In the present study, neuropathic pain was detected in 52.9 % of patients based on the DN4 questionnaire and in 38.6 % of patients based on the LANSS scale. According to DN4 questionnaire, the most frequently described neuropathic pain characteristic in KOA patients with neuropathic pain was a sensation of electric shock (48.7%) with a significant difference compared to KOA patients with non-neuropathic pain, and on physical examination up to 30% had touch hypoesthesia, while according to LANSS scale the most frequently knee pain quality symptom was pins and needles tingling or pricking (35.7%) followed by burning pain (34.3%) with a significant difference compared to KOA patients with non-neuropathic pain; however self-exam items revealed that 24.3% of patients had numbness or tenderness which felt when pressing on the painful area with a significant difference compared to patients with non-neuropathic pain and 14.3% had pins and needles, tingling or burning on rubbing on the painful area.
In agreement with our results, several studies reported the frequency of neuropathic pain in patients with KOA which was ranged from 17.6 to 51.9% (based on the DN4 questionnaire and/or LANSS scale) [26,27,28,29,30,31,32,33,34]. Other studies reported neuropathic pain in OA patients (based on the painDETECT questionnaire) in frequencies ranged from 20.7 to 66.7% [31, 35,36,37].
Aşkın et al.  refereed the wide variation in the reported prevalence of neuropathic pain in OA to differences in methodology between studies and pain assessment tools.
Our results revealed that KOA patients with neuropathic pain as detected by DN4 questionnaire have significantly higher WOMAC pain, WOMAC physical function scores, and significantly lower normalized pain OAKHQOL score than patients with non-neuropathic pain (P <0.0001, P= 0.015, P =0.018 respectively). Moreover, significantly higher Lequesne pain score and Lequesne index were found in KOA patients with neuropathic pain as detected by LANSS scale compared with patients without neuropathic pain (P=0.036, P=0.022 respectively).
In consistent with our results, Gölge et al.  found a highly significant difference between neuropathic and non-neuropathic groups as regards WOMAC pain score (p <0.001), Aşkın et al.  also found a significant difference between both groups as regards WOMAC physical function (p = 0.04), Narayan et al.  found a significant difference between both groups as regards WOMAC total score (p =0.024) and WOMAC physical function score (p = 0.008), Mahmoud et al.  study showed that the total score and normalized pain domain score were worst in the neuropathic group than the non-neuropathic group, and Radwan and Borai  found a highly significant difference between neuropathic and non-neuropathic pain groups as regards WOMAC physical function score, WOMAC pain score, and WOMAC total score (p < 0.001 for all of them).
We assessed serum β-NGF levels in the present study and we have found a significantly higher serum β-NGF levels in KOA patients than controls (P<0.001), and significantly higher serum β-NGF levels in KOA patients with neuropathic pain “as detected by DN4 questionnaire and LANSS scale” compared with patients without neuropathic pain (P= 0.01, P= 0.004 respectively). In agreement with our results, a study of Montagnoli et al.  reported significantly higher serum and synovial β-NGF levels in KOA patients than controls. To our knowledge, no previous studies in the literature assessed serum β-NGF levels in KOA patients with neuropathic pain versus patients without neuropathic pain.
In the present study, the correlations of neuropathic pain scores with sociodemographic data, physical function, quality of life, and disease severity were investigated and we found that DN4 score was positively correlated with WOMAC pain, WOMAC stiffness, and WOMAC physical function and negatively correlated with OAKHQOL pain scores (rs=0.459, P<0.001; rs= 0.258, P= 0.031; rs= 0.307, P= 0.010; rs = −0.337, P= 0.004 respectively), while LANSS scale was positively correlated with symptom duration, WOMAC stiffness, Lequesne pain, and Lequesne index (rs= 0.260, P= 0.020; rs= 0.343, P= 0.004; rs= 0.344, P= 0.004; rs= 0.322, P= 0.007) and negatively correlated with OAKHQOL physical, OAKHQOL mental health, OAKHQOL social support, and total OAKHQOL scores (rs= −0.258, P= 0.031; rs= −0.254, P= 0.034; rs= −0.283, P= 0.018; rs= − 0.261, P= 0.029 respectively). In this way, we consider that KOA patients with neuropathic pain may have longer symptom duration, severe pain, extreme disability, and worse quality of life than patients with non-neuropathic pain.
A number of evidences indicate that β-NGF plays a significant role in osteoarthritis, not only in pain and hyperalgesia by nociceptor sensitization, but also as a key element of the inflammatory process [40, 41]. Neuropathic pain is unresponsive to common analgesics, such as NSAIDs. Systemic central acting drugs as duloxetine, an antidepressant, have proven effective in controlling this type of pain in OA . Agents blocking NGF might have therapeutic utility for pain .
Female sex, age, and BMI are well-known risk factors for OA, as shown in previous studies [44,45,46]. Also, a low level of education was found to be a significant factor associated with OA . In the present study, we have found no correlation between neuropathic pain scores and risk factors. Study of Polat et al.  was consistent with our findings. In contrary to our results, Hochman et al.  found that the patients with neuropathic pain were younger and were more likely to be females, but there was no significant difference in level of education between patients with neuropathic pain and those without neuropathic pain.
Our results revealed no correlation between neuropathic pain scores and the Kellgren–Lawrence grades and this was consistent with the previous results of Narayan et al. , Polat et al. , and Radwan and Borai . In contrary to our results, Ohtori et al.  reported that neuropathic pain tended to be seen in patients with KL grades of late stages OA; however, the majority of our patients had mild to moderate OA.
In the present study, according to linear regression analysis, longer symptom duration, higher pain scores, lower OAKHQOL score, and higher serum βNGF levels were considered as significant risk factors for the development of neuropathic pain in KOA patients.
Our study has some limitations. Firstly, this study was cross-sectional with a relatively small number of patients. Secondly, we have not assessed for central sensitization by measuring pain pressure thresholds. Lastly, we did not assess β-NGF levels in synovial fluid and/or synovial tissue.