Our hypothesis that our knee JPS test would evoke greater response in somatosensory and motor cortices compared to simple knee flexion was confirmed by observations of greater response during angle reproduction in, for example, the precentral gyrus, middle frontal gyrus, insula and cingulate gyri. Our hypothesis that individuals with ACL reconstruction would show differences in brain response compared to asymptomatic controls was rejected due to a lack of significant differences between groups. Our hypothesis that knee JPS errors would correlate with response in related regions was on the other hand confirmed by associations between greater errors and BOLD response in the insula, anterior cingulate and supramarginal gyrus.
Our knee JPS test evoked response in the ipsilateral precentral gyrus for the right test side and cingulate gyri for both test sides. Response in these regions has also been observed among asymptomatic individuals during active knee flexion tasks of JPS13 and force matching32. Response in the right middle frontal gyrus was seen for both test sides and has been previously associated with switching between exogenous and endogenous attention33, relevant for our JPS task where the focus of attention changes from external instructions on a screen to internal sensations related to proprioception. Also common for both test sides was recruitment of the ipsilateral insula, previously associated with sensorimotor processes such as active and passive stepping motions34. This finding also aligns with the body image and body schema concepts of body representations35, in which current understandings attribute the insula with conscious perceptual representation of the body and memory36. The insula and anterior cingulate are further thought to play a key role in interoception, a term originally introduced by Sherrington37 to describe visceral sensations, but now often used as a broader term encompassing the subjective experience of the body state38 and even proprioception10. In fact damage to the insula due to stroke has been associated with poor position sense of the upper limbs39. The same study also found similar associations for the inferior frontal and superior temporal gyri, two areas that were activated during our JPS test in the current study. Thus, our findings add to previous evidence of neural networks involved in proprioceptive tasks and expand these to those associated with the lower limbs.
There were no significant between-group differences for brain response nor for knee JPS errors. Previous research investigating knee JPS and somatosensory evoked potentials of individuals 18 months after surgical reconstruction of the ACL also found a lack of difference compared with controls as evidence of sensory neurone regeneration40. The original version of the current supine knee JPS test also did not detect any significant differences in errors for a separate ACLR group approximately two years after surgical reconstruction compared with matched athletes21. In that study, less-active controls instead showed significantly greater JPS errors compared to the ACLR group, suggesting that activity level is a more important factor in this context. It is therefore possible that deficits in proprioception were not present among the individuals of our ACLR group, who were active and participated on average 23 months after surgical reconstruction. Additionally, a recent meta-analysis found that only knee JPS tests with passive rather than active movements differentiate between ACL-injured knees and those of asymptomatic controls9. The active movements of the current test, which also incorporated the hip, may further have contributed to the lack of between-group difference seen here. The target angles of 40° and 65° knee flexion used in our JPS test both lie close to the mid-range of motion for the joint. These angles may not have been optimal for elucidating differences between groups where joint receptors are the focus of investigation, given that they are believed to play a more predominant role towards the limits of joint rotation41. It is also possible that small group sizes (due to separating ACLR into right and left leg analyses), as well as the contrast to such a similar movement, may have reduced the sensitivity of our analysis.
Our results showed that greater knee JPS errors, i.e., poorer knee proprioceptive acuity, was associated with greater brain response in the ipsilateral insula for the left test side, as well as the ipsilateral anterior, paracingulate and supramarginal gyri for the right test side. These results are line with the Embodied Predictive Interoception Coding (EPIC) model proposed by Barrett and Simons42, which describes the process of active inference in interoception. Part of this model describes the role of the mid- and posterior insula in computing and transmitting prediction errors as well as the integration of other agranular visceromotor cortices such as the cingulate cortex in this process. This model may thus be relevant for proprioceptive tasks. The importance of the insula to position sense is further supported by findings previously mentioned here for the upper limbs following stroke, whereby lesions in this region were associated with greater errors when attempting to actively move the unaffected arm to the mirror-matched position of the passively moved contralateral arm39. Associated response for the right supramarginal gyrus is supported by previous findings of greater response during a proprioceptive task of force matching at the knee among asymptomatic females32.
A limitation of our study was the additional task of attempting a constant knee angular velocity, although this was similar for the contrast Flex condition. Further, the active movement of the whole lower limb meant that proprioceptive feedback was not isolated to the knee, but also incorporated the hip. Although this is more similar to everyday activities and thus may enhances the ecological validity of the task compared with a single-joint movement, compensations at the hip joint for potential deficits at the knee are possible. Despite the active task, head movement (range 0.14 - 0.88 mm) did not confound brain imaging analyses. Due to challenges in recruiting participants, we included ACLR individuals who had injured either knee, but a control group with only right-side dominance. Future studies with a greater number of and more homogenous participants are likely required to further elucidate potential group differences. Comparisons were thus made to the non-dominant and dominant legs of CTRL for the L-ACLR and R-ACLR groups respectively. Additionally, both sexes were represented in each of our groups. Two similar fMRI studies have however indicated potentially different functional brain connectivity between males43 and females44 who later suffered an ACL injury. Our eligibility criteria required a minimum physical activity level score of four according to the Tegner activity scale. Although we matched current activity level between groups, a sub-analysis using Wilcoxon Signed Ranks tests compared pre-injury and current activity levels within the ACLR groups and found that R-ACLR had significantly reduced their activity level (P = 0.011), but the level was not significantly changed among L-ACLR. A difference in activity level change from pre- to post-injury was thus a potentially confounding factor in our analyses. To explore this further, a Mann-Whitney U test to compare change in activity level between the groups found no significant between-group difference. A significant reduction in activity level following ACL injury is however considered a useful outcome for defining non-copers45. Future studies may thus benefit from analyses that consider coping classification of individuals with ACL injury.
In summary, our paradigm found greater BOLD response for a number of brain regions which have previously been associated with processes that can be linked to proprioception. These results thus indicate that the experimental design was successful in recruiting brain regions involved in proprioception. The lack of differences between groups is perhaps not surprising given mixed evidence of proprioceptive deficits among individuals nearly two years post-ACLR. The small sample sizes as a result of splitting the ACLR group into left and right-side injuries may however have been a contributing factor to the lack of significance. The significant correlations between knee JPS error and activation in some brain regions further indicates that demands were placed on the proprioceptive acuity of the participants. The novel integration of kinematics with fMRI thus provided added value to the paradigm by providing behavioural data as well as specific time frames for extraction of brain images to isolate such processes.
This is pioneering work that has attempted to capture brain response to a lower limb proprioception task using fMRI and simultaneous kinematics to quantify knee JPS. The identification of brain regions associated with lower limb proprioception tasks, such as knee JPS, provides new and valuable information regarding central processing of such tasks. Our unique paradigm demonstrates a method to expand these findings and provide further insights into brain response to proprioception at the knee and other joints as well as among different populations.