Validation and utility of ARDS subphenotypes identified by machine-learning models using clinical data: an observational, multicohort, retrospective analysis.

Pubmed ID: 35026177

Pubmed Central ID: PMC8976729

Journal: The Lancet. Respiratory medicine

Publication Date: April 1, 2022

Affiliation: Division of Clinical and Translational Research, Washington University School of Medicine, St Louis, MO, USA; Department of Anesthesia, Division of Critical Care, Washington University, St Louis, MO, USA. Electronic address: p.sinha@wustl.edu.

MeSH Terms: Humans, Retrospective Studies, Acute Lung Injury, Positive-Pressure Respiration, Machine Learning, Respiratory Distress Syndrome

Grants: R35 HL140026, K23 HL133489, R35 GM142992

Authors: Maddali MV, Churpek M, Pham T, Rezoagli E, Zhuo H, Zhao W, He J, Delucchi KL, Wang C, Wickersham N, McNeil JB, Jauregui A, Ke S, Vessel K, Gomez A, Hendrickson CM, Kangelaris KN, Sarma A, Leligdowicz A, Liu KD, Matthay MA, Ware LB, Laffey JG, Bellani G, Calfee CS, Sinha P, Rios F, Van Haren F, Sottiaux T, Lora FS, Azevedo LC, Depuydt P, Fan E, Bugedo G, Qiu H, Gonzalez M, Silesky J, Cerny V, Nielsen J, Jibaja M, Pham T, Wrigge H, Matamis D, Ranero JL, Hashemian SM, Amin P, Clarkson K, Bellani G, Kurahashi K, Villagomez A, Zeggwagh AA, Heunks LM, Laake JH, Palo JE, do Vale Fernandes A, Sandesc D, Arabi Y, Bumbasierevic V, Nin N, Lorente JA, Larsson A, Piquilloud L, Abroug F, McAuley DF, McNamee L, Hurtado J, Bajwa E, Démpaire G, Francois GM, Sula H, Nunci L, Cani A, Zazu A, Dellera C, Insaurralde CS, Alejandro RV, Daldin J, Vinzio M, Fernandez RO, Cardonnet LP, Bettini LR, Bisso MC, Osman EM, Setten MG, Lovazzano P, Alvarez J, Villar V, Milstein C, Pozo NC, Grubissich N, Plotnikow GA, Vasquez DN, Ilutovich S, Tiribelli N, Chena A, Pellegrini CA, Saenz MG, Estenssoro E, Brizuela M, Gianinetto H, Gomez PE, Cerrato VI, Bezzi MG, Borello SA, Loiacono FA, Fernandez AM, Knowles S, Reynolds C, Inskip DM, Miller JJ, Kong J, Whitehead C, Bihari S, Seven A, Krstevski A, Rodgers HJ, Millar RT, Mckenna TE, Bailey IM, Hanlon GC, Aneman A, Lynch JM, Azad R, Neal J, Woods PW, Roberts BL, Kol MR, Wong HS, Riss KC, Staudinger T, Wittebole X, Berghe C, Bulpa PA, Dive AM, Verstraete R, Lebbinck H, Depuydt P, Vermassen J, Meersseman P, Ceunen H, Rosa JI, Beraldo DO, Piras C, Ampinelli AMR, Nassar AP, Mataloun S, Moock M, Thompson MM, Gonçalves CH, Antônio ACP, Ascoli A, Biondi RS, Fontenele DC, Nobrega D, Sales VM, Shindhe S, Ismail DMABPH, Laffey J, Beloncle F, Davies KG, Cirone R, Manoharan V, Ismail M, Goligher EC, Jassal M, Nishikawa E, Javeed A, Curley G, Rittayamai N, Parotto M, Ferguson ND, Mehta S, Knoll J, Pronovost A, Canestrini S, Bruhn AR, Garcia PH, Aliaga FA, Farías PA, Yumha JS, Ortiz CA, Salas JE, Saez AA, Vega LD, Labarca EF, Martinez FT, Carreño NG, Lora P, Liu H, Qiu H, Liu L, Tang R, Luo X, An Y, Zhao H, Gao Y, Zhai Z, Ye ZL, Wang W, Li W, Li Q, Zheng R, Yu W, Shen J, Li X, Yu T, Lu W, Wu YQ, Huang XB, He Z, Lu Y, Han H, Zhang F, Sun R, Wang HX, Qin SH, Zhu BH, Zhao J, Liu J, Li B, Liu JL, Zhou FC, Li QJ, Zhang XY, Li-Xin Z, Xin-Hua Q, Jiang L, Gao YN, Zhao XY, Li YY, Li XL, Wang C, Yao Q, Yu R, Chen K, Shao H, Qin B, Huang QQ, Zhu WH, Hang AY, Hua MX, Li Y, Xu Y, Di YD, Ling LL, Qin TH, Wang SH, Qin J, Han Y, Zhou S, Vargas MP, Silesky Jimenez JI, González Rojas MA, Solis-Quesada JE, Ramirez-Alfaro CM, Máca J, Sklienka P, Gjedsted J, Christiansen A, Nielsen J, Villamagua BG, Llano M, Burtin P, Buzancais G, Beuret P, Pelletier N, Mortaza S, Mercat A, Chelly J, Jochmans S, Terzi N, Daubin C, Carteaux G, de Prost N, Chiche JD, Daviaud F, Pham T, Fartoukh M, Barberet G, Biehler J, Dellamonica J, Doyen D, Arnal JM, Briquet A, Hraiech S, Papazian L, Follin A, Roux D, Messika J, Kalaitzis E, Dangers L, Combes A, Au SM, Béduneau G, Carpentier D, Zogheib EH, Dupont H, Ricome S, Santoli FL, Besset SL, Michel P, Gelée B, Danin PE, Goubaux B, Crova PJ, Phan NT, Berkelmans F, Badie JC, Tapponnier R, Gally J, Khebbeb S, Herbrecht JE, Schneider F, Declercq PM, Rigaud JP, Duranteau J, Harrois A, Chabanne R, Marin J, Bigot C, Thibault S, Ghazi M, Boukhazna M, Ould Zein S, Richecoeur JR, Combaux DM, Grelon F, Le Moal C, Sauvadet EP, Robine A, Lemiale V, Reuter D, Dres M, Demoule A, Goldgran-Toledano D, Baboi L, Guérin C, Lohner R, Kraßler J, Schäfer S, Zacharowski KD, Meybohm P, Reske AW, Simon P, Hopf HF, Schuetz M, Baltus T, Papanikolaou MN, Papavasilopoulou TG, Zacharas GA, Ourailogloy V, Mouloudi EK, Massa EV, Nagy EO, Stamou EE, Kiourtzieva EV, Oikonomou MA, Avila LE, Cortez CA, Citalán JE, Jog SA, Sable SD, Shah B, Gurjar M, Baronia AK, Memon M, Muthuchellappan R, Ramesh VJ, Shenoy A, Unnikrishnan R, Dixit SB, Rhayakar RV, Ramakrishnan N, Bhardwaj VK, Mahto HL, Sagar SV, Palaniswamy V, Ganesan D, Mohammadreza Hashemian S, Jamaati H, Heidari F, Meaney EA, Nichol A, Knapman KM, O'Croinin D, Dunne ES, Breen DM, Clarkson KP, Jaafar RF, Dwyer R, Amir F, Ajetunmobi OO, O'Muircheartaigh AC, Black CS, Treanor N, Collins DV, Altaf W, Zani G, Fusari M, Spadaro S, Volta CA, Graziani R, Brunettini B, Palmese S, Formenti P, Umbrello M, Lombardo A, Pecci E, Botteri M, Savioli M, Protti A, Mattei A, Schiavoni L, Tinnirello A, Todeschini M, Giarratano A, Cortegiani A, Sher S, Rossi A, Antonelli MM, Montini LM, Casalena P, Scafetti S, Panarello G, Occhipinti G, Patroniti N, Pozzi M, Biscione RR, Poli MM, Raimondi F, Albiero D, Crapelli G, Beck E, Pota V, Schiavone V, Molin A, Tarantino F, Monti G, Frati E, Mirabella L, Cinnella G, Fossali T, Colombo R, Terragni P, Pattarino I, Mojoli F, Braschi A, Borotto EE, Cracchiolo AN, Palma DM, Raponi F, Foti G, Vascotto ER, Coppadoro A, Brazzi L, Floris L, Iotti GA, Venti A, Yamaguchi O, Takagi S, Maeyama HN, Watanabe E, Yamaji Y, Shimizu K, Shiozaki K, Futami S, Ryosuke S, Saito K, Kameyama Y, Ueno K, Izawa M, Okuda N, Suzuki H, Harasawa T, Nasu M, Takada T, Ito F, Nunomiya S, Koyama K, Abe T, Andoh K, Kusumoto K, Hirata A, Takaba A, Kimura H, Matsumoto S, Higashijima U, Honda H, Aoki N, Imai H, Ogino Y, Mizuguchi I, Ichikado K, Nitta K, Mochizuki K, Hashida T, Tanaka H, Nakamura T, Niimi D, Ueda T, Kashiwa Y, Uchiyama A, Sabelnikovs O, Oss P, Haddad Y, Liew KY, Ñamendys-Silva SA, Jarquin-Badiola YD, Sanchez-Hurtado LA, Gomez-Flores SS, Marin MC, Villagomez AJ, Lemus JS, Fierro JM, Cervantes MR, Mejia FJF, Gonzalez DR, Dector DM, Estrella CR, Sanchez-Medina JR, Ramirez-Gutierrez A, George FG, Aguirre JS, Buensuseso JA, Poblano M, Dendane T, Zeggwagh AA, Balkhi H, Elkhayari M, Samkaoui N, Ezzouine H, Benslama A, Amor M, Maazouzi W, Cimic N, Beck O, Bruns MM, Schouten JA, Rinia M, Raaijmakers M, Heunks LM, Van Wezel HM, Heines SJ, Buise MP, Simonis FD, Schultz MJ, Goodson JC, Rowne TSB, Navarra L, Hunt A, Hutchison RA, Bailey MB, Newby L, Mcarthur C, Kalkoff M, Mcleod A, Casement J, Hacking DJ, Andersen FH, Dolva MS, Laake JH, Barratt-Due A, Noremark KAL, Søreide E, Sjøbø BÅ, Guttormsen AB, Yoshido HHL, Aguilar RZ, Oscanoa FAM, Alisasis AU, Robles JB, Pasanting-Lim RAB, Tan BC, Andruszkiewicz P, Jakubowska K, Cox CM, Alvarez AM, Oliveira BS, Montanha GM, Barros NC, Pereira CS, Messias AM, Monteiro JM, Araujo AM, Catorze NT, Marum SM, Bouw MJ, Gomes RM, Brito VA, Castro S, Estilita JM, Barros FM, Serra IM, Martinho AM, Tomescu DR, Marcu A, Bedreag OH, Papurica M, Corneci DE, Negoita SI, Grigoriev E, Gritsan AI, Gazenkampf AA, Almekhlafi G, Albarrak MM, Mustafa GM, Maghrabi KA, Salahuddin N, Aisa TM, Al Jabbary AS, Tabhan E, Arabi YM, Trinidad OA, Al Dorzi HM, Tabhan EE, Bolon S, Smith O, Mancebo J, Aguirre-Bermeo H, Lopez-Delgado JC, Esteve F, Rialp G, Forteza C, De Haro C, Artigas A, Albaiceta GM, De Cima-Iglesias S, Seoane-Quiroga L, Ceniceros-Barros A, Ruiz-Aguilar AL, Claraco-Vega LM, Soler JA, Lorente MDC, Hermosa C, Gordo F, Prieto-González M, López-Messa JB, Perez MP, Pere CP, Allue RM, Roche-Campo F, Ibañez-Santacruz M, Temprano S, Pintado MC, De Pablo R, Gómez PRA, Ruiz SR, Moles SI, Jurado MT, Arizmendi A, Piacentini EA, Franco N, Honrubia T, Perez Cheng M, Perez Losada E, Blanco J, Yuste LJ, Carbayo-Gorriz C, Cazorla-Barranquero FG, Alonso JG, Alda RS, Algaba Á, Navarro G, Cereijo E, Diaz-Rodriguez E, Marcos DP, Montero LA, Para LH, Sanchez RJ, Blasco Navalpotro MA, Abad RD, Montiel González R, Toribio DP, Castro AG, Artiga MJD, Penuelas O, Roser TP, Olga MF, Curto EG, Sánchez RM, Imma VP, Elisabet GM, Claverias L, Magret M, Pellicer AM, Rodriguez LL, Sánchez-Ballesteros J, González-Salamanca Á, Jimenez AG, Huerta FP, Diaz JCJS, Lopez EB, Moya DDL, Alfonso AAT, Eugenio Luis PS, Cesar PS, Rafael SI, Virgilio CG, Recio NN, Adamsson RO, Rylander CC, Holzgraefe B, Broman LM, Wessbergh J, Persson L, Schiöler F, Kedelv H, Tibblin AO, Appelberg H, Hedlund L, Helleberg J, Eriksson KE, Glietsch R, Larsson N, Nygren I, Nunes SL, Morin AK, Kander T, Adolfsson A, Piquilloud L, Zender HO, Leemann-Refondini C, Elatrous S, Bouchoucha S, Chouchene I, Ouanes I, Ben Souissi A, Kamoun S, Demirkiran O, Aker M, Erbabacan E, Ceylan I, Girgin NK, Ozcelik M, Ünal N, Meco BC, Akyol OO, Derman SS, Kennedy B, Parhar K, Srinivasa L, McNamee L, McAuley D, Steinberg J, Hopkins P, Mellis C, Stansil F, Kakar V, Hadfield D, Brown C, Vercueil A, Bhowmick K, Humphreys SK, Ferguson A, Mckee R, Raj AS, Fawkes DA, Watt P, Twohey L, Thomas RRJM, Morton A, Kadaba V, Smith MJ, Hormis AP, Kannan SG, Namih M, Reschreiter H, Camsooksai J, Kumar A, Rugonfalvi S, Nutt C, Oneill O, Seasman C, Dempsey G, Scott CJ, Ellis HE, Mckechnie S, Hutton PJ, Di Tomasso NN, Vitale MN, Griffin RO, Dean MN, Cranshaw JH, Willett EL, Ioannou N, Gillis S, Csabi P, Macfadyen R, Dawson H, Preez PD, Williams AJ, Boyd O, De Gordoa LO, Bramall J, Symmonds S, Chau SK, Wenham T, Szakmany T, Toth-Tarsoly P, Mccalman KH, Alexander P, Stephenson L, Collyer T, Chapman R, Cooper R, Allan RM, Sim M, Wrathall DW, Irvine DA, Zantua KS, Adams JC, Burtenshaw AJ, Sellors GP, Welters ID, Williams KE, Hessell RJ, Oldroyd MG, Battle CE, Pillai S, Kajtor I, Sivashanmugave M, Okane SC, Donnelly A, Frigyik AD, Careless JP, May MM, Stewart R, Trinder TJ, Hagan SJ, Wise MP, Cole JM, MacFie CC, Dowling AT, Hurtado J, Hurtado J, Nuñez E, Pittini G, Rodriguez R, Imperio MC, Santos C, França AG, Ebeid A, Deicas A, Serra C, Uppalapati A, Kamel G, Banner-Goodspeed VM, Beitler JR, Mukkera SR, Kulkarni S, Lee J, Mesar T, Shinn Iii JO, Gomaa D, Tainter C, Mesar T, Cowley RA, Yeatts DJ, Warren J, Lanspa MJ, Miller RR, Grissom CK, Brown SM, Bauer PR, Gosselin RJ, Kitch BT, Cohen JE, Beegle SH, Gueret RM, Tulaimat A, Choudry S, Stigler W, Batra H, Huff NG, Lamb KD, Oetting TW, Mohr NM, Judy C, Saito S, Kheir FM, Schlichting AB, Delsing A, Elmasri M, Crouch DR, Ismail D, Blakeman TC, Dreyer KR, Gomaa D, Baron RM, Grijalba CQ, Hou PC, Seethala R, Aisiku I, Henderson G, Frendl G, Hou SK, Owens RL, Schomer A, Bumbasirevic V, Jovanovic B, Surbatovic M, Veljovic M

Cite As: Maddali MV, Churpek M, Pham T, Rezoagli E, Zhuo H, Zhao W, He J, Delucchi KL, Wang C, Wickersham N, McNeil JB, Jauregui A, Ke S, Vessel K, Gomez A, Hendrickson CM, Kangelaris KN, Sarma A, Leligdowicz A, Liu KD, Matthay MA, Ware LB, Laffey JG, Bellani G, Calfee CS, Sinha P, LUNG SAFE Investigators and the ESICM Trials Group. Validation and utility of ARDS subphenotypes identified by machine-learning models using clinical data: an observational, multicohort, retrospective analysis. Lancet Respir Med 2022 Apr;10(4):367-377. Epub 2022 Jan 10.

Studies:

Abstract

BACKGROUND: Two acute respiratory distress syndrome (ARDS) subphenotypes (hyperinflammatory and hypoinflammatory) with distinct clinical and biological features and differential treatment responses have been identified using latent class analysis (LCA) in seven individual cohorts. To facilitate bedside identification of subphenotypes, clinical classifier models using readily available clinical variables have been described in four randomised controlled trials. We aimed to assess the performance of these models in observational cohorts of ARDS. METHODS: In this observational, multicohort, retrospective study, we validated two machine-learning clinical classifier models for assigning ARDS subphenotypes in two observational cohorts of patients with ARDS: Early Assessment of Renal and Lung Injury (EARLI; n=335) and Validating Acute Lung Injury Markers for Diagnosis (VALID; n=452), with LCA-derived subphenotypes as the gold standard. The primary model comprised only vital signs and laboratory variables, and the secondary model comprised all predictors in the primary model, with the addition of ventilatory variables and demographics. Model performance was assessed by calculating the area under the receiver operating characteristic curve (AUC) and calibration plots, and assigning subphenotypes using a probability cutoff value of 0·5 to determine sensitivity, specificity, and accuracy of the assignments. We also assessed the performance of the primary model in EARLI using data automatically extracted from an electronic health record (EHR; EHR-derived EARLI cohort). In Large Observational Study to Understand the Global Impact of Severe Acute Respiratory Failure (LUNG SAFE; n=2813), a multinational, observational ARDS cohort, we applied a custom classifier model (with fewer variables than the primary model) to determine the prognostic value of the subphenotypes and tested their interaction with the positive end-expiratory pressure (PEEP) strategy, with 90-day mortality as the dependent variable. FINDINGS: The primary clinical classifier model had an area under receiver operating characteristic curve (AUC) of 0·92 (95% CI 0·90-0·95) in EARLI and 0·88 (0·84-0·91) in VALID. Performance of the primary model was similar when using exclusively EHR-derived predictors compared with manually curated predictors (AUC=0·88 [95% CI 0·81-0·94] vs 0·92 [0·88-0·97]). In LUNG SAFE, 90-day mortality was higher in patients assigned the hyperinflammatory subphenotype than in those with the hypoinflammatory phenotype (414 [57%] of 725 vs 694 [33%] of 2088; p<0·0001). There was a significant treatment interaction with PEEP strategy and ARDS subphenotype (p=0·041), with lower 90-day mortality in the high PEEP group of patients with the hyperinflammatory subphenotype (hyperinflammatory subphenotype: 169 [54%] of 313 patients in the high PEEP group vs 127 [62%] of 205 patients in the low PEEP group; hypoinflammatory subphenotype: 231 [34%] of 675 patients in the high PEEP group vs 233 [32%] of 734 patients in the low PEEP group). INTERPRETATION: Classifier models using clinical variables alone can accurately assign ARDS subphenotypes in observational cohorts. Application of these models can provide valuable prognostic information and could inform management strategies for personalised treatment, including application of PEEP, once prospectively validated. FUNDING: US National Institutes of Health and European Society of Intensive Care Medicine.