, reducing Phase II/III costs by 30–50%.", "A defensible competitive advantage: unlike static 'clustering' approaches (clinical phenotyping), SuStaIn captures the temporal dynamics of biomarkers. This provides an advantage for partnerships with CROs (e.g., IQVIA, Parexel) seeking to improve patient stratification in COPD trials, where the rate of FEV₁ decline is a key but noisy endpoint.", "A realistic commercialisation timeline: the proposed protocol (8–14 months, €20k–80k) is very lean. A minimum viable product (a SaaS algorithm for trial data analysis) could be operational within 18–24 months, targeting initially the R&D departments of Big Pharma (GSK, AstraZeneca, Boehringer Ingelheim) as a 'patient enrichment' tool." ], "weaknesses": [ "Data access barrier: Multi-cohort validation (Phase 3) requires access to high-quality longitudinal data (COPDGene, SPIROMICS, etc.). Obtaining data licences from academic consortia can take 6–12 months and cost >€50k, with no guarantee of exclusivity. This represents a critical bottleneck.", "Risk of IP commoditisation: The SuStaIn algorithm itself is open source (pySuStaIn). The intellectual property resides in the biomarker signatures specific to pulmonary diseases. If a competitor (e.g., Tempus, Owkin) applies the same algorithm to its own data, the competitive advantage erodes rapidly. The trajectory signatures must be patented as a diagnostic method.", "Low perceived value without omic validation: Large pharmaceutical companies pay for subtypes that are actionable (linked to a drug target). Without multi-omic validation (Phase 3), the tool remains a 'nice clustering' without mechanism. The risk is that the Phase 3 budget (>€100k) cannot be secured without a committed pharma partner." ], "critical_questions": [ "What is the revenue model? Sale of SaaS licences to CROs (€50k–100k per trial), or a consulting service for proprietary data analysis (€200k–500k per project)? The former is scalable; the latter is more profitable but less defensible.", "How is the intellectual property for the 'trajectory signatures' protected? Is a method patent (USPTO) on 'A method for identifying a rapid decliner subtype in COPD using SuStaIn on longitudinal FEV1 and DLCO data' envisaged? Without such protection, a competitor could replicate the workflow within three months.", "What is the plan for clinician adoption? A radiologist or pulmonologist will not use an algorithm that requires five PFT visits. The product must be integrated into an EHR workflow (e.g., Epic, Cerner) to be used routinely. Is an EHR partner in place?" ], "recommendation": "Phase 1 (in silico) is to be launched immediately in-house (€2k, 4 weeks) to validate feasibility on public data (COPDGene). In parallel, 2–3 early-stage biotechnology companies (e.g., 4D Molecular Therapeutics, Pulmatrix) are to be approached for a co-funded collaborative research agreement (€50k) that would cover Phase 2 and provide privileged access to their phase I data. The objective is to produce a publishable case study and file a provisional patent within 12 months, prior to seeking a licensing deal with a CRO or a mid-sized pharmaceutical company." }, { "reviewer_persona": "funding_strategist", "overall_score": 7.2, "verdict": "accept", "confidence": 0.75, "funding_programs": [ { "program": "ERC Starting Grant 2026", "agency": "European Research Council", "fit_score": 0.85, "typical_budget": "€1.5M - €2.5M (sur 5 ans)", "success_rate": "14-17% (2023-2024)", "next_deadline": "Octobre 2025 (panel LS4 - Physiology, Pathophysiology and Endocrinology)", "rationale": "L'hypothèse est parfaitement alignée avec les critères ERC: haut risque scientifique (application d'un algorithme de neurosciences à la pneumologie), potentiel de rupture (identification de sous-types mécanistiques), et PI jeune (idéal pour Starting Grant). Le protocole GO/NO-GO démontre une rigueur méthodologique qui rassure les évaluateurs. Le panel LS4 est le plus adapté pour la physiologie respiratoire et les maladies chroniques." }, { "program": "ANR AAPG 2025 (Appel à Projets Générique) - Projet de Recherche Collaborative", "agency": "Agence Nationale de la Recherche (France)", "fit_score": 0.75, "typical_budget": "€200k - €600k (sur 36-48 mois)", "success_rate": "15-20% (2024)", "next_deadline": "Mars 2025 (axe 'Santé et biotechnologies')", "rationale": "L'ANR finance des projets exploratoires à fort potentiel translationnel. La Phase 1 in silico (€20k-€50k) peut être présentée comme un projet court et focalisé, avec des objectifs clairs et falsifiables. L'ANR valorise les collaborations interdisciplinaires (mathématiques appliquées + pneumologie). Le budget modeste et le timeline court (8-14 mois) sont un atout pour un AAPG." }, { "program": "NIH R01 (Research Project Grant) - NHLBI", "agency": "National Institutes of Health (USA) - National Heart, Lung, and Blood Institute", "fit_score": 0.7, "typical_budget": "$500k - $1.5M (sur 4-5 ans)", "success_rate": "18-22% (NHLBI, 2023)", "next_deadline": "Février 2025 (cycle standard) ou Octobre 2025", "rationale": "Le NHLBI finance des projets sur la BPCO, l'asthme et les maladies pulmonaires restrictives. L'hypothèse répond à un intérêt majeur: la stratification des patients pour des essais cliniques personnalisés. Le NIH valorise les approches computationnelles et les validations multi-cohortes (Phase 3). Un R01 serait idéal pour financer la Phase 2 et 3, avec un accent sur l'accès aux cohortes américaines (SPIROMICS, COPDGene)." } ], "strengths": [ "A mechanistically strong and falsifiable hypothesis is presented: the application of a validated neuroscience algorithm (SuStAIn) to longitudinal pulmonary function data constitutes an original, high-risk/high-reward approach, with a clear GO/NO-GO protocol that reassures reviewers of the scientific rigour.", "Clinical breakthrough potential is identified: distinguishing distinct progression subtypes (e.g., rapid FEV₁ decline versus predominant emphysema) in heterogeneous diseases such as COPD and asthma could transform patient stratification and clinical trial design, addressing a major unmet need in respiratory medicine.", "The TRL and trajectory are well defined: the three-phase protocol (in silico, single cohort, multi-cohort) demonstrates progressive maturation from TRL 2–3 (concept verified on synthetic data) to TRL 5–6 (multi-cohort validation with omics data), which is ideal for translational funding calls." ], "weaknesses": [ "Risk of non-generalisability: SuStaIn was designed for high-dimensional data (brain imaging). PFT data (spirometry, volumes, DLCO) are of lower dimensionality and contain fewer sequential events, which may limit the algorithm's capacity to identify discrete and reproducible subtypes.", "Dependence on longitudinal data quality: the hypothesis relies on cohorts with repeated measurements over several years. Access to standardised, well-phenotyped longitudinal PFT data (with few missing data) represents a major bottleneck, particularly for restrictive diseases (e.g., PI) for which cohorts are rarer.", "Lack of immediate mechanistic validation: the protocol proposes multi-omic validation in Phase 3, but without guarantee that the subtypes identified by SuStaIn correspond to distinct biological pathways. A reviewer might request a mechanistic proof of concept (e.g., transcriptomic profiling) as early as Phase 2." ], "critical_questions": [ "How will access to high-quality longitudinal PFT data for Phase 2 (e.g., the ECLIPSE cohort, SPIROMICS, or UK Biobank) be guaranteed, and what negotiations are already under way with data stewards?", "What is the contingency plan if SuStaIn fails to identify subtypes with a difference of ≥50 mL/year in the Phase 1 in silico analysis? Are algorithm modifications (e.g., addition of covariates, dimensionality reduction) or a change of algorithm (e.g., latent class analysis) envisaged?", "How is the integration of omics data (proteomics, metabolomics) into Phase 3 planned to demonstrate mechanistic distinction, and what collaborations have already been established with multi-omics platforms (e.g., EMBL-EBI, BBMRI)?" ], "recommendation": "The primary funding target is the ERC Starting Grant (2026) to support the full protocol (Phase 1–3) over five years, with emphasis placed on originality and high risk. In parallel, an exploratory application is to be submitted to the ANR (AAPG 2025) for Phase 1 *in silico* (budget €20k–€50k) in order to generate robust proof-of-concept evidence prior to the ERC submission. Should the ERC application fail, a subsequent ERC Proof of Concept call (2027) is to be pursued if Phase 1–2 results are positive." } ], "meta_review": { "consensus_score": 6.59, "verdict": "publish_brief", "brief_quality_gate": true, "llm_verdict": "publish_brief", "llm_consensus_score": 7.2, "key_consensus": [ "The hypothesis is judged to be theoretically coherent and original, and is supported by a robust epistemological framework (SuStaIn) already validated in other domains, which justifies its transfer to respiratory diseases.", "The protocol is generally well-structured, with precise quantitative predictions (≥50 mL/year, HR ≥2.0) and a multi-cohort validation plan that strengthens external validity.", "The translational and clinical potential is high: patient stratification, enrichment of clinical trials, and a response to a major unmet need in pulmonology.", "Practical feasibility is conditional upon access to high-quality longitudinal data and the management of non-monotonicity in PFTs (exacerbations, treatments)." ], "key_disagreements": [ "The Contrarian considers that the non-monotonicity and intra-individual variability of the PFTs, together with the low number of time points (≥3), render the hypothesis unrealistic and will produce artefactual subtypes; the other reviewers judge these challenges to be surmountable with sensitivity analyses and adequate pre-processing.", "The Contrarian considers that the hypothesis of a finite number of discrete ordered sequences is an excessive simplification in the face of the continuum of phenotypes observed in COPD; the other reviewers acknowledge this risk but judge it to be testable through comparisons with alternative longitudinal clustering methods.", "The Methodologist and the Industrialist highlight a lack of control for the effect of the number of visits and the duration of follow-up, as well as a risk of confirmation bias regarding the z-score thresholds; the Domain Expert and the Funding Strategist emphasise the dependence on data quality and the risk of non-generalisability." ], "revision_guidance": [], "critical_path": "The demonstration of feasibility on realistic simulated data (including noise, fluctuations, and missing values) and the validation of subtype stability through sensitivity analyses (bootstrap, permutations) and comparison with alternative longitudinal clustering methods (growth mixture models) constitute the most decisive factor for the success or failure of this hypothesis.", "final_recommendation": "The panel recommends publication of this brief, as the hypothesis is promising, mechanistically strong, and addresses a major clinical need. The methodological reservations (non-monotonicity, low number of time points, risk of clustering artefacts) are real but can be addressed through rigorous sensitivity analyses and external validation. It is imperative to begin with a proof-of-concept study on simulated data and a well-characterised cohort (COPDGene or SPIROMICS) before proceeding to the multi-cohort and omic phases. The GO/NO-GO protocol and the falsifiable predictions are major assets for the credibility of the project." } } }