We also identified a peak in the Chr p next

We also identified a peak in the Chr8p23.1 next to HMGB1P46 when analysing the male only dataset, and the P value of the top SNP rs6986153 was 8.02×10 with an OR of 1.67. HMGB1P46 is a pseudogene of high mobility group box-1 (HMGB1). It is suggested that the induction of high mobility group box-1 in the dorsal root ganglion can contribute to pain hypersensitivity after peripheral nerve injury (Shibasaki et al., 2010). In addition, Feldman et al found that the persistent endogenous release of HMGB1 by sensory neurons contributes to tactile hyperalgesia in a neuropathic pain rat model (Feldman et al., 2012). The synthesis and release of HMGB1 from spinal neurons due to nerve injury facilitates the activity of both microglia and neurons which leads to symptoms of neuropathic pain (Nakamura et al., 2013). It is interesting to know that HMGB1 signalling and TLR pathways, to some extent, are overlapping together (Yu et al., 2006; Velegraki et al., 2012). There is evidence that pseudogenes are involved in the biological process. For example, the low level of high mobility group A1 (HMGA1) was also associated with a high level of HMGA1 pseudogene (HMGA1-p) mRNA (Chiefari et al., 2010). It was observed that knockdown of HMGA1-p RNA in the strontium chloride of diabetic patients led to partially restored HMGA1 mRNA levels which suggested a competing relationship between the two types of transcripts. It is therefore hypothesised that a competing relationship might also exist between HMGB1 and its pseudogenes. There were no SNPs found with a P value of less than 5×10 in the overall dataset, male only or female only datasets. Although a P value of 5×10 is generally adopted as the cut-off P value for GWAS significance, it has been suggested that this might be too stringent and risks missing important associations (i.e. false negatives) (Do et al., 2014). Using a lower threshold raises the chance of detecting associated SNPs, but also of detecting spurious associations (false positives), and we need to beware of that in interpreting this study. The narrow-sense heritability (variance explained by SNPs, excluding genetic variation due to dominance, epistasis, and environment) of diabetic neuropathic pain in the overall dataset was estimated to be 14.7%, which is similar to that found in our previous analysis (Meng et al., 2015). However when calculated by gender, we found males had a higher heritability (30.0%) than females (14.7%). Sex-specific heritability has been observed in other traits (Weiss et al., 2005). The reasons behind the different gender-specific heritabilities are unknown although it may result from parent-of-origin effects, interaction with sex chromosomes and the sex-specific hormonal environment. It is worth considering sex-specific genetic effects in future association studies of neuropathic pain. There are some reports indicating that genetic effects are different between genders in determining pain. Experiments in mice found that the Mc1r gene mediates kappa-opioid analgesia in female mice only. Correspondingly in a human study, females with two variant MC1R alleles showed greater analgesic responses from the kappa-opioid, pentazocine, than males and females who did not have the variant alleles (Mogil et al., 2003). In addition, polymorphisms in the OPRM1 gene have been reported to be associated with pressure-related pain sensitivity in men but not in women (Fillingim et al., 2005). Sato et al found that there were significant associations between the opioid receptor genes (OPRM1, OPRD1 and opioid OPRK1) and experimental pain sensitivity (Sato et al., 2013). Our results showing genetic differences associated with neuropathic pain between genders are consistent with these findings, though the biological mechanisms remain unclear and highlight the need for further research in this area. The heritability of neuropathic pain has been calculated as around 30% in rat models (Devor et al., 2005), similar to that measured here among men, though twice that found among women..