Interactive interdisciplinary pain research in adolescent and young adult females: a pilot investigation of brain, physiological, and emotional functioning following orthopedic surgery

Abstract

Objective

In this pilot investigation, we aimed to explore the neurological and biobehavioral mechanisms underlying pain outcomes in adolescent and young adult (AYA) females following orthopedic surgery, an area largely unexplored.

Methods

Functional near-infrared spectroscopy was used to investigate brain responses in the primary sensory cortex (sensory pain processing) and the prefrontal regions (emotional processing) in 24 AYA females who underwent orthopedic surgery within the previous 2 years compared to 20 group-matched controls without a surgical or chronic pain history. A battery of self-reported pain-related and emotional functioning measures (PROMIS; pain catastrophizing) were also administered. Cortical activations and functional connectivity (FC), involving the prefrontal (PFC) and somatosensory cortices (SMC), were assessed during resting state and a descending pain modulation task (conditioned pain modulation).

Results

In the control group, PFC-SMC FC in response to pain was significantly linked to anxiety, whereas this correlation was absent in the post-surgical cohort.

Conclusion

These results highlight distinct altered responses in sensory and emotional brain functioning in AYA females following orthopedic surgery. We suggest that such changes may be related to the involvement of the PFC-SMC communication in the maintenance of chronic pain.

Introduction

Nearly 5% of children and adolescents and 10% of young adults between the ages of 18–34 undergo surgery each year in the United States (Bicket et al., 2024; Rabbitts & Groenewald, 2020). One problematic outcome of surgery is the prevalence chronic post-surgical pain (CPSP), defined as pain persisting beyond 2 months after surgery, which affects approximately 20%–40% of children, adolescents, and young adults globally (Einhorn et al., 2024). CPSP can be influenced by various risk factors, particularly those related to brain function, such as through central sensitization or pain inhibitory processing at the cortical and subcortical levels. Additionally, these brain functions are significantly altered by psychological states (e.g., anxiety). As such, the result of uncontrolled or exacerbated pain can significantly diminish quality of life and, importantly, have deleterious outcomes later in life (Sieberg et al., 2022). Specifically, CPSP often leads to anxiety, depression, and other psychological disorders, which exacerbate pain perception within the central nervous system (CNS) and hinder recovery (Sieberg et al., 2022). Additionally, pain-related fear and pain catastrophizing (i.e., pain-related worry) can amplify the pain experience, creating a vicious cycle that is difficult to break (Chow et al., 2020). Given the variability in individual responses to pain and psychological interventions, CPSP management is complex and particularly challenging for clinicians (Thapa & Euasobhon, 2018). A tailored, personalized approach to treatment is imperative in addressing these multifaceted challenges effectively; however, in order to devise effective treatments, more integrative mechanistic research is warranted. Sieberg and colleagues proposed the need for increased rigor in research in pediatric CPSP and posited a model focused on the peri-surgical process (Sieberg et al., 2022). Specifically, this model involves ongoing and continuous evaluation and treatment of pre-mitigating factors to premorbid status, injury, and immediate post-injury treatments (including peri-surgical processes), objective assessment of pain chronification, and treatment rehabilitative processes.

Studies of CPSP in young people have primarily focused on orthopedic surgeries given the prevalence of these procedures during this stage of development (Batoz et al., 2016; Sieberg et al., 2023). For example, one study found that 13% of pediatric patients developed CPSP after orthopedic procedures, with 63% of these cases developing neuropathic pain (Batoz et al., 2016). Compared with all other surgical procedures, orthopedic surgery significantly increases the risk of moderate to severe CPSP (Thapa & Euasobhon, 2018). Overall, females are disproportionately affected by CPSP, with some evidence indicating a higher incidence of preoperative chronic pain, a greater likelihood of transitioning to CPSP, and more significant disruptions to sleep and daily activities compared to males (Kanaan et al., 2021). However, findings are mixed. For example, studies consistently indicate that females report higher pain sensitivity and intensity, often perceiving pain at lower thresholds (Casale et al., 2021). When considering pain treatment responses, some research has reported that females may experience less relief from opioid medications and greater susceptibility to side effects, which can lead to under-treatment (Angst et al., 2012), while other research has demonstrated no significant sex differences or even better outcomes in females for certain types of pain highlighting the complexity of the issue (Köckerling et al., 2020). These variations suggest that the type of surgery, the pain management approach, and individual patient factors may all contribute to outcomes. There is a poor understanding of why so many females experience CPSP, the mechanisms contributing to the maintenance and exacerbation of CPSP, and thus dismal treatment options, especially for young people. There is a need for interdisciplinary collaborations and multimodal investigations in order to better elucidate the complex biopsychosocial factors contributing to CPSP in young females.

Neuroimaging techniques, such as functional near-infrared spectroscopy (fNIRS) and functional magnetic resonance imaging (fMRI), offer valuable insights into the neural mechanisms underlying chronic pain and pain inhibitory processing. An increasing number of neuroimaging studies have enabled the identification of brain-based markers of CPSP (Peng et al., 2021). Using multimodal MRI, Ching et al. (2018) reported significantly increased resting-state functional connectivity (FC) in regions such as the insula, right angular gyrus, and middle occipital gyrus after pain stimulation in healthy adults, whereas changes were absent in people with chronic post-hysterectomy pain (Ching et al., 2018). However, research on young people remains limited, highlighting a critical gap in the literature. fNIRS is a non-invasive, portable method that uses near-infrared light to measure blood oxygenation changes in cortical areas (Ferrari & Quaresima, 2012). Its flexibility allows for research in clinical settings during pain-related tasks. With good temporal resolution, fNIRS effectively captures brain activity linked to pain perception and processing. Our team has pioneered the use of fNIRS in pain research, successfully identifying atypical hemodynamic responses in the prefrontal cortex (PFC) and the primary somatosensory cortex (SMC) in adolescent and young adults (AYA) undergoing knee arthroscopy (Karunakaran et al., 2023). These findings demonstrate the utility of fNIRS for monitoring pain and assessing patient risk for chronic pain.

Moreover, presurgical pain and presurgical psychological state, including catastrophizing and fear of pain, may disrupt homeostasis and contribute to allostatic load following surgery, which in turn results in significant risk factors for the development of CPSP (Karunakaran et al., 2023; Wiech & Tracey, 2009). For example, living with pain can result in negative affect, anxiety, and depression and there is also evidence that the reverse is true; negative mood and emotion can exacerbate pain such that the perception of pain is not necessarily linearly related to noxious input but is also strongly influence by psychological functioning (Wiech & Tracey, 2009). This bi-directional relationship can be explained by shared neural pathways in the brain that are responsible for both pain modulation and affective processing, including the periaqueductal gray (PAG), anterior cingulate cortex, and the anterior insula all of which impact pain and affective processing (Szabo et al., 2022; Wiech & Tracey, 2009).

This pilot study explored neurological and biobehavioral mechanisms underlying pain outcomes in AYA females following orthopedic surgery, with two main aims. First, we examined pain-evoked brain responses in AYA females who had orthopedic surgery within the past 2 years, compared with nonsurgical, pain-free controls. Pain responses were assessed using conditioned pain modulation (CPM), a validated method for evaluating inhibitory pain systems (Ramaswamy & Wodehouse, 2021). CPM is commonly used as reduced CPM efficiency is often seen in individuals with or at risk for chronic pain, indicating impaired CNS pain inhibition (Ramaswamy & Wodehouse, 2021). We hypothesized that the surgery group would show reduced FC and altered activation patterns during the CPM task compared to controls. Second, we aimed to assess the associations between these brain functional metrics and self-reported measures of pain and emotional functioning. Specifically, we hypothesized that neural alterations observed during the CPM task would correlate with self-reported measures, reflecting the potential long-term impact of orthopedic surgery on pain processing and emotional regulation. Given the limited research in young populations and the variability in previous findings, this hypothesis is exploratory and aims to generate preliminary evidence on the neural and psychosocial factors contributing to pain outcomes in AYA females.

Methods

Participants

Twenty-four AYA females who underwent orthopedic surgery within the previous 24 months and twenty group-matched, nonsurgical, pain-free controls completed this study. Patients who had undergone orthopedic surgery during the previous 2 years were recruited from advertisements in the Division of Sports Medicine at Boston Children’s Hospital and through flyers advertised in the hospital and local community. Controls were also recruited from advertisements at the hospital and the local community. Of note, this pilot study is part of a larger ongoing investigation deeply phenotyping CPSP risk across the lifespan (adolescents–older adults). Within the larger study 236 participants were screened eligible, while 42 participants were screened ineligible due to not meeting inclusionary criteria. For the present cohort pilot study, no one declined participation who was eligible. Further, no participants dropped out during the study session.

For surgical participants inclusion criteria consisted of having orthopedic surgery during the previous 2 years, being between the ages of 12 and 34 years old (consistent with other studies exploring pain in AYA females [Murray et al., 2022; Sieberg et al., 2013]), and on stable dosages of all medications during the past 30 days. All participants in the surgical and healthy control groups were biologically female and identified as female; however, the latter was not a criterion for inclusion. Participants were excluded if they reported a history of significant psychiatric or neurological disorders such as psychosis, schizophrenia, and/or bipolar disorder. Exclusionary criteria also included recent or current illicit recreational drug use and use of antipsychotic medication, and recent (within 12 months) traumatic brain injury. Additionally, we adopted a process that limits plasma levels of over-the-counter medication (OTC) that are taken “pro re nata” (PRN; unscheduled) at the time of scanning. Specifically, participants could not have taken a non-narcotic pain reliever within 48 hr prior to a study visit. If a participants had to take such a PRN medication within 48 hr, then their study visit was rescheduled. This is important as OTC analgesics, including nonsteroidal anti-inflammatory drugs and acetaminophen are known to alter brain responses (Ratner et al., 2018). Of note, none of the surgical cohort endorsed using opioid analgesics to manage their pain leading up to the study visit.

The same exclusion criteria were applied to the control group, and in addition, they could not have a history of chronic or recurrent pain, any surgeries requiring general anesthesia in their lifetime, and no history of chronic illnesses (e.g., cancer, heart disease, diabetes, or osteoporosis). Additionally, when assessing for transient pain (i.e., headache, muscle pull, menstrual cramps), controls could not have a current pain rating of greater than 3/10 on the day of the study visit. The study received Institutional Review Board approval at Boston Children’s Hospital. All potential participants were screened over email or the phone for eligibility. If eligible, participants provided written assent and/or consent (with parent/guardian consent for participants under age 18) and attended one approximately 3-hr study visit.

Self-report measures

Clinical information

Participants completed a battery of self-report questionnaires, including demographic (e.g., age, race, ethnicity) and medical history (e.g., surgery information, presence or absence of pain, neurological, psychiatric disorders). Medication use was assessed using the Medication Quantification Scale (Gallizzi et al., 2008) and any pain experienced on the day of the study visit was rated on a 0–10/10 numerical rating scale (NRS).

Pain-Related Functioning

Participants ages 18 and older completed the Pain Catastrophizing Scale (PCS), (Sullivan et al., 1995), while adolescents completed the child version of this measure (Crombez et al., 2003). Participants responded to 13 statements describing thoughts and feelings associated with pain. Total scores range from 0 to 52 for both the adult and pediatric questionnaires, with greater scores reflecting greater catastrophizing (i.e., pain-related worry). For adults, scores below 15 represent low catastrophizing, scores between 16 and 23 represent moderate catastrophizing, and scores above 24 represent high catastrophizing (Sullivan et al., 1995). For pediatric, scores below 15 represent low catastrophizing, 15–25 represent moderate catastrophizing, and scores above 26 indicate high catastrophizing (Pielech et al., 2014). High internal consistency of the PCS was present in this study, confirmed by Cronbach’s alpha score $\alpha =0.93.$

Pain interference was measured using the Pain Interference Patient-Reported Outcomes Measurement Information System (PROMIS) Questionnaire (Amtmann et al., 2010), which measures if pain hinders engagement with social, recreational, emotional, physical, and cognitive activities. Pediatric participants completed the pediatric version of the PROMIS measure. The scores are standardized to a mean = 50, SD = 10 in a general United States population (i.e., T-scores, https://www.healthmeasures.net/score-and-interpret/interpret-scores/promis/promis-score-cut-points). Higher scores reflect greater interference of pain and worse engagement with social, recreational, and other daily activities. This measure was also internal consistent within the study ($\alpha =\mathrm{\hspace{0.17em}}$0.92).

Emotional functioning

Participants completed the Depression and Anxiety PROMIS measures (Schalet et al., 2016). Both include eight questions to assess depressive and anxious mood, respectively. For both questionnaires, participants were asked to rate how frequently they experienced specific symptoms over the preceding 7 days. The total score assesses both the frequency and range of experienced symptoms over the past 7 days and uses T-scores, as described above, with higher scores indicating greater anxiety or depressive symptoms. Standardized scores are categorized according to the Health Measures score cut-off points: Within normal limits”, T-score $\le$55; “mild,” 55 $<$ T-score $\le$60; “moderate,” 60 < T-score $\ge$70; “severe,” 70 < T-score $\ge$ 80. For the pediatric surveys, the PROMIS categorization are as follows: “Within normal limits,” T-score $\le$ 50; “mild,” 50 $<$ T-score $\le$ 55; “moderate,” 55 < T-score $\ge$ 65; “severe,” T-score > 65 (https://www.healthmeasures.net/score-and-interpret/interpret-scores/promis/promis-score-cut-points [Amtmann et al., 2010]). Both PROMIS depressive symptoms and anxiety symptoms were internally consistent within the population ($\alpha =\mathrm{ }$0.95 and $\alpha =0.91$, respectively).

Participants also completed the NIH Toolbox Fear and Somatic Arousal measure (https://www.nihtoolbox.org/test/fear-somatic-arousal/) to assess autonomic arousal and perception of threat. The adult version comprises six items and participants are asked to rate the frequency of experiencing certain somatic symptoms (e.g., short of breath; tense muscles). Answers are on a 5-point scale ranging from “Not at all” to “Extremely.” Pediatric participants completed the 8-item NIH Toolbox Fear measure to assess somatic arousal and fear. Answers are also on a 5-point scale and range from “Never” to “Almost Always.” For both surveys, the scores again are standardized to a T-score to a mean = 50, SD = 10 in a general US population (NIH Toolbox Scoring and Interpretation Guide, https://www.epicrehab.com/epic/documents/crc/crc-201307-nih-toolbox-scoring-and-interpretation-manual%209-27-12.pdf). Higher scores suggest higher somatic arousal and fear. The Fear and Somatic around has strong internal consistency within the study population ($\alpha =\mathrm{ }$0.88).

Reliability and validity in self-report measures

The five self-repot measures in this analysis come from PROMIS, the NIH-Toolbox, and the PCS questionnaire. PROMIS tools for depression, anxiety, and pain interference have been thoroughly vetted for clinical practice and show strong internal consistency, validity, and reliability (DiRenzo et al., 2023; Quach et al., 2016). The NIH toolbox survey for Fear and Somatic arousal has been a reputable self-report measure in clinical practice for anxiety-related symptoms for a significant amount of time, demonstrating strong contrast validity and strong test–retest reliability (Carlozzi et al., 2017). The PCS is a widely established tool for assessing catastrophic thinking about pain across three levels: rumination, magnification, and helplessness. Research has repeatedly reported the PCS having strong validity as well as test–retest reliability (The Pain Catastrophizing Scale (PCS)—Pain-Related Fear, n.d.).

Experimental setup and fNIRS system

As shown in Figure 1A, 5 sources and 8 detectors (3 cm apart) were placed over the bilateral PFC, and four sources with 6 detectors over the bilateral SMC., yielding 12 long-seperation channels. Nine short-separation channels (1 cm apart) were placed over key regions of interest (ROIs) to capture superficial hemodynamic signals (Wyser et al., 2020). A total of six ROIs were selected, including bilateral lateral prefrontal cortices (lPFC), bilateral medial PFC (mPFC), and bilateral SMC. The selection of ROIs is based on established neuroscience literature highlighting dysfunction in cortical networks involved in pain processing and modulation. Specifically, the PFC is linked to higher-order cognitive processes related to pain, while the SMC handles sensory-discriminative aspects such as pain localization and intensity (Kolk & Rakic, 2022; Ong et al., 2019; Xu et al., 2019).

Figure 1.

(A) The experimental setup consists of two images, one with the participant during resting state where the Model CW7 and CW7 CE Systems processing unit, real-time data capture software (on leftmost computer) and the headcap holding sources and detectors (as seen on participant) are displayed. Before fNIRS data collection, participants were seated comfortably with a laptop at a fixed distance on an adjustable table. The fNIRS cap was secured with an adjustable chin strap, and participants received standardized task instructions. fNIRS data were collected using a multichannel continuous wave system (CW7, TechEn Inc., Milford, USA) at 25 Hz with two laser wavelengths (690 nm and 830 nm). A standard headcap was utilized to position the source and detector optodes. The fNIRS cap SD arrangement diagram shows the specific placement of the sources and detectors on the headcap to capture data from the six regions of interest (see footnote for abbreviations). (B) The Test Stimulus timeline depicts the steps taken to conducted to collect test stimulus data. Pressure is delivered to the non-dominant trapezius using an algometer until the participant indicated pain, at that point stimulus application is stopped, the pressure reached is recorded and the participant rates their pain on a scale from 0 to 10. After this short break, this cycle is repeated twice more and pressure and pain scores a record. The conditioned stimulus timeline depicted shows the participant submerging their hand in 4 °C cold water for 15 s, then receiving a pressure stimulus through the algometer until their indicate pain, they will then rate their pain on a scale from 0 to 10 before pulling their hand out the water. This was repeated once more. Note. L = left; mPFC = medial prefrontal cortex; lPFC = lateral prefrontal cortex; R = right; SD = source-detector; SMC = somatosensory cortex).

Functional near-infrared spectroscopy imaging data acquisition

To establish baseline brain responses, each participant underwent a resting state assessment lasting 10 min. During this period, participants were instructed to focus on a black cross displayed centrally on a laptop screen. The resting state session commenced following these instructions, with the black cross remaining on the screen for the entire 10-min duration.

The CPM paradigm employed in this study is based on a modified protocol from previous research (Martel et al., 2019). In brief, CPM is characterized by the reduction in perceived pain elicited by a test stimulus (TS) when accompanied by a conditioning stimulus (CS) applied contralaterally. In the design of this study, pressure pain applied to the trapezius of the participant’s non-dominant hand served as the TS, while immersion of the hand in cold water (4 °C, ±0.1 °C) acted as the CS (Figure 1B).

During the CPM TS period, pressure pain was administered to the trapezius muscle using a pressure algometer, and this procedure was repeated three times. In the TS+CS period, following the initial pressure pain stimulus to the trapezius, participants underwent a brief cold pressor task by immersing their hand in a circulating cold-water bath at approximately 4 °C (± 0.1 °C) for 15 s. Immediately after each immersion, while the hand remained in the cold water, the pressure stimulus was reapplied to the trapezius. Participants were instructed to maintain hand immersion for the duration of the procedure, with the option to withdraw if the sensation became intolerable. The neural response was quantified by comparing the magnitude of brain activation during the pressure stimulus in the CS condition and during the combined TS+CS condition. FC between ROIs was further analyzed based on the temporal patterns of brain activation in both conditions (TS and TS+CS). In total, 21 quantitative neuroimaging measures were generated per participant, comprising 6 regional brain activation magnitudes and 15 pair-wise FCs.

fNIRS pre-processing

The fNIRS data were preprocessed using the open-source Matlab toolbox (The MathWorks Inc., MATLAB version R2019b, Natick, MA, USA) and a customized script. Initial quality checks were performed using qt-nirs (https://github.com/lpollonini/qt-nirs) and the Homer2 toolbox (Huppert et al., 2009). A scalp-coupling index (Pollonini et al., 2014) was calculated for each channel, with a value of 0.5 indicating acceptable signal quality. Time-series data from the 10-min resting period were extracted for channels meeting quality criteria. Raw light intensity signals were converted to optical density, and temporal filtering was applied using a bandpass filter (0.01–0.1 Hz) to remove baseline fluctuations and physiological noise (e.g., cardiac (1–2 Hz), respiratory (0.2–0.4 Hz)) (Tong et al., 2012). Optical density values were then converted to hemoglobin concentration values using the modified Beer–Lambert law (Baker et al., 2014). Motion artifacts were detected and corrected using a spline algorithm (Jahani et al., 2018).

During the 10-min resting period and the CPM task involving TS and TS+CS, activation in the six ROIs was assessed by averaging changes in oxyhemoglobin concentrations—reflecting oxygen levels in brain regions—across all source–detector pairs within each hemisphere. Pairwise FC between ROIs was calculated using the Pearson correlation coefficient (Spearman, 1987), based on average changes in hemoglobin concentrations over the time series.

Statistical analyses

Self-report pain and emotional functioning measures

For self-report pain and emotional functioning measures, group differences were calculated using independent T-tests for parametric values (Kim, 2015) and Mann–Whitney U tests for non-parametric values (Mann & Whitney, 1947).

Neuroimaging measures

Paired-sample t-tests were initially conducted to compare the post-surgical group to the controls on brain functional patterns between the CPM TS phase and the TS+CS phase. For brain imaging measures that exhibited altered patterns between the two phases, Pearson’s correlation analyses (Spearman, 1987) were performed to examine the associations between these altered functional brain imaging measures and pain-related functioning variables. Furthermore, mean values of the brain imaging measures that showed significant correlations with self-report measures were compared to the magnitude of resting-state neural responses to determine whether activation was increased or decreased, and whether changes in FC were hyperactive or hypoactive relative to baseline. All statistical analyses were conducted using the Statistical Package for the Social Sciences (IBM SPSS Statistics, Version 27). Multiple comparisons were corrected using the false discovery rate at α = 0.05 (Genovese et al., 2002), with a significance threshold of p < .05 applied to all tests.

Results

Participant characteristics

The total sample included 24 post orthopedic surgery participants (14–31 years old, M age = 23.67 years, SD = 4.13) and 20 age-matched and handedness-matched nonsurgical, pain-free controls (16–28 years old, M age = 22.40 years, SD = 3.15). Demographic and clinical data are included in Table 1. There were no significant differences in age, race, ethnicity, and handedness between the surgical and nonsurgical groups. Surgical patients had a mean of 3.58 ± 3.76 surgeries in their lifetime, a number that reflects all lifetime surgeries, including non-orthopedic procedures. Of note, nine participants in the control group reported small procedures such as tooth extraction and skin biopsies; however, none of the procedures required general anesthesia and all procedures were at more than two years prior to the study visit with no resultant pain or complications.

Table 1.

Demographics and clinical characteristics of female orthopedic surgery patients and pain-free controls.

		Surgical patients (n = 24)	Controls (n = 20)	Test statistic
Demographics
Age, years		23.67 ± 4.13	22.40 ± 3.15	t = 1.13, p = .267
		(14–31)	(16–28)
Ethnicity, n (%)				χ 2 (1, N = 44) = 0.058, p = .810
	Hispanic/Latino	3 (13)	3 (15)
	Not Hispanic/Latino	21 (87)	17 (85)
Handedness, n (%)	Left	22 (92)	18 (90)	χ 2 (1, N = 44) = 0.037, p = .848
	Right	2 (8)	2 (10)
Race, n (%)				χ 2 (4, N = 44) = 10.73, p = .03
				${\mathit{\chi }}_{\mathit{williams’correction}}^2$ = 10.49
	White	20 (84)	12 (60)
	Black/African American	0 (0)	3 (15)
	South Asian/Asian	1 (4)	3 (15)
	Hispanic/Latino	0 (0)	2 (10)
	Multiple	3(12)	0 (0)
Clinical variables
Lifetime surgeries		3.58 ± 3.76	0 ± 0.00	–
		(1–19)	–

Note. Data are expressed as mean SD or number of patients (%). Continuous data were analyzed using an independent sample t-test for parametric values.

Surgical characteristics

Orthopedic surgical procedures within the last two years are described in Table 2, within the majority of participants undergoing either knee or hip surgeries (75%). The mean time since the latest orthopedic surgery ranged from >1 month to 21 months prior to the study visit, with a mean of 11.17 ± 5.51 months since undergoing an orthopedic surgery. Two participants had <1 orthopedic surgery within the 24 months prior to the study visit.

Table 2.

Surgical data by participant.

Cohort summary
Type of surgery, n (%)		Time since orthopedic surgery (months)
Knee	8 (33)	11.17 ± 5.51 (0–21)
Hip	10 (42)
Other	6 (25)

Individual data
Patient	Age (years)	Laterality	Orthopedic surgical procedures	Number orthopedic surgeries	Time since last orthopedic surgery (months)
Knee surgery
1	14	L	Medial distal femoral hemiepiphysiodesis	1	7
2	15	R	ACL reconstruction with bone tendon graph and meniscus repair	2	6
3	18	L	Arthroscopic plica exision	1	7
4	18	R	Arthroscopic partial lateral release; medial patellofemoral ligament reconstruction	2	0a
5	22	L	Arthroscopic meniscus repair; two additional arthroscopic knee surgeries	3	15
6	21	R	ACL reconstruction and lateral meniscus repair; knee arthroscope	2	14
7	25	L	ACL and meniscus repair; Scar tissue removal surgery; Scar tissue removal/fat pad resection surgery	3	2a
8	24	L	Arthroscopy and ACL reconstruction with hamstring autograph	1	4
Hip surgery
9	25	R	Arthroscopic labral repair and acetabularplasty	1	16
10	25	L	Femoral derotational osteotomy and hardware removal; hip realignment surgery; shoulder stabilization surgeries	?	12
11	27	L	Hip arthroscopy with labral acetabuloplasty, osteoplasty	1	8
12	29	L	Hip arthroscopy with labral repair, osteochondroplasty, capsular repair; spinal fusion for scoliosis	2	13
13	26	L/R	Left and right hip periacetabular osteotomy with femoral osteochondroplasty	2	11
14	31	R	Hip periacetabular osteotomy, anterior arthrotomy, osteochondroplasty	1	13
15	23	L	Hip arthroscopy, labral repair; hip periacetabular osteotomy, hardware removal from periacetabular osteotomy, arthroscopy and labral reconstruction with fascia late allograph, chondroplasty, synovectomy	4	9
16	27	R	Hip arthroscopic labral repair	1	19
17	24	L/R	Left Hip femoral derotation osteotomy; left foot and ankle reconstruction surgeries (×8); right foot and ankle reconstruction surgeries (×3);	13	8
18	27	L/R	Right hip periacetabular osteotomy; left and right reconstruction hip surgeries due to hip dysplasia and gunstock deformity	7	19
Other
19	23	L	Achilles tendon repair; left thumb repair	2	11
20	25	R	T4–T6 posterior rib head restabilizing surgery; T4–T6 restabilizing surgery (second pass after failed initial surgery)	2	18
21	26	N/A	T4–T11 spinal fusion; spinal implant removal	2	16
22	21	N/A	Temporomandibular Joint Replacement; scoliosis spinal fusion; bimaxillary osteotomy	3	11
23	26	L	Fifth metacarpal repair	1	21
24	25	N/A	Microdiscectomy L5S1	1	8

Note. ACL = anterior cruciate ligament; L = left; MPFL = medial patellofemoral ligament; OCD = osteochondritis dissecans; R = right.

^aThese participants had more than one surgery within a 24-month period.

Self-report measures

Descriptives of self-reported pain levels are included in Table 3. On the day of the visit, surgical participants were asked to rate their pain from 0 to 10, with 10 being the worst pain imaginable. Patients rated a mean of 2.88 ± 2.54 for their pain. Reasons for pain ratings included pain from their surgical site and headaches. Over the last 3 months prior to the visit, 96% of patients rated experiencing any degree of pain in their surgical sites. Of those experiencing pain, 83% reported that pain to be at the surgical site. 79 percent of patients endorsed pain for 3 months or longer, with 75% of that group stating that pain to be at the surgical site. Patients experienced a mean of 15.46 ± 13.03 days of pain per month. Healthy controls rated a mean 0.40 ± 0.22 when asked about their baseline pain levels on the day of the study visit. Most healthy controls reported no pain, while five reported pain levels between 0 and 3 on a scale of 0–10. This pain was confirmed as transient pain (headache, menstrual cramps, muscle soreness) rather than chronic.

Table 3.

Pain experience variables for the surgical group, and emotional functioning self-report data for surgical patients and pain-free controls.

		Surgical patients (n = 24)
Pain scores, (0–10)
	Pain level at visit	2.88 ± 2.54
		(0–7)
	Pain level in the last week	3.92 ± 2.30
		(0–7)
Pain experience variables, n (%)
	Pain in the last 3 months	23 (96)
	Pain at the surgical site	20 (83)
	Pain longer than 3 months	19 (79)
	Pain at the surgical site	18 (75)
	Average # of pain days per month	15.46 ± 13.03
		(0–30)a

Self-report measures	Surgical patients	Nonsurgical controls	Test statistic
	(n = 24)	(n = 20b)
Pain catastrophizing	13.17 ± 10.01	13.85 ± 8.35	U = 222.00, p = .671
	(0–38)	(3–38)
Severity, n (%)
Low catastrophizing	15	12
Moderate catastrophizing	7	7
High catastrophizing	2	1
Fear/somatic arousal, T-Score	56.67 ± 9.61	51.74 ± 9.73	t = 1.66, p = .104
	(44.4–87.9)	(36.3–70.3)
Pain interference, T-Score	54.25 ± 5.28	46.06 ± 5.98	U = 78.50, p < .001***
	(40.7–68.2)	(40.7–56.7)
Severity, n (%)
Normal limits	15	14
Mild	6	3
Moderate	3	3
Severe	0	0
Anxiety, T-Score	54.16 ± 9.30	53.86 ± 7.65	t = 0.12, p = .909
	(37.1–69.9)	(33.5–65.5)
Severity, n (%)
Normal limits	9	10
Mild	7	5
Moderate	8	5
Severe	0	0
Depression, T-Score	53.75 ± 8.03	46.15 ± 7.63	t = 3.20, p = .003**
	(37.1–68.2)	(37.1–60.9)
Severity, n (%)
Within normal limits	12	17
Mild	7	2
Moderate	4	1
Severe	1	0

Note. Data are expressed as mean ± SD or number of participants (%). Continuous data were analyzed using independent-sample t-test for parametric values and Mann–Whitney U test for non-parametric values.

^* p < .05;

^** p < .01;

^*** p < .001.

^a30 day/month was considered pain everyday.

^b n = 19 for fear/somatic arousal T-score.

Emotional functioning measures are reported in Table 3. The post-surgical cohort reported significantly higher depressive symptoms as well as pain interference compared to nonsurgical controls (t = 3.20, p = .003; U = 78.50, p < .001, respectively). There were no significant differences in pain catastrophizing, fear/somatic arousal, or anxiety symptoms between the two groups.

Brain imaging measures

In the surgical cohort, we observed a significant increase in activation of the left SMC during the TS+CS condition compared to the TS condition (t = 2.154, p = .047). FC analyses revealed a significant decrease in connectivity between the right mPFC and left mPFC during TS+CS relative to TS (t = −2.533, p = .022). Additionally, we noted significant reductions in FC between the left SMC and right mPFC (t = −2.857, p = .011), as well as between the left mPFC and left SMC (t = −3.372, p = .004) when comparing the TS+CS response to the TS condition alone. In the control group, there was a significant decrease in FC between the left mPFC and right lPFC during TS+CS compared to TS (t = −5.604, p < .001). Furthermore, FC between the left lPFC and right mPFC (t = −2.127, p = .059) and between the left lPFC and right SMC (t = −2.147, p = .057) did not reach statistical significance. Further comparison of the mean values of neuroimaging measures described above to resting-state neural responses within each group revealed a decrease in FC between the left lPFC and the right SMC, alongside an elevation in activation of the left SMC, relative to the resting state.

Associations between brain imaging and self-report measures

Distinct correlation patterns between the brain functional metrics and self-report pain and emotional functioning measures were observed in the surgical cohort compared to the controls, as illustrated in Figure 2. In the control group, a significant positive correlation was observed between the FC in the left lPFC and the right SMC and anxiety symptom scores during the TS+CS condition (r = 0.500, p = .034). However, this correlation was not observed in the post-surgical group (r = −0.224, p = .371). Conversely, in the post-surgical group, activation of the left SMC in response to the pain stimulus during the TS+CS condition demonstrated a numerical trend toward a negative correlation with anxiety symptom scores (r = −0.391, p = .065), though this did not reach statistical significance. Notably, this pattern was not evident in the control group (r = 0.073, p = .739).

Figure 2.

Presentation of significant correlations between brain imaging metrics and psychological measures in females after orthopedic surgery derived from Pearson’s correlation analysis. Results show a significant positive correlation between the L. lPFC & R.SMC connectivity during the TS+CS condition and anxiety symptom scores in pain-free controls (Graph A) that is not present in orthopedic females group (Graph C). A marginally negative correlation between the activation of the left SMC in response to the pain stimulus during the TS+CS condition and anxiety symptom scores is present in orthopedic females (Graph D). This relationship is not present in pain free controls (Graph B). Note. Anxiety_T = t scores of anxiety symptoms; L = left; l_PFC = lateral prefrontal cortex; m_PFC = medial prefrontal cortex; R = right; r = Pearson correlation coefficient; SMC = somatosensory cortex; p = level of significance.

Discussion

This pilot study, to the best of our knowledge, represents the first neuroimaging investigation of pain-evoked neural responses in AYA female patients following orthopedic surgery. The findings provide insight into the potential neural mechanisms underlying CPSP, particularly focusing on alterations in the PFC and SMC, which appear to be closely associated with the persistence of CPSP and psychological functioning.

Post-surgical pain

When considering our results, first, it is important to note that having post-surgical pain was not an inclusionary criterion for the study, yet the majority of participants in the surgical cohort reported ongoing pain at the surgical site in the moderate range, with the average pain days per month exceeding 15. Interestingly, though pain interference was generally low, the postsurgical pain group endorsed significantly higher depressive symptoms.

FC results

Our results indicate that the surgical group exhibited increased activation in the left SMC during the TS+CS condition compared to the TS condition alone. This heightened activation suggests that the SMC, which plays a pivotal role in the sensory-discriminative aspects of pain may exhibit increased responsiveness to painful stimuli in individuals following surgery and who may be at risk for developing CPSP. Such hyperactivation may contribute to the heightened pain sensitivity observed in these patients and could play a role in perpetuating the chronic pain experience. Moreover, significant reductions in functional communication between the PFC and SMC were observed in the surgical group during the TS+CS condition. The PFC, especially the medial PFC, is crucial for higher-order cognitive processes, including pain modulation and emotional regulation. The reduced FC between the PFC and SMC may indicate a disruption in the neural pathways that regulate pain, which could potentially impair the brain’s ability to modulate pain signals effectively. This disruption could exacerbate the experience of pain, as the PFC’s role in downregulating the SMC’s response to pain stimuli may be compromised. Consequently, the diminished PFC-SMC connectivity observed in this study may underlie the persistent and possibly worsening pain experiences reported by the post-surgical group.

Brain regions measured

These findings are consistent with other studies that have reported aberrant PFC responses in chronic pain conditions. For instance, research utilizing fMRI has demonstrated altered resting-state FC between the PFC and various brain regions, such as the PAG, sensorimotor networks, and the insula in people with chronic pain (Baliki et al., 2014; Yu et al., 2014). Specifically, studies have shown increased connectivity between the PFC and pain-processing regions, which may reflect maladaptive pain modulation mechanisms. Moreover, in pain conditions primarily impacting females, like primary dysmenorrhea and fibromyalgia, decreased PFC connectivity has been associated with increased pain perception and emotional disturbances (Flodin et al., 2014; Gao et al., 2016). These parallels further underscore the role of the PFC in pain chronification and the importance of intact PFC function in pain modulation.

Brain and behavior

The relationship between neural response patterns and anxiety was also examined. In the control group, a positive correlation was observed between FC in the left lPFC and right SMC and anxiety symptoms during the TS+CS condition. This finding suggests that, in a pain-free state, increased communication between the PFC and SMC may serve as a compensatory mechanism to modulate anxiety during pain experiences by maintaining a balanced integration of cognitive and sensory processing. However, this compensatory mechanism appears to be disrupted in post-surgical patients, as the correlation was absent in this group. Interestingly, the post-surgical group exhibited a negative correlation between SMC activation and anxiety symptoms during the TS+CS condition; however, this correlation did not reach statistical significance. While this study is exploratory and based on a relatively small sample, reporting such patterns is valuable for guiding future research directions rather than establishing definitive conclusions. Although this result was not statistically significant, it may hold relevance for consideration in larger, fully powered studies that can determine whether this pattern persists. This clarification ensures transparency in our reporting while aligning with best practices for pilot investigations. This negative correlation suggests that heightened SMC activity might be associated with a reduction in the PFC’s ability to regulate anxiety effectively in patients with CPSP. The maladaptive neural response pattern observed may contribute to the worsening of anxiety symptoms, further complicating the emotional and psychological recovery process. Our findings also align with previous research that has demonstrated the role of the PFC in the top-down and bottom-up control of pain sensation and emotional regulation. The PFC has been extensively implicated in modulating pain through its connections with various brain regions involved in pain processing, including the PAG and amygdala (Chen & Heinricher, 2019; Yu et al., 2014). Disruptions in these connections, as observed in our study, likely contribute to the difficulties in pain modulation and the exacerbation of anxiety seen in patients after surgery and with CPSP. Additionally, structural changes in the PFC, such as decreased gray matter density and alterations in white matter integrity, have been associated with chronic pain and further highlight the central role of the PFC in pain and emotional regulation (Bushnell et al., 2013; Chen & Heinricher, 2019). Higher levels of depressive symptoms in the surgical cohort are consistent with previous research produced by this lab (Moreland et al., 2024) modeling a dose–response relationship between pain and depression that suggests that depression may be a risk factor for the development of post-surgical pain (Giusti et al., 2021). Indeed depression has been shown to impact pain intensity, pain symptoms, and physical functioning(Maallo et al., 2021).

Prevention and early intervention

Aligning with the afore-mentioned peri-surgical pain model proposed by Sieberg and colleagues (Sieberg et al., 2022) aimed at the prevention of pediatric CPSP, the authors also detail potential solutions to prevent and treat pain early during the three peri-surgical phase: pre-surgical, intra-operative, and post-surgical. While these potential solutions are detailed in that publication, some proposed include: (1) the development and implementation of multicenter data repositories that include the administration of biobehavioral assays (e.g., bedside quantitative sensory testing, inflammatory/genetic markers, pain and emotional functioning questionnaires) prior to surgery, which are both feasible to administer and predictive of post-surgical outcomes; (2) utilizing patient and provider data stored in electronic health records and clinical research informatics tools to study and identify risk factors for the development of CPSP, with the ultimate goal of developing global clinical data research networks that have the capability of deep clinical phenotyping;(3) appropriate presurgical preparation and education that target pain risk and coping; and (4) randomized control trials as well as mechanistic clinical trials targeting CPSP treatment. In cases when pain chronification has already occurred, treatments need to be employed early in the clinical course to help reverse or diminish chronic pain–associated comorbidities.

Caveats

This study has several limitations. First, in the surgical group, pre-existing chronic pain and time since surgery may have influenced neural and emotional responses, limiting generalizability. While the control group was intentionally selected to include only pain-free individuals to serve as a “clean” comparison group, some participants in the surgical cohort had a history of chronic pain prior to their most recent surgery. Second, while we observed reduced within-PFC and PFC-SMC connectivity during the TS+CS condition relative to TS alone in pain-free controls, as well as a negative correlation between SMC activation and anxiety symptoms within the patient group, these findings did not reach statistical significance. While these patterns align with existing literature on pain modulation and emotional regulation, caution is necessary in their interpretation. Future studies with larger, well-powered samples will be required to determine whether these effects are robust and clinically meaningful.

Third, our neuroimaging measures were confined to cortical regions, potentially omitting important CPSP-related changes in subcortical and brainstem areas. Future investigations should integrate deeper brain regions, such as the basal ganglia and amygdala, which are critical in pain processing and modulation (Borsook et al., 2010), to gain a more comprehensive understanding of the impact of CPSP on the developing brain. Fourth, while our pilot study had a relatively small sample size, it aligns with other recent fNIRS studies on pain that have reported similar sample sizes (Karunakaran et al., 2023). However, we acknowledge that the limited sample size may affect the generalizability of our results. Despite this limitation, our study establishes essential feasibility benchmarks and measurement parameters, laying the groundwork for future, larger-scale studies.

Another notable limitation is the lack of racial and ethnic diversity in the sample. Since pain perception, neural responses, and emotional functioning can be shaped by cultural, genetic, and socioeconomic factors (Fillingim, 2017), this homogeneity may limit the generalizability of our findings. Future studies should include participants from more diverse backgrounds to enhance the validity and applicability of the results, and offer a fuller understanding of how these factors influence pain and emotional responses. The study also included a broader age range, from adolescence to young adulthood. While this is consistent with other studies of chronic pain in youth (Murray et al., 2022; Sieberg et al., 2013), and chronic pain at any ages is linked to CNS changes (Hestbaek et al., 2006), developmental factors like puberty and mood disorder onset, may influence brain metrics and pain perception. Larger sample sizes and prospective designs across the lifespan are needed to explore how age and development affect the onset, maintenance, and exacerbation of chronic pain. Finally, this study did not include male participants; however, the focus on AYA females is important as pain in women has historically been overlooked (As-Sanie et al., 2016). Future studies also including male participants should be conducted to allow for the assessment of potential sex differences in CPSP risk among AYAs.

Conclusion

This pilot study offers preliminary insights into the neural and psychological mechanisms potentially underlying CPSP in AYA females who have undergone recent orthopedic surgery. The findings suggest that altered neural responses, particularly within the PFC and SMC, may play a role in the persistence of chronic pain and the exacerbation of anxiety symptoms. These neural alterations underscore the importance of developing and implementing interventions that address both the biological and psychosocial dimensions of CPSP to improve pain management, and enhance the quality of life for those affected. Moreover, access to healthcare remains a significant concern, as young females may encounter barriers due to gender-based disparities, such as biases in the healthcare system or socio-economic factors, which could limit their ability to seek care. These barriers may further complicate CPSP management, as delayed or inadequate treatment could worsen both pain and its associated psychological impact.

Data availability

The data used in this study are available upon request. Researchers can request access to the data by contacting csieberg@mgh.harvard.edu

Author contributions

Ziyan Wu (Conceptualization [lead], Data curation [equal], Formal Analysis [lead], Investigation [lead], Methodology [lead], Software [lead], Supervision [equal], Writing—original draft [lead], Writing—review & editing [lead]), Margaret Moreland (Formal Analysis [supporting], Investigation [equal], Project administration [lead], Supervision [supporting], Writing—original draft [supporting], Writing—review & editing [supporting]), Maya Jotwani (Formal Analysis [supporting], Investigation [supporting], Writing—original draft [supporting], Writing—review & editing [supporting]), Melissa Christino (Methodology [supporting], Supervision [supporting], Writing—original draft [supporting], Writing—review & editing [supporting]), David Borsook (Conceptualization [equal], Methodology [lead], Resources [lead], Software [lead], Supervision [equal], Writing—original draft [supporting], Writing—review & editing [supporting]), and Christine B. Sieberg (Conceptualization [lead], Data curation [equal], Formal Analysis [supporting], Funding acquisition [lead], Investigation [equal], Methodology [lead], Project administration [lead], Resources [lead], Supervision [lead], Writing—original draft [equal], Writing—review & editing [lead])

Funding

This research is funded by a K23 Supplement (K23GM123372-04S1); Loan Repayment Award (L30GM134514); and R35 MIRA Award (R35GM142676) all from the National Institute of General Medical Sciences awarded to CBS, as well as a grant from the Cathedral Fund awarded to CBS.

Conflict of interest

Dr. David Borsook discloses that he consults for Luminous Mind Inc and Pattern Computer Inc. but this role is not in conflict with this manuscript. No other authors or co-authors on this manuscript have a conflict of interest or any disclosures.