Chronic pain is a highly prevalent health problem with tremendous cognitive, social, and economic costs to the
individual and society. Negative affect and associated disorders–including depression, anxiety, PTSD,
neuroticism, catastrophizing, and multisensory sensitivity–confer risk for the development of chronic pain after
surgery or injury. These correlations have been shown at the epidemiological, genetic, and brain structural levels,
which suggests that negative affect may be a consistent risk factor for multiple types of chronic pain, and
therefore an effective target for prevention and treatment. Quantitative predictive models based on negative
affect and related risk factors could help identify patients who are likely to develop pain after injury or surgery
and those who would respond better to psychological, behavioral, pharmacological, or multimodal treatment.
Such models could be readily adopted in clinical screening, but for this to happen in a useful and equitable way,
several barriers must be overcome. First, negative affect is often seen as superfluous to the biological
mechanisms that drive pain. Second, existing evidence linking them is restricted to statistical associations
between aggregate measures, rather than precise predictive models, and effect sizes are moderate but below
the level needed for clinical utility. Third, the relevant psychosocial risk factors and the extent of their effects is
likely to vary across ancestry, sex, and culture. This study capitalizes on our group’s work over the past 2 years
on large-sample genetic, phenotypic, and neuroimaging data analysis to make forward progress on these
challenges. In Aim 1, we will use modern statistical learning techniques to develop quantitative predictive models
linking negative affect and related psychosocial risk factors to multiple types of chronic pain in the 500,000-
person UK Biobank sample, linking psychosocial risk profiles to (a) Genome Wide Association data in >400,000
individuals of European descent and >20,000 samples of African and Asian descent, and (b) MRI measures of
brain structure and function on >60,000 individuals. In Aim 2, we investigate how these models apply to, and
how they can be customized for, diverse U.S. populations in the 477,000-person All Of Us study, including
customizations for racial/ethnic, sex, and socioeconomic characteristics. In Aim 3 (exploratory), we will assess
whether these risk profiles predict post-surgical pain in All Of Us (in approximately 40,000 individuals who have
undergone surgery) and in the U.S. Acute to Chronic Pain Signatures study, which tracks 2,800 patients
longitudinally pre- and post-surgery (knee arthroplasty or thoracic surgery) with psychosocial, functional,
imaging, and multi-omics measures. Together, this work will provide new, quantitative models of psychosocial
risks for post-surgical pain and beyond that are both readily scalable for clinical use and grounded in a genetic
and neurobiological framework.
