Translating temporal multi-disease pathways in multiple cancer types using real-world clinical data

Although several pre-cancer comorbidities have been described, the role of medically interpretable preceding diagnoses as multi-disease pathways (MDPs) remains underexplored in the U.S. population. We analyzed electronic health record data from Ochsner Health (Louisiana) spanning 2013-2022, encompassing more than 4.1 million patients. Disease trajectories were identified by analyzing pairs of diagnoses that exhibited significant temporal correlations. To construct three-diagnosis trajectories, we merged overlapping pairs (e.g.,D1→D2 and D2→C (cancer) were combined to form D1→D2→C). For each patient, all three-step diagnostic sequences were extracted, and the pairs within each triplet were tested for directiona lsignificance. We applied this analysis to ten high-burden cancers, breast, melanoma, lung, prostate, pancreatic, liver, bladder, leukemia, ovarian, and colorectal, which yielded 501 (11 significant), 507 (10),194 (2), 349 (8), 49, 132 (3), 111, 84, 39, and 155 directional associations, respectively. Significant directional trajectories to cancer included breast lump → abnormal breast imaging, joint disorder → overuse/soft-tissue disorder, joint disorder → shoulder lesion, acute sinusitis → acute upperrespiratory infection, osteoporosis → other bone disorder, knee osteoarthritis → joint disorder, majordepression → anxiety disorder, acute pharyngitis → acute upper respiratory infection, cough → breathingab normality, joint disorder → other soft-tissue disorder, and finger/toe deformity → bonedensity/structure disorder preceding breast cancer; acute viral hepatitis → liver fibrosis/cirrhosis, chronicviral hepatitis → liver cirrhosis (unspecified), and liver fibrosis/cirrhosis → other liver disease precedingliver cancer; urinary frequency → sexual dysfunction, joint disorder → shoulder lesion, benign prostatichyperplasia → abnormal tumor markers, knee osteoarthritis → joint disorder, erectile dysfunction →abnormal tumor markers, joint disorder → other soft-tissue disorder, obstructive/reflux uropathy →benign prostatic hyperplasia, and osteoarthritis → joint disorder preceding prostate cancer; osteoarthritis of knee → joint disorder → soft-tissue disorder preceding lung cancer; and actinic keratosis → seborrheickeratosis, osteoarthritis → joint disorder, allergic/vasomotor rhinitis → chronic sinusitis, and joint disorder→ soft-tissue disorder preceding melanoma cancer. An important consideration in this approach is that while these MDPs are statistically significant, they may not represent causative associations. For example, conditions such as erectile dysfunction may lead to age-appropriate screening in a high cancer risk population. Likewise, a diagnosis of breast lump may directly imply cancer prior to formal diagnosis. Biologically plausibility of all significant MDPs should be investigated individually.