The Relationship Between Plasma BDNF and Pain in Older Adults With Knee Osteoarthritis

Abstract

Osteoarthritis (OA) is the most prevalent cause of chronic pain and disability in people aged ≥45 years, with the knee being the most affected joint. Neurotrophic factors like brain-derived neurotrophic factor (BDNF), which promotes neurogenesis and neuroplasticity, have been shown to significantly affect chronic pain. This study aimed to investigate the relationship between resting plasma BDNF levels and clinical pain and quantitative sensory testing measures in older adults with knee OA pain. For this secondary analysis, a previously reported dataset was used comprised of older adults with knee OA who underwent quantitative sensory testing. A comprehensive generalized linear model (GLM) was built to understand the relationships between BDNF and important covariates, followed by the elastic net (EN) method for variable selection. GLM was then performed to regress BDNF levels against only the variables selected by EN. The mean age of the sample was 60.4 years (SD = 9.1). Approximately half of the participants were female (53%). Plasma BDNF levels were positively associated with heat pain threshold and the numeric rating scale of pain. Future mechanistic studies are needed to replicate and extend these findings to advance our knowledge of the underlying mechanisms of BDNF in knee OA and other chronic pain conditions.

Of all health problems, chronic pain remains one of the most economically burdensome conditions, affecting approximately one in three people in the United States, with an annual cost of more than $635 billion (Gaskin & Richard, 2012). Osteoarthritis (OA), which commonly occurs in weight-bearing joints, is a leading cause of chronic pain in older adults, leading to functional impairment, disability, and decreased quality of life (Chu et al., 2014; Lawrence et al., 2008; Neogi, 2013). The knee is the most frequently affected joint (Hunter et al., 2014; Silverwood et al., 2015; Wallace et al., 2017) and knee OA affects an estimated 265 million people worldwide, resulting in 8.3 million years lived with disability (James et al., 2018).

Pain is the most common symptom of knee OA and it is regarded as a complex phenomenon involving multiple mechanisms including hyperactivity of the central nervous system (CNS), often referred to as central sensitization (Ahn et al., 2019; Lluch et al., 2014; Woolf, 2011), and peripheral input from the affected joint, referred to as peripheral sensitization (Richner et al., 2014; Syx et al., 2018). However, the mechanism behind the peripheral input concept remains unclear, as knee OA pain severity has been shown in several studies to lack direct correlation with radiographic joint damage (Bedson & Croft, 2008; Ribeiro et al., 2020). This disparity in the severity of pain and the extent of joint damage suggests that factors other than peripheral sensitization play a role in knee OA pain. Indeed, central sensitization, a phenomenon characterized by changes in the central pain processing network that result in disproportionate amplification of central neuronal response to either normal or sub-threshold sensory input, leading to hypersensitivity to pain, (Meeus & Nijs, 2007; Woolf, 2004) has often been proposed as a possible explanation for the lack of correlation between pain intensity and radiographic joint damage in knee OA patients (Bedson & Croft, 2008; Lluch et al., 2014). Moreover, recent compelling evidence also indicates that some knee OA patients experience pain that is centrally mediated (Lluch et al., 2014; Soni et al., 2019). Furthermore, both quantitative sensory testing analyses and functional magnetic resonance imaging studies have confirmed the role of central sensitization in influencing pain in knee OA patients (Gwilym et al., 2009; Soni et al., 2019).

Central sensitization may occur as a result of dysregulation in neurotrophic factors, leading to synaptic plasticity (Latremoliere & Woolf, 2009; Obata & Noguchi, 2006). One of the most studied neurotrophic factors with a role in the facilitation of central sensitization is brain-derived neurotrophic factor (BDNF). BDNF is a synaptic modulator that is produced by nociceptor neurons and then transported and released into the spinal cord (Zhou & Rush, 1996) in an activity-dependent manner (Balkowiec & Katz, 2000), where it facilitates central sensitization to pain (Heppenstall & Lewin, 2001; Latremoliere & Woolf, 2009; Pezet & McMahon, 2006; Vanelderen et al., 2010). BDNF is found in the CNS, muscle tissue, and peripheral blood (Gomes et al., 2013; Yamamoto & Gurney, 1990), where it can alter pain pathways through different complex mechanisms (Nijs et al., 2015).

BDNF has also been shown to play a role in neuron survival and differentiation by binding to and activating the tropomycin receptor kinase B (TrkB) and activating downstream signaling pathways (Hennigan et al., 2007; Reichardt, 2006; Song, Li et al., 2008; Young et al., 2007). This downstream signaling pathway implicated in the regulation of pain transmission is one of the brainstem descending pathways connecting the periaqueductal gray, the rostral ventromedial medulla, and the spinal cord (Morgan et al., 2006). Because of these and other characteristics, BDNF is thought to have therapeutic potential in the management of chronic pain (Binder & Scharfman, 2004; Massa et al., 2010; Nagahara & Tuszynski, 2011; Paul & Hughes, 2003; Smith, 2014). Eaton et al. demonstrated this in an animal study by successfully producing significant analgesic effects by genetically inducing overexpression of BDNF in rats (Eaton et al., 2002). In a related study, Azoulay and colleagues identified low serum BDNF levels in patients with peripheral neuropathic pain (Azoulay et al., 2020), which was consistent with their previous findings in lymphoma and myeloma patients (Azoulay et al., 2019; Azoulay et al., 2019), implying that lower BDNF levels could be linked to hyperalgesia.

Given that knee OA pain is a complex phenomenon that is difficult to treat and a leading cause of disability in older adults, biomarkers that can help predict disease progression or the efficacy of therapeutic interventions are overdue. However, despite our current understanding of the role of biological mediators in pain sensitivity in chronic pain conditions such as knee OA, there is a lack of clinical evidence on the exact role neurotrophic factors such as BDNF play in the modulation of pain in patients with knee OA. Therefore, this study aimed to examine the relationship between resting plasma BDNF levels and pain in older adults with pain associated with knee OA.

Materials and Methods

The present report is a secondary analysis of data retrieved from a previously reported single-center, experimenter- and participant-blinded, randomized, sham-controlled pilot clinical study that took place at the University of Florida Institute on Aging, to evaluate the efficacy and safety of 5 daily sessions of transcranial direct current stimulation (tDCS) with the anode electrode placed over the primary motor cortex of the hemisphere contralateral to the pain-affected area of the body and with the cathode electrode placed over the supraorbital region ipsilateral to the affected area on clinical pain severity and mobility performance in older persons with knee OA. The trial, which was conducted between September 2015 and August 2016, was registered at ClinicalTrials.gov (NCT02512393). All study procedures were approved by the Institutional Review Board of the University of Florida and all participants provided written informed consent. The original study methods including detailed selection criteria and study enrollment processes were described in full elsewhere (Ahn et al., 2017). Briefly, 40 English-speaking older adults aged 50–70 years meeting the inclusion criteria were originally enlisted in the study, of which blood samples of 38 subjects were available. The inclusion criteria comprised self-report of unilateral or bilateral knee OA pain, according to the American College of Rheumatology criteria. Participants with specific conditions that could potentially confound symptomatic OA-related outcome measures or coexisting diseases that could hinder the completion of the protocol (e.g., prosthetic knee replacement or non-arthroscopic surgery to the affected knee, peripheral neuropathy, rheumatoid arthritis, cognitive impairment, history of brain surgery, tumor, seizure, stroke, or intracranial metal implantation) were excluded.

Procedures

After giving informed consent, participants were surveyed for demographic information such as age, gender, and ethnicity. The severity of OA was assessed using the Kellgren-Lawrence scores (range = 0-4, with higher scores indicating severe OA severity; Damen et al., 2014). Participants identified their most affected knee in terms of pain severity, which then served as their index knee. Clinical pain was evaluated using multiple validated pain scales including the Numeric Rating Scale (NRS; range = 0–100; Hjermstad et al., 2011) for current knee pain, the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC; range = 0-96; Bellamy et al., 1988) for OA-related symptoms, and the Short-Form McGill Pain Questionnaire (SF-MPQ-2; range = 0–220; Melzack, 1987) for a detailed evaluation and characterization of pain symptoms. Following clinical assessment, the researchers collected a 6-ml blood sample from each participant via venipuncture according to standard phlebotomy protocol, then processed and stored the sample until analysis, as described below. Participants then underwent quantitative sensory testing.

Blood was drawn into ethylenediaminetetraacetic acid plasma tubes. Within 30 min of collection, samples were centrifuged at 1,600 × g for 15 min at 4°C. Next, plasma was removed, aliquoted, and stored at −80°C. The concentration of BDNF was determined by sandwich enzyme-linked immunosorbent assay kits according to the procedures supplied by the manufacturer (DuoSet, R&D Systems, Minneapolis, MN, USA) and routinely performed in our laboratory (Mourao et al., 2019; A. L. Rocha et al., 2017; N. P. Rocha et al., 2018). All samples were assayed in duplicate, and the technician was blind to the sample’s clinical information. The detection limit for the assay was 5 pg/ml. The intra- and inter-assay coefficients of variability were < 10%.

A quantitative sensory testing battery was completed, including heat pain threshold, pressure pain threshold, and cold pain sensitivity. The heat pain threshold was measured via a computer-controlled TSA-II neurosensory analyzer (Medoc Ltd., Ramat Yishai, Israel). The thermal stimuli were delivered to the index knee using an ascending method of limits. From a baseline of 32°C, the thermode temperature increased by 0.5°C per second until the participants responded by pressing a button to stop the heat stimulus. Participants were given and instructed to press a button when the sensation “first becomes painful” to assess the heat pain threshold. We also measured pressure pain threshold by applying blunt mechanical pressure at a constant rate of 0.3 kgf/cm²/s to the index knee via a handheld digital pressure algometer (Wagner, Greenwich, CT, USA). Participants were instructed to notify the experimenter when the sensation “first becomes painful” to assess the pressure pain threshold. For cold pain sensitivity, participants were asked to rate the cold pain intensity (0–100) from the immersed hand up to the wrist in a cold-water bath at a temperature of 12 C.

Statistical Analysis

SAS Version 9.4 (SAS Institute, Cary, NC) was used for all statistical analyses. Normality assumption assessment, outlier detection, and inspection for multicollinearity were conducted prior to subsequent analyses. Two participants (one from each group) were omitted from all statistical analyses due to missing data. A comprehensive generalized linear model (GLM) was built to understand the relationships between BDNF levels and important covariates, including: tDCS treatment group indicator (active vs. sham); age; gender; race; Kellgren-Lawrence scores; body mass index (BMI); clinical pain (i.e., NRS, WOMAC, and SF-MPQ-2); and quantitative sensory testing (i.e., heat pain threshold, pressure pain threshold, and cold pain sensitivity). The elastic net (EN) method was used for variable selection, and then GLM was performed to regress BDNF levels against the variables selected by EN only.

Results

Demographic Characteristics

Of the 38 subjects enrolled in this study, approximately half (52.6%, n = 20) were female, as shown in Table 1. The mean age of this cohort was 60.40 years (standard deviation [SD] = 9.1), the mean plasma BDNF was 296.1 pg/ml (SD = 302.9), and the baseline heat threshold was 41.18°C (SD = 4.53).

Table 1.

Demographic and Clinical Characteristics of Participants.

Characteristics	Value
Age, years, mean (SD)	60.4 (9.13)
Gender, n (%)
Male	18 (47.4)
Female	20 (52.6)
Race, n (%)
White	18 (47.4)
Asian	20 (52.6)
Kellgren–Lawrence Score, n (%)
0 (none)	12 (31.6)
1 (doubtful)	7 (18.4)
2 (minimal)	10 (26.3)
3 (moderate)	8 (21.1)
4 (severe)	1 (2.6)
BMI, kg/m2, mean (SD)	26.47 (3.78)
BDNF, pg/ml, mean (SD)	296.09 (302.99)
NRS, mean (SD)	23.68 (12.56)
WOMAC, mean (SD)	24.55 (13.52)
SF-MPQ-2, mean (SD)	5.16 (6.20)
Heat pain threshold, oC, mean (SD)	40.94 (4.51)
Pressure pain threshold, kgf, mean (SD)	2.56 (1.18)
Cold pain sensitivity, mean (SD)	66.84 (26.19)

Note. N = 38. SD = standard deviation; BMI = body mass index; BDNF = brain-derived neurotrophic factor; NRS = numeric rating scale; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-MPQ-2 = Short-Form McGill Pain Questionnaire.

Relationship Between BDNF and Pain

The BDNF levels followed an exponential distribution, and no outliers were detected within the sample. The GLM regression results suggested that there was a significant relationship between BDNF and NRS after adjusting for other confounders. Regressing BDNF against heat pain threshold and NRS with confounding factors (i.e., age and gender) is presented in Table 2. In the log scale, the estimated coefficient of heat pain threshold was 0.10 (p-value = .04), and the estimate of NRS was 0.03 (p-value = .03).

Table 2.

Results of Regression Analysis.

Variable	log(β)	SE	t Value	p Value
Treatment group	NA	NA	NA	NA
Age	NA	NA	NA	NA
Gender	NA	NA	NA	NA
Race	0.18	2.25	0.08	0.94
Kellgren–Lawrence score	NA	NA	NA	NA
BMI	0.04	0.05	0.71	0.48
NRS	0.03	0.01	2.25	0.02
WOMAC	NA	NA	NA	NA
SF-MPQ-2	NA	NA	NA	NA
Heat pain threshold	0.10	0.05	2.04	0.04
Pressure pain threshold	NA	NA	NA	NA
Cold pain sensitivity	−0.01	0.01	−0.84	0.40

Note. SE = standard error; BMI = body mass index; NRS = Numeric Rating Scale; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-MPQ-2 = Short-Form McGill Pain Questionnaire; NA = variables were dropped after the variable selection step.

Discussion

In this secondary analysis, the relationship between resting-state plasma levels of BDNF, a well-studied neurotrophic factor, and pain was investigated. Our findings included a statistically significant relationship between plasma levels of resting-state BDNF and heat pain threshold and numeric rating scale of pain in older adults with knee OA. These results indicate that BDNF levels may influence pain in chronic pain states and, therefore, may be useful as predictors of deteriorating chronic pain conditions as well as evaluating the efficacy of various pain relief therapies.

Our findings are consistent with previous clinical studies evaluating the effect of BDNF in painful conditions. Stefani et al. (2012) examined pain modulation via heat and pressure pain thresholds in healthy subjects. Their study revealed that higher levels of BDNF were correlated with higher pain thresholds in females. In another study performed by Leifsdottir et al. (2020), participants with rheumatoid arthritis were examined for the effects of serum BDNF on pain and mood perception. They concluded that high serum BDNF levels were linked to a lower number of tender points and a higher pain threshold. Furthermore, in a related animal study, Eaton et al. (2002) demonstrated a near-complete improvement of thermal hyperalgesia in rats that received an over-expressed BDNF gene. While these studies corroborate our findings, it is important to emphasize that dissimilar findings were reported in other studies. For example, in a cross-sectional study to establish the association between serum BDNF levels and heat pain threshold in healthy adult males, Dussán-Sarria et al. (2018) reported that serum BDNF levels were negatively correlated to the heat pain threshold. Importantly, however, circulating BDNF levels are known to be influenced by gender (Lommatzsch et al., 2005) and age (Oh et al., 2016), and the Dussán-Sarria et al. study was restricted to males with an average age of 25.7 years (SD = 5.28).

The exact mechanism underlying the relationship between BDNF levels at rest and heat pain threshold is not clear. Previous clinical studies have focused on the potential negative effects of BDNF increase in chronic painful conditions (Diz et al., 2017; Garraway et al., 2003; Kerr et al., 1999; Merighi et al., 2008; Thompson et al., 1999). While the role of BDNF in the sensitization of pain pathways seems to be adaptive in the early stages of injury, it can become maladaptive in the chronic stage (Nijs et al., 2014). Moreover, BDNF possesses the capacity to augment pain pathways at different levels, including the brain, spinal cord, and peripheral nociceptors. Interestingly, this multilevel effect of BDNF could help explain the complex pathophysiology of chronic pain in knee OA because the pain has a strong CNS component but manifests primarily in the periphery. Researchers, on the other hand, cannot overlook the physiological effects of BDNF, which include contributing to the growth and survival of neurons, functioning as a neurotransmitter modulator, and playing an important role in neural plasticity (Hennigan et al., 2007; Reichardt, 2006; Song et al., 2008; Young et al., 2007). Indeed, BDNF exerts these effects by binding to TrkB to form the BDNF–TrkB system, which activates downstream signaling pathways in the midbrain periaqueductal gray–rostral ventromedial medulla circuits, suggesting their role in synaptic activity related to pain modulation (Ren & Dubner, 2007). The administration of BDNF into the midbrain of rats, for example, resulted in an analgesic effect, therefore providing compelling evidence that BDNF has antinociceptive properties (Siuciak et al., 1995). Thus, while we found a connection between BDNF and heat pain threshold, as well as a significant association between BDNF and subjective self-reported pain (NRS), other mechanisms may be involved in its role in pain processing.

The present study has some limitations that need to be acknowledged when interpreting the results. First, since we enrolled a limited number of adults, we were unable to conduct sensitivity analyses such as sub-group analyses, which would have revealed if the association between BDNF and heat pain threshold, as well as BDNF and self-reported pain score (NRS), was modified by characteristics such as BMI, age, race/ethnicity, or disease severity. Second, because the study was a secondary analysis, we were restricted to data obtained as part of the original study. As a result, we lacked information on certain key variables such as depression, a known cofounder of BDNF levels (Diniz et al., 2010; Teixeira et al., 2010). Third, we measured BDNF levels once for per participant instead of taking multiple measurements per participant within 24 hr. Because plasma BDNF levels in both men and women can differ significantly throughout a 24-hr cycle, with peak timing that differs greatly between individuals, serial BDNF measurements are recommended to assist in the discovery of more accurate outcomes (Cain et al., 2017). However, it is important to note that the collection of data used in the current study preceded the recommendation by Cain et al. Finally, we relied on plasma BDNF levels in the current study, and there is undeniably a debate regarding the most appropriate blood-based biological matrix for BDNF measurement. This debate revolves around the fact that platelets store BDNF and can release it upon stimulation, not necessarily involving molecule stability (Teixeira et al., 2010). It is worth mentioning that different groups have used plasma as the main matrix with consistent results, and the findings were comparable to those with serum. Moreover, several authors have preferred using plasma because, during serum preparation, platelets can be activated to produce and secrete molecules that may affect the peripheral levels of BDNF.

The results of this study lay a solid basis for subsequent research. First, future studies should consider these limitations by enrolling larger samples that can support sensitivity and subgroup analyses and also should consider serial BDNF measurements over 24 hr. Second, future studies involving knee OA patients should endeavor to compare findings with healthy age-matched volunteers. Third, the mechanisms underlying the relationship between resting plasma BDNF levels and heat pain threshold as well as self-reported pain scores (NRS) should be investigated further. Finally, the importance of BDNF as a valid predictor of deteriorating chronic pain conditions and a measure of the efficacy of various pain relief therapies must be evaluated.

Conclusion

In conclusion, the concentration of plasma BDNF levels were positively associated with heat pain threshold and numeric rating scale of pain. Future mechanistic studies are needed to replicate and extend these findings to advance our knowledge of the underlying mechanisms behind the relationship between BDNF and pain in knee OA and other musculoskeletal pain conditions.