Discovering Signature Disease Trajectories in Pancreatic Cancer and Soft-tissue Sarcoma from Longitudinal Patient Records

  • 0McWilliams School of Biomedical Informatics, The University of Texas Health Science Center at Houston, Houston, TX, USA.

Summary

This summary is machine-generated.

This study reveals key disease progression pathways for rare cancers like pancreatic cancer and soft tissue sarcoma (STS). These findings can help identify early clinical markers for these challenging diseases.

Area Of Science

  • Oncology
  • Medical Informatics
  • Data Science

Background

  • Understanding disease progression is crucial for clinical insights.
  • Rare cancers, including pancreatic cancer and specific types of soft tissue sarcoma (STS), present unique challenges in diagnosis and treatment.
  • Longitudinal patient data offers a valuable resource for studying disease trajectories.

Purpose Of The Study

  • To discover signature disease trajectories for pancreatic cancer, soft tissue sarcoma of the trunk and extremity (STS-TE), and soft tissue sarcoma of the abdomen and retroperitoneum (STS-AR).
  • To identify potential early clinical markers for these rare cancers by analyzing diagnosis sequences.
  • To validate a methodology for uncovering disease pathways using electronic health records.

Main Methods

  • Leveraged the IQVIA Oncology Electronic Medical Record database for patient data.
  • Employed matched cohort sampling, statistical computation, and right-tailed binomial hypothesis testing to identify significant diagnosis pairs.
  • Visualized disease trajectories up to three progressions, both preceding and following the primary rare cancer diagnoses.

Main Results

  • Identified 266 significant diagnosis pairs for pancreatic cancer, 130 for STS-TE, and 118 for STS-AR.
  • Discovered numerous multi-step (2-hop and 3-hop) disease trajectories before and after the primary diagnoses across all studied rare cancers.
  • Validated the discovered trajectories using the UTHealth Electronic Health Records, confirming the method's feasibility and reliability.

Conclusions

  • The study successfully mapped significant disease trajectories for three rare cancer types.
  • Key clinical features within these trajectories may serve as potential early diagnostic markers for rare cancers.
  • The developed approach is generalizable to other diseases and real-world longitudinal patient data analysis.