Patient demographics, knee function and radiological assessment
In total, 83 patients including 21 males and 62 females were enrolled in the study. The average age was 63.4±10.4 years. The average WOMAC score was 25.5 (21.7-29.3), while the average ROM was 112.0° (95% CI 108.0-116.1). The patient demographics, knee function, pain intensity and expression of inflammatory mediators in different K-L grade groups are illustrated in Table 1. With increasing age, ROM became more limited (Figure 1A), and the WOMAC score increased (Figure 1B), moreover, worsening K-L grade also led to limited ROM (r=-0.499, p<0.001) and increased WOMAC score (r=0.493, p<0.001) (Figure 2A and Figure 2B), which indicated that knee function was lost gradually.
To explore the differences in inflammatory mediators among different K-L groups, the expression of inflammatory mediators was compared using one-way ANOVA. We found the expression of IL-1β (p=0.002) and IL-6 (p=0.023) was significantly different among four K-L groups (Figure 3A and Figure 3B). Increased expression of IL-1β and IL-6 was found in the early stage of KOA, which suggested that more severe inflammation may exist in the early stage of KOA. However, no significant differences were found for other inflammatory mediators among the K-L groups.
Significant correlations among different pain scoring systems
To evaluate the intensity of pain in KOA, nociceptive pain was measured using the NRS and VAS, while neuropathic pain was determined using the PainDETECT questionnaire. The WOMAC pain score was recorded to measure function-related pain. The NRS and VAS scores were 3.04 (95% CI 2.63-3.44) and 3.55 (95% CI 3.09-4.03) cm, respectively. The WOMAC pain score was 5.05 (95% CI 4.29-5.81), while the PainDETECT pain score was 2.89 (95% CI 2.12-3.67).
To further investigate the consistency among different pain scoring systems, correlation coefficients were determined using the Spearman test. As expected, the four measurements of pain were highly correlated with each other and exhibited homogeneity (Figure 4A and Figure 4B).
Pain affected knee function and correlated to age and K-L grade
To determine the correlations between pain and patient demographic parameters, knee function or radiological grade, pain was analyzed for correlations with age, ROM, WOMAC score and K-L grade using the Spearman test. As expected, with an increase in age or K-L grade, the NRS, VAS and WOMAC pain scores increased (Figure 5A and Figure 5B). Correspondingly, knee function was gradually lost, as shown by an increase in the WOMAC score (Figure 5C) and limited ROM (Figure 5D). However, neuropathic pain scores correlated to only the WOMAC score (r=0.384, p<0.000) and not to age (r=0.108, p=0.331), K-L grade (r=0.176, p=0.112) or ROM (r=-0.179, p=0.105). In conclusion, age and K-L grade were important factors correlated to KOA pain, and KOA pain significantly affected knee function.
Correlations among pain, patient characteristics and inflammatory cytokines
Inflammatory cytokines play critical roles in pain initiation and development (5). Among inflammatory cytokines, IL-1β, IL-6 and TNF-α are the most important inflammatory cytokines involved in KOA. Even though IL-1β, IL-6 and TNF-α have been widely investigated in both experiment models and clinical samples, the expression of these cytokines in KOA pain is still controversial, and different research studies have reached conflicting conclusions (7, 18, 19). Moreover, as illustrated in the introduction, M-CSF has been found to be important in KOA (10, 11). However, whether SF M-CSF is involved in KOA pain is still unknown.
To investigate the correlations between pain and these inflammatory cytokines, the expression of IL-1β, IL-6, TNF-α and M-CSF was measured by ELISA and analyzed by the Spearman test. As is shown in Table 2, IL-1β and IL-6 were very weakly negatively correlated to VAS score, and TNF-α was negatively correlated to VAS and NRS scores. However, M-CSF did not show any correlation to any of the pain scores. Moreover, WOMAC-assessed pain and neuropathic pain were not correlated to any of the inflammatory cytokines. In short, the above evidence demonstrated that inflammatory cytokines significantly correlated to KOA pain. The expression of IL-1β, IL-6 and TNF-α may be higher in patients with less pain.
To further determine the correlations between the expression of inflammatory cytokines and patient characteristics, the expression of IL-1β, IL-6, TNF-α and M-CSF was analyzed for correlations with age, K-L grade, ROM and WOMAC score using the Spearman test (Table 3). We found IL-1β and IL-6 were negatively correlated to age, K-L grade and WOMAC score, while IL-6 was positively correlated to ROM. In addition, TNF-α was negatively correlated to age. In line with the results shown in Figure 3, the expression of IL-1 and IL-6 was relatively high in the early stage compared with the late stage, and the expression of these two inflammatory cytokines was associated with knee function.
No correlations were found between pain and measured catabolic proteases
OA progression is usually accompanied by high matrix catabolism, therefore, we hypothesized that the high expression of catabolic proteases in the SF enhances the breakdown of cartilage and causes severe synovitis and inflammation, which may lead to more pain. Among these catabolic proteases, MMPs and ADAMTS5 play important roles in KOA through involvement in the degradation of ECM type II collagen and aggrecan, respectively (20, 21). However, whether these proteases are involved in KOA pain is still unknown.
To evaluate the correlations between these catabolic proteases and pain, MMP-3, MMP-13 and ADAMTS5 levels were measured by ELISA and correlated with pain scores using the Spearman test (Table 4). The results showed that only MMP-13 was weakly correlated to NRS score (r=0.267, p=0.015). However, no correlations were found between pain scores and other measured catabolic enzymes. This result demonstrated that the measured catabolic cytokines may not be involved in KOA pain.
No correlations were found between pain and measured neuropeptides
It is believed that neuropeptides sensitize knee nociceptors and are related to knee pain (19). SP, NPY, CGRP and BK are considered important neuropeptides involved in OA pain. First, the SF expression of SP was previously found in KOA and rheumatoid arthritis (22). The concentration of SP is significantly higher in painful temporomandibular joints than in painless temporomandibular joints (23). Second, the expression of NPY in dorsal root ganglion was found to be significantly correlated to OA pain in animal models (24), and SF NPY expression was found to be correlated to OA pain (25). Third, CGRP was reported to be involved in the progression and prognosis of KOA and widely distributed in the nociceptive pathways in the peripheral and central nervous systems, which correlated to numerous types of pain (26-28). Last but not least, BK is an inflammatory mediator that can lead to vasodilation and inflammation induction. Although BK has been reported to correlate with biomarkers for cartilage degradation and inflammation in OA (29), the role of BK in pain still needs to be further investigated. Therefore, it is necessary to explore whether these neuropeptides play roles in OA pain.
To address this issue, the expression of these neuropeptides was analyzed and correlated with different pain scoring systems using the Spearman test (Table 5). However, no correlations were found between any of the measured pain scoring systems and these neuropeptides, which suggests that these neuropeptides may not contribute to KOA pain.
Interactions and correlations among inflammatory mediators
To investigate the associations among the measured inflammatory mediators, correlation coefficients were defined among all of the inflammatory mediators using the Spearman test. The outcome is shown in Figure 6. On the one hand, all the types of inflammatory mediators had internal associations. Among the inflammatory cytokines, IL-1β was highly positively correlated to IL-6 (r=0.93, p<0.000), while TNF-α was negatively correlated to M-CSF (r=-0.362, p=0.004). Interestingly, of the measured matrix-catabolizing proteases, only MMP-3 and ADAMTS5 showed a very weak correlation (r=0.239, p=0.030), and no other correlations were found among the matrix proteases. To our surprise, almost all of the measured neuropeptides were correlated to each other, which demonstrated the obvious interactions among these neuropeptides. On the other hand, correlations were found among different types of inflammatory mediators. For example, TNF-α and M-CSF were correlated to all of the measured neuropeptides, while MMP-3 and MMP-13 were correlated to only SP. These results demonstrated that inflammatory cytokines, catabolic proteases and neuropeptides interacted with each other.