Predicting persistent pain after shoulder surgery

The goal of our study published in PAIN (January 2015) was to identify risk subgroups for predicting post-operative pain outcomes following shoulder arthroscopy. Our subgroups were made up of genetic and psychological factors believed to be players in the transition from acute to chronic pain conditions. We used a parallel cohort design and the first cohort (n = 190) experienced exercise-induced shoulder pain. The goal here was to identify risk subgroup candidates in a controlled setting for further validation in a more clinically relevant setting. The second cohort was an operative shoulder pain cohort (n = 150), assessed pre-operatively and followed for 12 months for post-operative pain outcomes. The take home messages from this study were that 1) a risk subgroup comprised of genetic predisposition to pain sensitivity (based on COMT variation) and pain catastrophizing appears to be robust in that it was predictive of outcomes in both cohorts; 2) this risk subgroup could be considered as a candidate in other operative cohorts because there is nothing specific to shoulder pain in the risk factors; and 3) these data could be used as a starting point for future attempts at personalized pain relief. For those of you interested in some more background on this study here is the University of Florida release.

Beyond what was said in that piece, a few other points come to mind for the inquisitive readers of Body in Mind:

1. I hope this study gives motivation for others to consider exercise-induced injury as a viable pain phenotype for genetic studies. Most genetic studies have used transient induced experimental pain techniques (e.g. thermal, pressure, and/or chemical stimuli) and I wonder if their nature limits those phenotypes. While not a perfect match for clinical conditions, the pain generated following exercise-induced injury lasts for several days and seems to result in overlap with the lower third of some of our clinical samples. Admittedly the pain does resolve so it is not quite what happens with chronic pain but perhaps more studies using that phenotype would provide a different perspective on previous genetic findings using the transient phenotype methods.
2. The cutoff for pain catastrophizing in the subgroup was low. This was surprising to us but was empirically determined so we went with it. In our previous studies the cutoff scores were determined based on distribution of the data (i.e. 75^th percentile scores) and this study suggests even lower amounts of pain catastrophizing may be a risk factor when coupled with decreased genetic predisposition for pain modulation. You could play the chicken and egg game here with those factors, but since we measured them at the same time the best we can say is that the co-occurrence was what increased the risk.
3. The conundrum of predicting post-operative pain outcomes in musculoskeletal disorders is that pain is present pre-operatively. So it is difficult to determine whether your variable of interest is a true prognostic factor if the pre-operative differences carry over into the post-operative. In this cohort we did NOT see pre-operative differences for pain intensity which was encouraging (for full disclosure – we did report a pre-operative difference in our pilot study when a higher pain catastrophizing score was used to determine the subgroup – see George et al, PAIN 2008). For continuous measures of pain intensity (0-10 scores) – the largest difference in the subgroups was at 3 months, with the statistical difference remaining at 6 months, and then no longer there at 12 months. However, the categorical recovery criterion that was used (and determined a priori) was noticeably different at 12 months suggesting that the continuous pain scores may not have told the entire story. Also these findings suggested that the two groups started roughly at the same pre-operative intensity so the differences in recovery at 12 months can at least be partially attributed to risk subgroup status.
4. We did not do a good job assessing peri-operative pain management and the subsequent rehabilitation course. The original plan was to capture this by self report or medical record but we had a heck of time getting usable data. What we did capture was in the paper with sensitivity analyses were some key factors that could confound the results (e.g. depressive symptoms, medication status, and rotator cuff tear size), but there were not at the level of detail that would satisfy all people interested in this topic. Given the current interest and controversy in the importance of peri-operative pain management more work can be done here in subsequent studies.
5. The future is to see if these risk subgroups can be used to provide personalized pain management. Basically we are going to test whether combinations of pharmaceutical and behavioral interventions can make the high risk subgroup look more like the low risk subgroup. Our plan is to do that first in the pre-clinical setting and then hope to have enough “juice” to move it forward to a clinical trial. I think if we can successfully do that then we might be onto something. The funding for that project should be coming in soon, so I am hoping to get to spend another 3-5 years on this topic. And if all goes well will be able to update the readers of this blog with what happened… and the circle of life continues.
6. Please make sure you take a look at the co-author list, and the acknowledgments. This was truly a massive team effort, involving many disciplines and people at different stages of their careers. For example, beside the primary data this study served as a generator for some very interesting PHD and post-doctoral projects. I am grateful to have had everyone pitching in on this project as it helped immensely, much more than I could convey to those that helped.

Thank you to the powers that be at Body in Mind for the chance to write a bit more on our recent article in PAIN.

References

George SZ, Wallace MR, Wu SS, Moser MW, Wright TW, Farmer KW, Borsa PA, Parr JJ, Greenfield WH 3rd, Dai Y, Li H, & Fillingim RB (2015). Biopsychosocial influence on shoulder pain: risk subgroups translated across preclinical and clinical prospective cohorts. Pain, 156 (1), 148-56 PMID: 25599310

George SZ, Wallace MR, Wright TW, Moser MW, Greenfield WH 3rd, Sack BK, Herbstman DM, & Fillingim RB (2008). Evidence for a biopsychosocial influence on shoulder pain: pain catastrophizing and catechol-O-methyltransferase (COMT) diplotype predict clinical pain ratings. Pain, 136 (1-2), 53-61 PMID: 17686583

About Steven George

At the University of Florida, Dr. George is Associate Professor, Director of the Doctor of Physical Therapy Program, and Director of the Brooks-PHHP Research Collaboration. His teaching responsibilities include two courses in the Evidence-Based Practice sequence and his research interests involve the utilization of biopsychosocial models for the prevention and treatment of chronic pain. His research projects have been supported by awards from the National Institutes of Health, Department of Defense, Orthopaedic Section of the American Physical Therapy Association, University of Florida, and Foundation for Physical Therapy. With doctoral students and collaborators, Dr. George has authored over 160 peer-reviewed publications in physical therapy, rehabilitation, orthopedic, and pain research journals. He currently serves as an Editorial Board Member for Physical Therapy and an International Editorial Review Board Member for Journal of Orthopaedic & Sports Physical Therapy.