Perturbations in Neuroinflammatory Pathways Are Associated With a Worst Pain Profile in Oncology Patients Receiving Chemotherapy

Patients and Settings

This study is part of a larger, longitudinal study of the symptom experience of oncology outpatients receiving chemotherapy.^60,75 Eligible patients were ≥18 years of age; had a diagnosis of breast, gastrointestinal, gynecological, or lung cancer; had received chemotherapy within the preceding four weeks; were scheduled to receive at least two additional cycles of chemotherapy; were able to read, write, and understand English; and gave written informed consent. Patients were recruited from two Comprehensive Cancer Centers, one Veteran’s Affairs hospital, and 4 community-based oncology programs.

Study Procedures

The study was approved by the Institutional Review Board at each of the study sites. Of the 2234 patients approached, 1343 consented to participate (60.1% response rate). The major reason for refusal was being overwhelmed with their cancer treatment. Eligible patients were approached in the infusion unit during their first or second cycle of chemotherapy by a member of the research team to discuss study participation and obtain written informed consent. Blood for ribonucleic acid (RNA) isolation was collected at the enrollment assessment. Medical records were reviewed for disease and treatment information. For this study, a total of 717 patients provided a blood sample for the analyses (Supplemental Figure 1). Of these 717 patients, 357 patients had their samples processed using RNA sequencing (i.e., RNA-seq sample) and 360 patients had their samples processed using microarray (i.e., microarray sample) technologies.

Instruments

Data Analysis

RNA-seq Performance

Of the 357 patients in the RNA-seq sample, 72 were in the None and 120 were in the Severe pain classes. Median library threshold size was 9,042,589 reads. Following the application of quality control filters, 10,881 genes were included in the final analysis. The common dispersion was estimated as 0.26493, yielding a biological coefficient of variation of 0.5147 well within the expected value for clinical samples.^41,51

RNA Microarray Performance

Of the 360 patients in the microarray sample, 86 were in the None and 111 were in the Severe pain classes. All of these samples demonstrated good hybridization performance for biotin, background negative, and positive control assays on the arrays. Limma was used for background correction, quantile normalization, and log2 transformation.⁷⁶ Following quality control filters, 46,542 loci were included in the final analysis.

Differences in Demographic and Clinical Characteristics

Of 192 patients with phenotypic data in the RNA seq sample (Table 1), compared to the None class, the Severe class was more likely to be female; had fewer years of education; a higher body mass index; a lower performance status; a higher number of comorbidities; and a higher comorbidity burden; were more likely to have adult care responsibilities; had a lower annual income, and were more likely to self-report diagnoses of anemia or back pain.

Of 197 patients with phenotypic data in the microarray sample (Table 2), compared to the None class, the Severe class had fewer years of education; a lower performance status; a higher number of comorbidities, a higher comorbidity burden, and a higher number of prior cancer treatments; were more likely to be not married or partnered; had lower annual income; were less likely to be employed; were less likely to exercise regularly; were more likely to have a current or previous history of smoking; and were more likely to self-reported diagnoses of anemia, depression, osteoarthritis, or back pain.

Differences in Pain Characteristics

As summarized in Table 3, no differences were found between the two Severe pain classes in any of the pain characteristics evaluated. In brief, the majority of the patients had both cancer and non-cancer pain; were experiencing chronic pain; had worst pain scores in the severe range; and reported moderate levels of interference.

Logistic Regression Analyses

In the logistic regression analysis for the RNA-seq sample (Table 4), five variables were retained in the final model (i.e., gender, income, adult care responsibilities, KPS score, self-reported diagnosis of back pain) and were used as covariates in the gene expression analysis. Patients who were female, had a lower annual income, had adult care responsibilities, had a lower functional status, and self-reported a diagnosis of back pain were more likely to belong to the Severe class.

In the logistic regression analysis for the microarray sample (Table 4), eight variables were retained in the final model (i.e., married/partnered, exercise on a regular basis, current or history of smoking, KPS score, number of prior cancer treatments, self-reported diagnoses of anemia, depression, back pain) and were used as covariates in the gene expression analysis. Patients who were not married or partnered did not exercise on a regular basis, had a current or history of smoking, had a lower KPS score, had a higher number of prior cancer treatments, and self-reported a diagnosis of depression or back pain were more likely to belong to the Severe class.

Perturbed Signaling Pathways Associated With Worst Pain Severity

Of the 13 surrogate variables identified for the RNA-seq sample, none were associated with class membership. The final differential expression model for this sample included 13 surrogate variables and five phenotypic characteristics. Of the 15 surrogate variables identified for microarray sample, two were associated with class membership and were excluded from the final model. The final differential expression model for this sample included 13 surrogate variables and eight phenotypic characteristics. For both samples, a total of 3,868 genes were included in the PIA analyses. Using Fisher’s Combined Probability method, across the two samples, 51 KEGG signaling pathways were significantly perturbed at an FDR of 0.015 (see Supplemental File 2). Of these, nine were related to neuroinflammatory mechanisms (Table 5).

Complement and Coagulation Cascades Pathway

As noted in a recent review,⁸² complement signaling is important in directing neuronal responses to tissue injury and nerve trauma. After an acute injury, a complex interplay occurs between nociceptive neurons and immune cells to promote healing and facilitate guarding of the site of injury. The complement works by activating immune cells and stimulates these cells to release inflammatory mediators. However, in the setting of chronic pain, persistent or unbalanced signaling of complement factors occurs.⁸² In fact elevated levels of several key complement factors (e.g., C3a, C5, C5a) were found in patients with rheumatoid arthritis^35,40,59 and osteoarthritis.^81,87

Phagosome Pathway

While little is known about the interaction between phagocytes and nociceptor signaling, emerging evidence suggests that phagocytes (i.e., macrophage, dendritic cells) can contribute to the development and maintenance of pain.²⁷ For example, macrophages release a variety of immune mediators that bind to receptors on nociceptors. This binding induces neuronal hyperexcitability and hypersensitivity.²⁷ In addition, microglia play an essential role in the initiation of neuroinflammation by releasing the complement components C1q and C3 that induce phagocytosis by binding to neuronal surfaces.⁹ For example, following high-frequency stimulation-induced spinal long-term potentiation in rats,⁹⁴ the number of activated microglia in the dorsal and ventral horn increased, which suggests an association between microglial activation, spinal plasticity, and chronic pain hypersensitivity. In addition, in a recent preclinical study,⁹³ blockade of spinal microglia function significantly attenuated neuropathic pain through the inhibition of neuroinflammation.

Cytokine-cytokine Receptor Interaction Pathway

As noted in one review,⁴⁹ in response to tissue injury, nociceptors induced a number of pro- and anti-inflammatory mediators that directly bind to and activate cytokine receptors. This bidirectional interaction between pain and inflammation leads to hyperexcitability and hypersensitivity of nociceptor neurons (i.e., peripheral sensitization).³³ For example, in both preclinical and clinical studies, interleukin (IL)-6, tumor necrosis factor-alpha (TNF-α), and IL-1β appear to be involved in the development and maintenance of pain associated with cancer,^21,72 rheumatoid arthritis,¹ and peripheral neuropathies.^83,92 In contrast, in a recent preclinical study,⁴² IL-10 attenuated pain hypersensitivity following cisplatin administration, which suggests that IL-10 may decrease neuroinflammation. In another preclinical study,⁹⁰ IL-4 receptor knockout mice showed upregulation of pro-inflammatory mediators. Furthermore, cytokine signaling in the periphery is transmitted to the central terminals of the nociceptors and the brain.²⁸ As a result, pro-inflammatory cytokines activate microglia that contributes to the development and maintenance of central sensitization.^39,93 For example, following sciatic nerve chronic constriction injury, differentially perturbed cytokine-cytokine receptor interaction pathways were found in dorsal horn tissues.¹³

Chemokine Signaling Pathway

Chemokines are chemotactic cytokines that control the movement of circulating peripheral immune cells (e.g., T lymphocytes, natural killer cells, B cells, dendritic cells) by mediating cell-to-cell communication.³⁴ In addition, they activate G-protein-coupled receptors.^10,28 The chemokine (C-X-C motif) ligand-(CXCL) receptor pair serves as a mediator for glia-neuron communication^61,95 that when activated leads to persistent hyperexcitability and neuroplasticity in peripheral nociceptors.²⁸ Subsequently, this chemokine signaling alters nociceptive transduction through activation of chemokine receptors in dorsal root ganglia (DRG) cells.⁴⁷ As noted in one review of rodent models of neuropathic pain,⁸⁵ upregulation of the expression of C-C motif chemokine ligand 2 (CCL2) and its receptor (CCR2) in DRG neurons was identified. In addition, in another preclinical study of autologous nucleus pulposus-induced pain,⁹⁵ increases in chemokine CCL2/CCR2 signaling in DRG and spinal cord were associated with the maintenance of lumbar disc herniation-induced pain. Of note, the administration of CCR2 antagonist decrease mechanical allodynia.

Additional preclinical studies provide evidence that chemokines and their receptors play a crucial role in cancer pain,^61,62,89 visceral pain,⁴ and inflammatory pain.^11,88 For example, chemokines appear to be involved in the regulation of neuronal excitability, neurotransmitter release, and neuronal survival⁸⁴ by enhancing the activity of the N-methyl-D-aspartate (NMDA) receptors in dorsal horn neurons.²⁶ In a recent preclinical study of inflammatory pain,¹² CXCL1/CXCR2 signaling induced an enhancement of NMDA-induced currents in spinal cord neurons. The authors suggested that CXCL1/CXCR2 drives hyperactivity of NMDA receptors, which in turn mediates persistent inflammatory pain through the induction and maintenance of central sensitization.¹²

Janus Kinase-signal Transducer and Activator of Transcription (JAK-STAT) Signaling Pathway

As noted in one review on the association between the JAK-STAT signaling pathway and pain,¹⁰ this pathway is involved in both pro-and anti-nociceptive mechanisms through numerous inflammatory responses. For example, IL-6 binding to the IL-6 receptor induces the activation of the JAK-STAT transduction pathway.⁵⁴ Phosphorylated JAK1 and 2 and STAT3 are translocated to the nucleus, leading to the expression of target genes and an increase in the release of proinflammatory cytokines.^55,92 In contrast, IL-4 binding to the IL-4 receptor results in the activation of JAKs 1 and 3 and consequently STAT6, which in turn leads to the inhibition of the production of proinflammatory cytokines.¹⁰ In fact, a JAK-STAT inhibitor was approved by Food and Drug Administration and is used to treat rheumatoid arthritis.³¹ Given the positive findings across other studies,^2,48,65 the use of a JAK-STAT inhibitor may warrant investigation in oncology patients with severe pain.

Calcium Signaling Pathway

Primary afferent neurons express multiple types of voltage-gated calcium channels (VGCCs), including N- and T-type channels.⁸ Calcium signaling mediated by these VGCCs is involved in the development and maintenance of chronic pain, including neuropathic⁸⁰ and inflammatory⁷³ pain, through the induction of a multifaceted cascade of signaling molecules.³⁰ This calcium signaling-related cascade begins with an influx of calcium ions into the post-synaptic neuron through NMDA and a-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptors for glutamate and/or VGCCs.⁸ Subsequently, modulation of the ion channel pool induces sensitization and hyperexcitability of sensory neurons by releasing glutamate, an excitatory neurotransmitter at central nerve endings.^30,50

Of note, several preclinical studies suggest that N-type calcium channel (Cav2.2) knockout mice have attenuation of inflammatory and neuropathic pain.⁶⁹ Upregulation of N-type calcium channels in rat DRG neurons was associated with neuropathic pain.^46,91 In addition, the deletion of the T-type calcium channel genes (e.g., Cav3.2) in rats was associated with major antinociceptive effects.^7,15 L-type calcium channels (Cav1.2 and Cav1.3) are primarily located on post-synaptic channels that are involved in dorsal horn hyperexcitability and short- and long-term neuronal plasticity.⁶⁸ As demonstrated in a preclinical study,⁶⁶ L-type calcium channels contribute to the integration of afferent inputs and the maintenance of hyperexcitability in DRG neurons by controlling plateau potentials.

Retrograde Endocannabinoid Signaling Pathway

Retrograde endocannabinoid signaling is implicated in several forms of short- and long-term synaptic plasticity.⁶ The endocannabinoid system includes cannabinoid receptor subtypes 1 (CB1) and 2 (CB2), as well as their ligands, namely endocannabinoids.⁶ The endocannabinoids synthesized in response to activation of post-synaptic metabotropic glutamate receptors (mGluRs) travel retrogradely to bind CB1 receptors and impede neurotransmitter release through inhibition of presynaptic VGCCs.³⁸ In addition, retrograde endocannabinoid signaling decreases presynaptic neurotransmitter release and balances glutamate/GABAergic transmission.³⁸ Interestingly, while CB2 receptors are not found in the healthy brain, upregulation of CB2 receptors on microglia appears to be induced by neuroinflammatory processes.³⁸ Therefore, increased endocannabinoid signaling may be associated with anti-inflammatory and neuroprotective phenotypes in microglia that suggests the therapeutic potential of targeting CB1 or CB2 receptors.²³