The protocol for conducting the review was prepared using the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols guidelines (PRISMA-P) (Shamseer et al., 2015), and has previously been peer-reviewed and published (Liang et al., 2021). Additionally, this systematic review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) (Registration number: CRD42020212128).

This systematic review included relevant articles that fulfilled the following criteria: (1) original research, (2) available in full-text, (3) written in English, (4) had clear diagnostic criteria for PFP, and (5) measured at least one of the following variables in participants with PFP: (1) spinal reflex excitability measured using H-reflex amplitudes, (2) corticospinal excitability measured using TMS-elicited MEP amplitudes, motor threshold (MT), or stimulus-response (SR) curves, (3) cortical reorganization assessed using TMS cortical mapping or structural magnetic resonance imaging (MRI), and (4) functional changes of the brain using functional MRI (fMRI). Case studies, case series, and articles focused on lower extremity injuries unrelated to PFP, were excluded.

A literature search for records published prior to December 10, 2021 was conducted using the following databases: PubMed, Medline via OVID, Embase, and Web of Science. Search strings and medical subject headings (MeSH) keywords related to the themes of PFP, and non-invasive assessments of brain and spinal cord functions were used. Search strings included (“patellofemoral pain” OR patellofemoral pain syndrome” OR “anterior knee pain” OR “PFP”) AND (“reciprocal inhibition” OR “corticospinal” OR “corticospinal excitability” OR “spinal reflex” OR “spinal excitability” OR “stretch reflex” OR “Hoffmann^* reflex” OR “H-reflex” OR “H:M ratio” OR “primary motor cortex” OR “motor evoked potential^*” OR “MEP^*” OR “motor threshold” OR “cortex reorganization” OR “transcranial magnetic stimulation” OR “TMS” OR “neural excitability” OR “mapping” OR “magnetic resonance imaging” OR “functional magnetic resonance imaging”). The asterisks indicated any potential suffix for each respective term and were input into each database.

Four investigators (SB, AM, KM, and DN) manually searched and inspected the literature published before December 10, 2021. The detailed flow chart for study selection is shown in Figure 1. Specifically, the titles of the articles from database searches were initially identified for relevance and reviewers manually screened the reference list of each title for additional articles. This was followed by screening of the abstracts of all titles yielded. If abstracts were not published in English, indicated the record was not an original research article, did not have full-text, or did not include H-reflex, MEP amplitudes, MT, SR curves, MRI, fMRI, cortical reorganization or mapping as a primary outcome variable, these records were removed from the study. The full-texts of records passing the abstract screening were then read in their entirety to determine their eligibility using the inclusion and exclusion criteria. Records that lacked an abstract or had an abstract with insufficient information were also read in their entirety and assessed using the inclusion and exclusion criteria. Methods for conducting this systematic review were developed using the Guidelines for Meta-Analysis and Systematic Reviews of Observational Studies (Stroup et al., 2000).

Quality of eligible studies were analyzed utilizing the Quality Assessment Tool for Observational Cohort and Cross Sectional Studies from the National Institutes of Health (NIH¹) by four investigators (SB, AM, KM, and DN). The assessment tool included 14 questions, and responses were either yes, no, not applicable (NA), or not reported (NR). A score of 0 was given for “no” and 1 for “yes.” The methodological quality for each study was then categorized subjectively by the assessors as good, fair, or poor, in reference to previous literature (Alfuraydan et al., 2020), as the NIH Quality Assessment Tool for Observational Cohort and Cross-sectional Studies does not have cut-off ranges for scoring. The investigators independently scored each study and reconvened at a later date and compared results. Any discrepancies in the quality rating between investigators were discussed and adjusted accordingly.

Risks of bias were conducted using the Cochrane risk-of-bias tool (Higgins et al., 2011) by two investigators (K-YH and JL). We assessed 8 items, including (1) Random sequence generation (selection bias), (2) Allocation concealment (selection bias), (3) Blinding of participants and personnel (performance bias), (4) Blinding of outcome assessment (detection bias): self-reported outcomes, (5) Blinding of outcome assessment (detection bias): objective measures, (6) Incomplete outcome data (attrition bias), (7) Selective reporting (reporting bias), and (8) No asymptomatic control group (other bias). The response options are low risk of bias, unclear risk of bias, high risk of bias, and NA. If the item was NR it was graded as unclear risk, and if the item was NA it was answered with NA. The investigators independently scored each study and reconvened at a later date to compare results. Any discrepancies in the risk-of-bias rating between investigators were discussed and adjusted accordingly.

Quality of evidence was assessed by two investigators (K-YH and JL) using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach (Higgins and Green, 2011). We evaluated three clusters of literature, including spinal reflex excitability, corticospinal excitability, and cortical reorganization. The investigators independently scored each category and reconvened at a later date to compare results. Any discrepancies in the evaluations between investigators were discussed and adjusted accordingly. If a category contained only one study that did not allow between-study comparisons, the analysis of evidence quality was not performed (de Oliveira Silva et al., 2020).

Available data were synthesized via meta-analysis using Review Manager (RevMan5, The Cochrane Collaboration). As the effect sizes of the eligible studies were assumed to represent a random sample from a particular distribution of effect sizes, a random-effects model was used (Borenstein et al., 2010; Pigott and Polanin, 2019). P < 0.05 was considered statistically significant.

A total of 119 records were identified through database search. After duplicates were removed, the abstracts of 30 studies were screened to determine further evaluation. Of the 30 abstracts screened, 17 new records were identified from the reference lists and the abstracts of these records were screened with the same criteria. All of the 47 records had abstracts available for the abstract screening process. Out of the total 47 records in the abstract screen process, 39 were disqualified as the articles were not published in English (n = 1), were not an original research article (n=8), did not have full-text (n = 1), or did not use H-reflex, MEP amplitudes, MT, SR curves, MRI, fMRI, cortical reorganization or mapping as a primary outcome variable (n = 32). Three articles were disqualified for more than one reason. Eight studies met the eligibility criteria and of those, 5 reported H-reflex, 2 reported MEP amplitude and MT, and 1 reported cortical reorganization. None of the qualified records reported MRI or fMRI. Figure 1 shows the PRISMA flowchart for literature search and selection. The participant information for each study is summarized in Table 1.

The total quality assessment scores of the articles ranged from 6 to 8 (Table 2). The maximum possible quality assessment score was 14. Based on the details reported by the articles and the total scores calculated, it was determined that the study quality was similar (fair) across the studies (Alfuraydan et al., 2020). The analysis of risks of bias of the articles revealed two primary sources of bias, including blinding of participants and personnel and blinding of outcome assessment: objective measures (25% high risk and 62.5% unclear risk). Among the 8 included studies, only one study clearly stated that blinding procedure was used in their work. Given that the studies included in this review were all cross-sectional, observational studies, the first 2 items regarding participant randomization and allocation were rated as NA (Figure 2).

Five articles with an emphasis in neurophysiological changes at the level of the spinal cord were examined with a collective sample size of 141 participants (90 PFP and 51 asymptomatic). Studies examined the H-reflex amplitudes of the VM, VL, and RF muscles, in participants with chronic PFP. An earlier study conducted in 1995 examined knee pain in participants with PFD, and reported no change in VM, VL, and RF H-reflex amplitudes before and after cold application to the knee for pain relief, suggesting that pain did not have an effect on quadriceps H-reflex amplitudes in participants with knee pain from PFD (Leroux et al., 1995).

Our meta-analysis of the data extracted from 2 more recent studies (de Oliveira Silva et al., 2016; Pazzinatto et al., 2019) revealed reduced VM H-reflex amplitudes in participants with PFP compared to asymptomatic participants. It was suggested that reduced spinal reflex excitability may be a reason underlying increased pain and decreased function and muscle strength among participants with PFP. Specifically, a correlation was reported between the VM H-reflex amplitude and pain, chronicity, and function in participants with PFP, where participants with greater VM H-reflex amplitude have less pain, more recent symptoms and better physical function (de Oliveira Silva et al., 2017). The recent study conducted by Waiteman et al. (2022) further demonstrated that VM H-reflex amplitude is negatively associated with higher variability of submaximal torque production and positively associated with maximal isometric torque production of the quadriceps muscles in participants with PFP. Their findings highlight the relationship between altered H-reflex excitability and impaired modulation of motor unit firing rates and motor unit recruitment in participants with PFP.

In addition, VM H_max/M_max ratio was found to have high sensitivity and specificity for discriminating between females with and without PFP (de Oliveira Silva et al., 2016). It was also reported that patellar tendon reflex correlated strongly with VM H-reflex in participants with and without PFP (Pazzinatto et al., 2019), suggesting that testing the patellar tendon reflex in a clinical setting may be beneficial for identifying a potential contributing factor to PFP (Pazzinatto et al., 2019).

Three articles with an emphasis in brain neurophysiological and structural changes were examined with a collective sample size of 66 participants (24 participants with PFP, 10 with AKP, 32 asymptomatic; On et al., 2004; Rio et al., 2016; Te et al., 2017).

The quality assessment tool and risk-of-bias assessment tool used were beneficial for gaining an overall sense of consistency, or lack thereof, of the studies combined. The quality for the studies evaluated was considered fair, with scores between 6 and 8. The risk-of-bias analysis also revealed the key bias of the current studies in this area, such as blinding of participants and personnel, blinding of outcome measures, and no asymptomatic control group. These assessment tools can be used for future research to qualitatively address the potential risks of selection bias, detection bias, or confounding variables. For example, to optimize the study quality and reduce bias, future studies can improve by specifying inclusion and exclusion criteria for all participants, conducting a sample size analysis, implementing a blinding procedure, using asymptomatic control groups, and stating clear diagnostic criteria of PFP.

A recent systemic review highlighted the importance of patient education in the management of PFP, including understanding the anatomy of the joint, physiology of pain, activity modification, and quadriceps muscle strengthening (de Oliveira Silva et al., 2020). Our systemic review findings regarding central nervous system changes in participants with PFP highlight the need for including neurophysiological dysfunctions of the quadriceps muscles in patient education of this cohort. Furthermore, a potential intervention is to alleviate central inhibition of the affected muscles via transcranial direct current stimulation (tDCS), which is a non-invasive brain stimulation that modulates cortical excitability (Lefaucheur et al., 2017). It has been shown that a 4-week (12-session) tDCS protocol focusing on corticomotor excitation of the knee extensor musculature improved quadriceps muscle strength and symptoms in participants with PFP (Rodrigues et al., 2020).

In addition, Glaviano et al. (2019) have theorized that central inhibition of gluteal musculature is a possible mechanism of hip muscle dysfunction and faulty movement patterns in individuals with PFP, which may lead to prolonged dysfunction and pain. However, this theory has not yet been tested in individuals with PFP. Overall, most available literature examined the spinal level and cortical level separately, thus evidence for contributions and/or adaptations from spinal or descending supraspinal sources remain limited. As the existing studies examined the spinal and cortical changes utilized a cross-sectional study design, it remains unclear whether such changes are the risk factors of PFP or the consequents of chronic pain (Pazzinatto et al., 2017). Taken together, immediate future efforts should examine cortical and spinal levels concurrently to elucidate the relative contributions of supraspinal vs. spinal influences contributing to altered knee and hip muscle function. Ideally, a larger scale, longitudinal study that tracks the central nervous system adaptations in young adults until the appearance of PFP may facilitate the understanding the underlying neurophysiological mechanism of PFP.