1710

Towards specific biomarkers of chronic pain: Modelling behavior readouts of developing pain to resting-state functional connectivity using a developmental trajectory in a mouse model of chronic pain from bone cancer
David Buehlmann1,2, Giovanna Diletta Ielacqua1, Joanes Grandjean3, Jael Xandry4, and Markus Rudin1,4

1Inst. for Biomedical Engineering, ETH & University of Zurich, Zurich, Switzerland, 2Neuroscience Center Zurich, Zurich, Switzerland, 3Singapore Bioimaging Consortium, Singapore, Singapore, 4Inst. for Pharmacology & Toxicology, University of Zurich, Zurich, Switzerland

Synopsis

We performed longitudinal behavioral readouts of pain and resting-state fMRI in a mouse model of chronic pain from breast cancer derived tibial bone metastases. The developmental trajectory of behavioral readouts was used to model the fMRI response to extract pain-specific functional connectivity changes during development of a chronic pain state. The specificity of these functional readouts was supported through inhibition of osteolytic activity, reducing the nociceptive input. Thereby, characteristic functional changes could be reduced and in some regions prevented. These results emphasize the specificity of the functional readouts for developing chronic pain and could be used to evaluate novel treatments.

Purpose

Chronic pain affects 15% of the world’s population and poses a major burden on patients’ lives due to insufficient treatment options1. Development of novel treatments has been largely unsuccessful due to poor translatability of preclinical findings, typically based on behavioral readouts of pain. They probe simple reflexes or innate responses, which are a result of sub-spinal processing rather than actual pain generated in the brain2. This could be problematic in particular in chronic pain conditions where emotional and affective components were found to be main contributors in the chronification of pain3. We therefore performed longitudinal analysis of functional connectivity based on resting-state fMRI to find potential signatures of developing chronic pain in the brain, which might serve for evaluating novel treatments. Specificity of these readouts was tested by modulating the nociceptive input using peripherally active treatments.

Material & Methods

Mouse models of chronic pain from bone cancer were prepared by intramedullar injection of EO771 (tebu-bio) breast cancer cells into the tibia of female C57BL/6 (n=10). Functional changes in the brain were assessed longitudinally using a Bruker Biospec 94/30 small animal MR system with a four-element receive only cryogenic phased array coil (Bruker BioSpin AG) with a linearly polarized room temperature volume resonator for transmission. Resting-state functional MRI (rs-fMRI) data was acquired using a gradient-echo echo-planar imaging (GE-EPI) sequence: TR=1000ms, TE=12ms, FA=60°. Ex-vivo digital radiographs were acquired (35kW/10s) using a planar x-ray system (MX20, Faxitron). Animals were anesthetized using a combination of medetomidine-hydrochloride (0.05mg/kg as bolus, 0.1mg/kg/h maintenance, s.c.), intubated and ventilated with low dose isoflurane (0.5%) in 20%O2 / 80%air mixture at 80breaths/min. For immobilization, pancuronium bromide was administered s.c. in a bolus of 0.5mg/kg. Behavioral readouts of pain were assessed in terms of time spent guarding and number of flinches in a 2min period. Rs-fMRI data was analyzed using independent component analysis (ICA). Longitudinal evolution of obtained components was modelled on a developmental trajectory obtained from behavioral readouts of pain. Statistical differences between groups were then tested using a general linear model with non-parametric permutation testing, accounting for multiple comparisons with threshold-free cluster enhancement and Bonferroni correction. Treatment with zoledronic acid (Zoledronate Onco Sandoz, Sandoz) was administered prospectively 7days post tumor injection (10mg/kg, s.c.).

Results

Nociceptive behavior was significantly increased 14 and 21days post injection in tumor-bearing compared to sham treated animals (Fig.1). Both guarding and flinching indicated similar longitudinal evolution of nociceptive behavior, supporting the trajectory used for modelling the fMRI response. Most robust functional changes were found in the cingulate (p=0.0004) and prefrontal/anterior cingulate cortex (p=0.0006). Additionally, piriform (p=0.0008) and temporal associative cortices (p=0.0002) were found to be affected as well as the ventral striatum (p=0.003). Less robust changes were observed in the ventral (p=0.009) and dorsal (p=0.033) hippocampus (Fig.2). Prospective treatment with zoledronic acid has been shown to effectively reduce osteolysis4 and efficacy could be confirmed in ex-vivo digital radiographs (Fig.3). Tibiae of treated mice indicated only minor signs of osteolysis, which were highly abundant in vehicle-treated animals. Reduced peripheral input significantly modulated the central readouts described before (Fig.4). Within-network connectivity in the cingulate (p=0.0053) and prefrontal/anterior cingulate (p=0.012) cortex and ventral striatum (p=0.0022) was found significantly reduced but still remained above statistical significance although it did not survive Bonferroni correction. Changes in the piriform (p=0.39), temporal associative (p=0.24), dorsal (p=0.15) and ventral (p=0.071) hippocampus did not reach significance after treatment (Fig.5).

Discussion

Longitudinal rs-fMRI data was modelled on a trajectory obtained from behavioral readouts of pain to find regions specific to developing chronic pain. Several regions were found to be affected, following the inverse developmental trajectory of behavioral readouts. The observed functional alterations in brain connectivity were found highly robust and reproducible with a maximum effect size of 30%. These functional changes could be modulated by peripherally active therapy, supporting the specificity of the presented readouts. Complete inhibition of central effects was not achieved, since only one aspect of pain was targeted in the applied treatment regime. The maximum effect-size of the treatment in the prefrontal/anterior cingulate cortex of 20% is comparable to clinical data of patients suffering from metastatic bone cancer treated with zoledronic acid5.

Conclusion

By using behavioral readouts of pain we found direct functional correlates of developing chronic pain in the brain. These readouts are modular through application of treatments, therefore seem to be specific. Given the reproducibility and effect-size, these readouts could be used to evaluate novel treatments. Since pain is ultimately generated in the brain, these central readouts might be much more specific than nocifensive behavioral readouts, which potentially enhances translatability of preclinical pain research.

Acknowledgements

This study was supported by the Swiss National Science Foundation (grant SNF 160310, MR).

References

1. Murray CJL, Lopez AD. Measuring the Global Burden of Disease. N Engl J Med. 2013;369(5):448-457. doi:10.1056/NEJMra1201534.

2. Mogil JS. Animal models of pain: progress and challenges. Nat Rev Neurosci. 2009;10(4):283-294. doi:10.1038/nrn2606.

3. Baliki MN, Apkarian A V. Nociception, Pain, Negative Moods, and Behavior Selection. Neuron. 2015;87(3):474-491. doi:10.1016/j.neuron.2015.06.005.

4. Casanova M, Herelle J, Thomas M, et al. Effect of combined treatment with zoledronic acid and parathyroid hormone on mouse bone callus structure and composition. Bone. 2016;92:70-78. doi:10.1016/j.bone.2016.08.012.

5. Saad F, Gleason DM, Murray R, et al. Long-Term Efficacy of Zoledronic Acid for the Prevention of Skeletal Complications in Patients With Metastatic Hormone-Refractory Prostate Cancer. JNCI J Natl Cancer Inst. 2004;96(11):879-882. doi:10.1093/jnci/djh141.

Figures

Nocifensive behavior was assessed as number of flinches and time spent guarding in a 2min period. Guarding was defined as holding the affected paw aloft while ambulatory, while flinching was defined as holding the affected paw aloft while not ambulatory. Significantly increased behavioral signs of pain were observed at 14 and 21 d.p.i. in tumor bearing, compared to sham treated animals. Statistical analysis was performed using multiple t-tests (* p > 0.05; ** p > 0.01; *** p > 0.001).

Independent functional components affected during development of a chronic pain state (Pir Piriform cortex; TAc Temporal associative cortex; Pfc Prefrontal cortex; Cg Cingulate cortex; vStr ventral Striatum; v/dHp ventral/dorsal Hippocampus). Longitudinal subject-level spatial maps after dual regression were modeled on a evolutionary trajectory based on behavioral readouts of pain. The presented independent components are following the inverse trajectory in tumor-bearing animals, but remain constant in controls. Statistical analysis was performed using non-parametric permutation testing, accounting for multiple comparisons using threshold-free cluster enhancement and Bonferroni correction. P-values are indicated as heatmaps before (red - yellow) and after Bonferroni correction (blue - light blue).

Ex-vivo digital radiographs of tumor-bearing limbs 22 d.p.i. in untreated (vehicle) and treated (zoledronic acid) animals. Yellow arrows indicate regions affected by osteolysis. These regions of decreased bone-density are much more apparent in vehicle-treated animals compared to zoledronic acid-treated animals, indicating treatment efficacy.

Independent functional components affected during development of a chronic pain state in tumor-bearing animals treated with zoledronic acid compared to controls (Pfc Prefrontal cortex; Cg Cingulate cortex; vStr ventral Striatum). Functional changes were still observed in the Pfc, Cg and vStr, although much less robust. Changes in remaining components previously described to be affected in untreated, tumor-bearing animals could be prevented. Statistical analysis was performed using non-parametric permutation testing, accounting for multiple comparisons using threshold-free cluster enhancement and Bonferroni correction. P-values are indicated as heatmaps (red - yellow). None of the components survived Bonferroni correction.

Parameter estimates after dual regression of most robustly affected components. β-value estimation reveals robust changes in tumor-bearing compared to sham-treated animals, indicating a very similar evolutionary profile as observed in behavioral readouts of pain. These changes could be rescued almost to a control-level after application of peripheral treatment. Statistical analysis was performed using 2way ANOVA, accounting for multiple comparisons (* p > 0.05; ** p > 0.01; *** p > 0.001).

Proc. Intl. Soc. Mag. Reson. Med. 25 (2017)
1710