Abstract: “Development and Validation of a Multimodal Ultrasound-Based Biomarker for Myofascial Pain” This proposal responds to RFA-AT-22-003 to address the need for a biomarker indicative of myofascial pain. Myofascial pain can affect many regions of the body and it is a key component of chronic low back pain in particular. Patients with chronic low back pain have a range of musculoskeletal pathologies perpetuating their pain syndrome in addition to the myofascial components, such as facet arthritis or stenosis. Hence, there is a significant clinical need to identify the components of chronic low back pain related to myofascial pain beyond use of the physical exam only. Such a biomarker would have immediate clinical diagnostic uses, as well as being important as a phenotyping tool and outcome measure in clinical trials. Advances in ultrasound technology have resulted in identification of several abnormalities in myofascial tissues related to myofascial pain, beyond identifying trigger points. In addition to echogenicity changes, these include shear wave elastography of muscles and fascia, and dynamic fascia tissue deformation capturing abnormalities in movement/ glide of fascia tissue during lumbar flexion. Despite the clinical need and the available technology, no comprehensive study has integrated these ultrasound measures to validate a biomarker for the myofascial component of chronic low back pain. First, we propose to perform two detailed ultrasound assessments and standardized physical exams (including pressure algometry for painful trigger points) on 160 subjects each with and without chronic low back pain, divided into 4 phenotypic groups with and without painful trigger points. We will correlate the ultrasound measures to the clinical phenotype. Second, we will then use deep learning approaches to construct explainable machine learning models which integrate these measures to classify and predict the myofascial components of chronic low back pain, with latent and/or active trigger points. As performance metrics, we will report on area under the curve (AUC), sensitivity, and specificity. Third, in the R33 phase will perform a single blinded, randomized controlled trial of dry needling versus sham needling in 80 patients with chronic low back pain and active trigger points. We will collect the ultrasound measures and perform a standardized examination for myofascial pain prior to the intervention and at a one-week follow-up. We will test the ability and performance metrics of the deep learning models to predict the intensity of myofascial pain prior to injection and changes in myofascial pain post-needling. We anticipate that this work will lead to a software module which can be incorporated into existing clinical ultrasound machines for assessment of the myofascial components of musculoskeletal pain.

Narrative: This will be two phase study to develop and validate a biomarker for lumbar myofascial pain based on ultrasound obtained measurements of the lumbar muscles and fascia. We will use advanced machine learning approaches and validation in a randomized controlled trial.