What Is My Patient's Pain Phenotype? An Exercise in Transferring Patient Clustering Solutions between Institutions and across Patient Samples

Abstract

Persistent post-mastectomy pain (PPMP) is a consequence of breast surgery that is variable in duration and severity across patients. The biopsychosocial model of pain acknowledges the impact of psychosocial and biophysical factors to modulate pain outcomes. The Rapid OPPERA Algorithm (ROPA) empirically clusters patients based on their biophysical and psychosocial characteristics and identified 3 clusters of patients based on 4 phenotypic characteristics with different risk of chronic pain. We aimed to understand the translatability of this type of phenotypic-based clustering, developed in patients with TMD, to a cohort undergoing breast cancer surgery, to predict pain severity and functional pain outcomes. Women (N:228, age:55.8±12.4, 87% White) scheduled for breast cancer surgery were prospectively enrolled in a longitudinal observational study. Preoperative pressure pain threshold (PPT) was measured using brief bedside QST, and anxiety, depression, and somatization were assessed using validated questionnaires. At 2-weeks, 3, 6, and 12-months after surgery, participants reported surgical area pain severity and frequency, and the impact of pain on cognitive/emotional and physical functioning. The ROPA clustering algorithm was applied using preoperative scores for anxiety, depression, somatization, and PPT. We determined differences between the identified clusters using general estimating equations linear regressions for longitudinal pain-related outcomes during the first year after surgery. Patients fell into three identified clusters: Adaptive (low psychosocial factors, low PPT; n=61), Pain Sensitive (moderate psychosocial factors, low PPT; n=125), and Global Symptoms (high psychosocial factors, moderate PPT; n=42). The Global Symptoms cluster, compared to other clusters, reported significantly greater pain severity and worse functional pain outcomes (p's<.05). The ROPA clustering algorithm identified a group at higher risk of PPMP, based on 4 preoperative phenotypic measures. Findings suggest that pain phenotyping may extend across patients with different diagnoses and clinical settings, highlighting the importance of personalized medicine, treating the "person type" as much as the diagnosis. Grant support from NIH/NIGMS: K23 GM110540.