br loss increased facture risk and increased
loss, increased facture risk and increased risk of developing multiple myeloma (Jorgensen et al., 2015). Our data does not suggest that the exacerbated pain-behavior should be driven by an increased tumor load in AFC5261-treated animals, as no diﬀerence was found between any of the groups. In vivo bioluminescence is only directly correlated to the number of living Calcein-AM until the tumor has reached a certain size (Klerk et al., 2007), after which the signal tends to stagnate, most likely due to a combination of quenching of the signal by increased optical density of the tumor mass, increasing necrosis in the middle of the tumor and limited availability of substrate in the core of the tumor. In this model, we typically see a stagnation of the signal from around day 14 (Appel et al., 2017; Falk et al., 2015a), and for these reasons the biolumines-cence signal was only statistically evaluated until day 14.
Interestingly, Huang et al. have previously demonstrated that chronic inhibition of the P2X7R by small-interference RNA or daily injection of Brilliant Blue G (BBG) into the rostral ventromedial medulla (RVM) has an anti-nociceptive eﬀect in the Walker256 rat model of cancer-induced bone pain (Huang et al., 2014). The same group has later demonstrated that a similar approach targeting the spinal cord by intrathecal injections also had anti-nociceptive eﬀect (Yang et al., 2015). In both studies, the Walker256 mammary carcinoma cell line was used, in Sprague-Dawley rats and Wistar rats, respectively. One reason for the discrepancy between their studies and the data presented in the study could therefore be linked to the diﬀerent glia response observed in models of cancer-induced bone pain. Whereas it is
generally well-established that the Walker256-induced models presents with robust glia response in the spinal cord (Lan et al., 2010; Liu et al., 2017; Mao-Ying et al., 2012; Wang et al., 2012), the result for the MRMT1-induced models used in this study have been more conflicting (Dore-Savard et al., 2010; Ducourneau et al., 2014). In line with the study by Ducourneau et al., we do not observe glia activation in the MRMT1-model (unpublished data). The diﬀerence in glia response in the models might therefore add to the inconsistency between the stu-dies. Another explanation for the opposing eﬀect could be related to the type of pain modality tested. Whereas some studies have adapted pain assessment tool from the fields of neuropathic and inflammatory pain, and hence relies on assessment of mechanical and thermal paw hy-persensitivity, other have focused more on measures that more directly asses bone-related pain. We and others have previously proposed that quantification of evoked paw hypersensitivity is not an optimal mea-sure for cancer-induced bone pain, as it is not necessarily directly linked to the severity of the bone pain (Falk et al., 2015b; Guedon et al., 2016). Guedo et al. recently made an elegant study, demonstrating that the mechanical paw hypersensitivity observed in the NTCT mouse model of cancer-induced bone pain can be blocked without aﬀecting the bone pain component, measured as weight-bearing and spontaneous noci-fensive behavior (Guedon et al., 2016). This is in line with our previous observation in the MRMT1 rat model, where we found that some ani-mals develop a numbness in the paw in the late phase (Falk et al., 2015b) making assessment of paw hypersensitivity far from optimal in this model. A part of the explanation for the conflicting results could therefore be caused by a more pronounced anti-nociceptive eﬀect on paw hypersensitivity, and a smaller, if any, eﬀect on weight-bearing and limb use. The inconsistency among the diﬀerent studies with chronic P2X7R inhibition could also be explained by diﬀerent sites of action. Huang et al. and Yang et al. induced the inhibition of the P2X7R directly in the central nervous system and hence evaluated the local eﬀect of P2X7R inhibition in the RVM or the lumbar segment of the spinal cord. This approach is unlikely to influence the bone environ-ment around the tumor and the peripheral nervous system, and the inconsistency might therefore in addition be aﬀected by local eﬀects vs. systemic eﬀects. Lastly, as the P2X7R is highly polymorphic and various splice-variant are expressed in vivo, the variations reported might re-flect diﬀerences in splice-variant specificity and/or diﬀerences in the PK profiles of the diﬀerent compounds tested.