Failed back surgery syndrome (FBSS) is a chronic neuropathic condition characterized by chronic leg or lower back pain (cLBP) following one or more lumbar spinal surgeries (Thomson, 2013). Literature has reported the incidence of FBSS in up to 40% of the cases (Alizadeh and Sharifzadeh, 2021). Disabling neuropathic pain usually persists or worsens with time and dramatically affects quality of life (Mondello et al., 2018). When other therapies have failed, treatment paradigms resort to spinal cord stimulation (SCS), which provides electric therapy through implanted electrodes for pain relief (Alizadeh and Sharifzadeh, 2021; Hosseini et al., 2021). In standard clinical practice, a subjective trial phase is carried out to determine patients who will benefit from SCS therapy before the permanent implantation of the device (North et al., 2020; De Andres et al., 2021). In the literature, successful trials varies from 40 to 80%, which is influenced by patient’s age, SCS program settings, pain localization and patient’s history of spine surgery (Odonkor et al., 2020). In addition, a third of patients who had failed SCS trial may still benefit from permanent implant in a 6–18 months follow-up (Oakley et al., 2008), yet patients with successful trial may show decreased efficacy in a 4 years follow-up (Teton et al., 2019). Therefore, despite the clinical efficacy of SCS, its mechanisms of pain relief and patient selections are still not completely understood. Both spinal segmental and supraspinal pathways are complementary systems in the effectiveness of SCS (El-Khoury et al., 2002; Barchini et al., 2012; De Groote et al., 2020b).

Functional magnetic resonance imaging (fMRI) studies in patients suffering from cLBP have shown alterations of brain activity related to pain perception and emotion (Hashmi et al., 2013; Sun et al., 2021). Resting state fMRI (rsfMRI) is a powerful tool for mapping the intrinsic brain functional connectivity (FC). Previous studies have observed widespread changes in cLBP patients, highlighting the disrupted default mode network (DMN) along with elevated FC related to the medial prefrontal cortex, anterior insula, amygdala, and sensorimotor regions (Kregel et al., 2015; Jiang et al., 2016; Mao et al., 2022a,b). Studies that focused on the FBSS patient population demonstrated generally decreased connectivity within the DMN (Kornelsen et al., 2013) and increased connectivity associated with multiple resting state networks (RSNs), including the salience, central executive, and sensorimotor networks (Kolesar et al., 2017). Although phantom studies have demonstrated safety of SCS stimulation via externalized neurostimulators in both 1.5 and 3T MRI (Moens et al., 2012), only a few studies have looked into the effects of SCS on resting state functional connectivity in FBSS. To our knowledge, only two research groups (Moens and Rezai) have published studies on rsfMRI of FBSS patients undergone permanent SCS implantation (Deogaonkar et al., 2016; De Groote et al., 2019, 2020a). SCS systems have received FDA approval for 1.5T MRI; however, it remains off-label to scan under 3T MRI (Thornton, 2017). With careful considerations of all the parameters of FDA guidelines (SAR, B0 values, head-only transmit, and scan duration), we conducted 3T rsfMRI in a single FBSS patient, who had received permanent SCS implantation. In this case study, we demonstrate the safety of 3T MRI in a patient with a permanently implanted SCS to provide evidence for future studies.

Increasing evidence suggests that spontaneous brain network communication measured via rsfMRI is intrinsically dynamic. FC among resting state networks (RSNs) have been demonstrated to have dynamic fluctuations rather than stationary throughout the entire scan time, especially relevant to the cognitive and attentional functions (Allen et al., 2014). A sliding-window technique followed by k-means clustering has been applied for the assessment of the dynamic characteristics; the resulting dynamic states (or brain states) represent highly replicable sub-patterns of FC during resting state (Allen et al., 2011, 2014). With such progress in neuroimaging analysis, Kucyi and Davis first introduced the concept of dynamic pain connectome in the brain (Kucyi and Davis, 2015). In patients suffering from low-back-related leg pain, temporal variabilities were found increased from healthy control groups (Pei et al., 2020). Additionally, immediate changes of rsFC were observed in cLBP after a pain-inducing maneuver and a manual therapy (Yu et al., 2014; Isenburg et al., 2021). Therefore, the second aim of this case study was to evaluate our hypothesis that FBSS presents alterations in dynamic FC from healthy controls. Such findings may be helpful in identifying potential biomarkers in the FBSS patient population and/or the supraspinal effects of SCS.

A 46-year-old female, who had right L5-S1 microdiscectomy and revision microdiscectomy 4 and 5 months earlier, presented with chronic low back and leg pain. Her pain was described as constant burning sensation radiating unilaterally down the right lower extremity, with accompanying numbness and weakness in her right leg. Her pain would get worse as the day extended. At the time of the initial presentation, an 8/10 visual analog scale (VAS) was rated, and the Oswestry Disability Index (ODI) was scored 26/50 (modified score: 23). As no significant nerve root compression was noted on the patient’s lumbar MRI, she was diagnosed with FBSS. After her initial evaluation and non-operative therapy (including aquatherapy, physical therapy, chiropractor care, acupuncture, and epidural steroid injection), it was recommended that she consider SCS for her chronic pain reduction. All risks and benefits of each therapy (both non-operative and SCS) were explained to the patient. She had a successful SCS trial (8 days) and experienced approximately 70% pain relief with improvements in activity tolerance. The patient felt most pain relief while on differential target multiplexed (DTM) programming. As such, permanent SCS implantation was performed 2 weeks later. Following standard clinical care, two SCS permanent percutaneous electrodes (Medtronic, Inc., Minneapolis, MN, USA) were implanted in patient’s thoracic epidural space. One electrode was positioned at the top of T8 and the other electrode in the mid body of T9, so that they were at the anatomical midline with contacts of leads covering T8-10. A left buttock incision was made for the placement of a rechargeable implantable pulse generator (Medtronic Intellis Neurostimulator). The surgery was successful with no complications.

During her follow-up visit 2 weeks after her permanent SCS implantation and before the start of SCS treatment, the patient noted well-controlled incisional pain without the need of any additional pain medication. She reported an 8/10 VAS and was on her regular pain medications (Duloxetine, 60 mg by mouth daily; Gabapentin, 600 mg by mouth 2 times a day and 1,200 mg at bedtime; Percocet, 5–325 mg tablet, 1 tablet by mouth every 4 h as needed for up to 3 days; Naloxone, 4 mg per actuation spray, 1 spray as needed). Other clinical assessments were also collected, including the ODI, Pain Catastrophizing Scale, Hospital Anxiety Scale, Hospital Depression Scale, Pittsburgh Sleep Quality Assessment, and Central Sensitization Inventory (Table 1). rsfMRI was acquired on the same day. SAR values was monitored throughout the scan, which stayed way below the Medtronic MR guidelines and FDA regulations (3 W/kg head only). The patient was closely monitored and did not report any discomfort during and after the MRI acquisition process. She was also checked by a neurosurgeon with no complications or side effects noticed due to MRI. These measurements will serve as the baseline for future follow-up visits for monitoring the SCS therapy.

In this case study, we scanned a FBSS patient with SCS system implanted (off-stimulation) on a 3T MRI device with careful safety considerations. Being part of an ongoing prospective study, our first case demonstrated the safety and feasibility of 3T MRI scanning for patients with implanted SCS system. The rsfMRI data in this study will also serve as the baseline for intra-subjective studies about the effects of SCS therapy on supraspinal FC in FBSS. Here, we assessed the resting state network connectivity using both static and dynamic FC analyses and compared to healthy controls. The static FC pattern from the FBSS patient showed wide-spread alterations among all the RSNs. The dynamic FC analysis clustered all window-matrices, separately for each group, into four brain states. Visual evaluation of the matrix patterns revealed that that brain state 1 and 2 of both groups had relatively similar patterns compared to brain state 3 and 4. Similar fraction rates were observed in both groups: about half of window-matrices occurred as brain state 1. Regarding the mean dwell time shown by the bar graph, the FBSS case seemed to prefer a single state (State 1) over the other three states. As there’s only one case in the FBSS group, future statistical analysis with larger sample size is necessary to make any concrete conclusions.

Increasing evidence suggest the temporal fluctuations of brain FC in chronic pain and warrant the necessity to consider the dynamic characteristics of pain connectome in contemporary research (Kucyi and Davis, 2015; Pei et al., 2020). Effective treatments of cLBP have shown the ability to reverse abnormal FCs toward healthy controls (Seminowicz et al., 2011; Li et al., 2014), indicating the plasticity of the pain connectomes. Changes of rsFC were also observed in cLBP subjects after 10 min pain-inducing maneuver, which further demonstrated the dynamic characteristics of FC in chronic pain (Yu et al., 2014). In a previous dFC study of a group of healthy participants, it was demonstrated that pain perception threshold was associated with the frequency of a DMN-dominant brain state, whereas the touch threshold was associated with another sFC-like brain state (Yuan et al., 2019). Other dFC studies found that the mean dwell time in a less frequent brain state, out of 5–6 total states, was significantly higher in the chronic migraine group versus healthy controls (Zou et al., 2021) and in subjects with post-brain-traumatic headache vs. those without (Li et al., 2021). Pei et al. applied the sliding-window method (without clustering) and revealed a higher temporal variation of window-matrices in subjects with low-back-related leg pain compared to healthy controls (Pei et al., 2020). The current case study demonstrated with qualitative results the promising application of dFC in FBSS, as well as cLBP, populations. Moreover, we observed different FC patterns between the two groups in the less frequent brain state 3 and 4, indicating the potential of dFC analysis to detect more sensitive and specific differences between FBSS and HCs.

A number of studies have used rs-fMRI to assess alterations of the intrinsic brain functional connectivity in cLBP compared to healthy controls. Widespread findings have been found in the current literature. In general, cLBP patients showed decreased FC within the DMN (Kornelsen et al., 2013; Jiang et al., 2016; Ayoub et al., 2019; Ng et al., 2021a), along with increased FC between the DMN and other networks, including the emotional, cognitive, and/or central executive networks (Kornelsen et al., 2013; Jiang et al., 2016; Ma et al., 2020; Ng et al., 2021b; Mao et al., 2022b,c) in static FC studies. Particularly, medial prefrontal cortex (mPFC) seems playing an important role in cLBP (Baliki et al., 2011, 2014; Yu et al., 2014). The lack of brain activities in common noxious regions indicated that chronic pain may not be caused by nociceptive process, but rather associated with emotional and cognitive processes and potentiated through connectivity of these regions (Ng et al., 2017; Pahapill et al., 2020). In addition, enhanced rsFC in cLBP were also observed in the sensorimotor networks (Mao et al., 2022a), visual networks (Shen et al., 2019; Lamichhane et al., 2021a,b), reward network (Yu et al., 2020), and DMN-PAG couplings (Yu et al., 2014). Given the widespread findings in rsFC patterns, machine-learning, although on small sample size, showed some capabilities to differentiate cLBP patients from healthy controls (Tu et al., 2019; Lamichhane et al., 2021a,b). These reports were based on heterogeneous cLBP populations with different etiologies (neuropathic vs. nociceptive).

For homogeneous cohorts, studies that focused on the FBSS population revealed similar alterations in cortical networks (i.e., decreased FC within the DMN, and increased FC between the DMN and salience, central executive, and sensorimotor networks) compared to healthy controls (Kornelsen et al., 2013; Kolesar et al., 2017; Pahapill et al., 2020). Moreover, Pahapill et al. reported decreased FC between the striatum and other networks, decreased FC between the PAG and DMN, and increased FC between the PAG and sensorimotor networks (Pahapill et al., 2020). There is shortage of rs-fMRI studies that specifically exams homogeneous groups, hence it remains unclear whether neuropathic chronic pain express differentiable connectivity patterns from nociceptive chronic pain patients. However, a recent EEG study measured the brain responses to nociceptive stimulation and revealed significant differences in coupling strengths between patients with nociceptive and neuropathic etiologies. Their findings provided positive evidence for finding potential neuroimaging biomarkers of cLBP subtypes (Goudman et al., 2020). We believe that dynamic FC has the potential to provide new biomarkers in future studies in FBSS, better patient selections for SCS treatments, and/or the mechanisms of SCS.

There are several limitations that should be considered in this case study. First, only a single case was analyzed and compared with healthy controls, that was not age-matched to the case. Although trends were observed between the FBSS case and healthy controls, statistical analysis were lacking due to the sample size. Second, the ROIs were extracted based on a pre-determined atlas including 8 RSNs in the cerebral area, without subcortical regions. Recent studies that explored the effects of SCS for FBSS patients have shown FC alterations in the descending pain modulatory system (DPMS), which includes the periaqueductal gray (PAG) and the rostral ventromedial medulla (RVM) in the brainstem (De Groote et al., 2019). Our study would likely not capture those effects given that active stimulation therapy had not occurred in the implant yet. Only changes related to the 8-day trial therapy may have been captured, if contributory. Nonetheless, this study demonstrated that the quality of MR images was comparable with those acquired from patients without SCS implants, which did not limit the study of subcortical and brainstem regions. The implanted electrode and pulse generator were located with sufficient distance from the brain to not introduce artifact in the imaging region. Therefore, we are planning to conduct a larger study in the future.

This case study demonstrated that, with prior safety considerations, patients with SCS implantation (during off-stimulation) may safely undergo 3T MRI. The preliminary study of a FBSS patient provided support for our ongoing prospective study, which aims to explore the effects of SCS on brain functional connectivity. Our results also demonstrated that dynamic FC in fMRI studies in chronic pain provides more granularity with FC changes associated with different brain states. Further studies with larger sample sizes and inclusions of subcortical brain regions are warranted to better understand the brain functional connectomes in subtypes of chronic pain, as well as the mechanisms of SCS therapy. In addition, future studies should compare changes in FC over time in patient undergoing SCS therapy.

The datasets presented in this article are not readily available because institutional regulations require the establishment of formal data sharing agreements before patient identifying information included in this dataset or metadata can be shared. Deidentified raw data can be provided upon request to the corresponding author, after approval from the institutional review board. Requests to access the datasets should be directed to JM, jingya.miao@jefferson.edu.

The studies involving human participants were reviewed and approved by the Thomas Jefferson University Institutional Review Boards. The patients/participants provided their written informed consent to participate in this study.

JM and MA contributed to the concept and design of the study. IA, LK, KF, DM, KT, and PN helped in the acquisition and verification of data. JM, IA, FM, KH, CM, and MA interpreted the data. JM drafted the manuscript and has full access to all the data in the study and responsible for the integrity of the data and the accuracy of the data analysis. FM, KH, CM, and MA critically revised the manuscript for important intellectual content. MA supervised the study. All authors contributed to the article and approved the submitted version.