Automated real-world data integration improves cancer outcome prediction

Justin Jee,Christopher Fong,Karl Pichotta,Thinh Ngoc Tran,Anisha Luthra,Michele Waters,Chenlian Fu,Mirella Altoe,Si-Yang Liu,Steven B. Maron,Mehnaj Ahmed,Susie Kim,Mono Pirun,Walid K. Chatila,Ino de Bruijn,Arfath Pasha,Ritika Kundra,Benjamin Gross,Brooke Mastrogiacomo,Tyler J. Aprati,David Liu,JianJiong Gao,Marzia Capelletti,Kelly Pekala,Lisa Loudon,Maria Perry,Chaitanya Bandlamudi,Mark Donoghue,Baby Anusha Satravada,Axel Martin,Ronglai Shen,Yuan Chen,A. Rose Brannon,Jason Chang,Lior Braunstein,Anyi Li,Anton Safonov,Aaron Stonestrom,Pablo Sanchez-Vela,Clare Wilhelm,Mark Robson,Howard Scher,Marc Ladanyi,Jorge S. Reis-Filho,David B. Solit,David R. Jones,Daniel Gomez,Helena Yu,Debyani Chakravarty,Rona Yaeger,Wassim Abida,Wungki Park,Eileen M. O'Reilly,Julio Garcia-Aguilar,Nicholas Socci,Francisco Sanchez-Vega,Jian Carrot-Zhang,Peter D. Stetson,Ross Levine,Charles M. Rudin,Michael F. Berger,Sohrab P. Shah,Deborah Schrag,Pedram Razavi,Kenneth L. Kehl,Bob T. Li,Gregory J. Riely,Nikolaus Schultz,Aaron Lisman,Gaofei Zhao,Ino de Bruijn,Walid K. Chatila,Xiang Li,Aarman Kohli,Darin Moore,Raymond Lim,Tom Pollard,Robert Sheridan,Avery Wang,Calla Chennault,Manda Wilson,Hongxin Zhang,Robert Pimienta,Surya Rangavajhala,Guru Subramanian,Jowel Garcia,Naveen Rachuri,Kevin Boehm,Mitchell Parker,Henry Walch,Subhiksha Nandakumar,Jordan Eichholz,Ayush Kris,Paolo Manca,Xuechun Bai,Tejiri Agbamu,Justin U,Xinran Bi

DOI: https://doi.org/10.1038/s41586-024-08167-5

IF: 64.8

2024-11-07

Nature

Abstract:The digitization of health records and growing availability of tumour DNA sequencing provide an opportunity to study the determinants of cancer outcomes with unprecedented richness. Patient data are often stored in unstructured text and siloed datasets. Here we combine natural language processing annotations 1,2 with structured medication, patient-reported demographic, tumour registry and tumour genomic data from 24,950 patients at Memorial Sloan Kettering Cancer Center to generate a clinicogenomic, harmonized oncologic real-world dataset (MSK-CHORD). MSK-CHORD includes data for non-small-cell lung ( n = 7,809), breast ( n = 5,368), colorectal ( n = 5,543), prostate ( n = 3,211) and pancreatic ( n = 3,109) cancers and enables discovery of clinicogenomic relationships not apparent in smaller datasets. Leveraging MSK-CHORD to train machine learning models to predict overall survival, we find that models including features derived from natural language processing, such as sites of disease, outperform those based on genomic data or stage alone as tested by cross-validation and an external, multi-institution dataset. By annotating 705,241 radiology reports, MSK-CHORD also uncovers predictors of metastasis to specific organ sites, including a relationship between SETD2 mutation and lower metastatic potential in immunotherapy-treated lung adenocarcinoma corroborated in independent datasets. We demonstrate the feasibility of automated annotation from unstructured notes and its utility in predicting patient outcomes. The resulting data are provided as a public resource for real-world oncologic research.

multidisciplinary sciences