Sensory data-driven classification of complex regional pain syndrome: a multicentre cohort study

The underlying pathophysiological mechanisms of complex regional pain syndrome (CRPS) have been under debate in recent years. Previous studies identified mechanistic CRPS subtypes, which were characterised as 'cold' and 'warm' or peripheral and central phenotypes. Our aim was to examine CRPS patients through comprehensive somatosensory testing and identify potential subgroups using unbiased statistics. Thus, our approach differs fundamentally from previous top-down approaches which stratified along predefined pathophysiological mechanisms.

In total, 604 patients (age: 51.9 [SD 13.4] yr, female: 436, disease duration: 1.6 [SD 2.9] yr) with CRPS (type I: n=520; type II: n=84) underwent Quantitative Sensory Testing according to the DFNS protocol (German Research Network on Neuropathic Pain). We assessed 13 parameters, including thermal and mechanical detection and pain thresholds, indicating one distinct sensory profile for each participant. We conducted a hypothesis-free cluster analysis in a training set (A, n=380), which was re-evaluated in a validation set (B, n=224) to account for possible overfitting of the data and a comparison towards human pain surrogate models.

We identified three sensory phenotypes in the training set, which were confirmed in the validation set. The largest group showed pronounced hypersensitivity towards cold and heat pain (n=382), a second group showed loss of thermal and mechanical sensation (n=200), and a third, small, but consistent, group exhibited strong allodynia and mechanical hyperalgesia (n=22).

A novel bottom-up approach can stratify CRPS based on sensory phenotypes, adding to a mechanistic understanding through a comparative analysis of human pain surrogate models.