In this study, we show that RVO patients had significantly higher plasma and aqueous levels of complement proteins compared with cataract patients. Six (C4, C4b, C3b/iC3b, CFB, CFI, and CFH) out of 13 complement proteins in the plasm were significantly higher in RVO patients compared to the controls. In aqueous humor, apart from C3, all other complement proteins were significantly higher in RVO patients. In our study, we excluded participants with inflammatory/autoimmune diseases and those who were taking immunosuppressant medications. RVO patients were younger than cataract patients, but the differences in complement proteins remained after adjusting for age. Our results suggest that complement activation may be critically involved in RVO development and RVO-mediated retinal pathologies.
The occlusion of the retinal vein/venules in RVO is due to the formation of thrombosis. Although the mechanism underlying abnormal thrombosis in RVO remains elusive, the complement and coagulation pathways are closely related. There are multiple cross-talks between the components of the two cascades. For example, plasma kallikrein can affect the generation of C3 and C5 fragments directly or indirectly[12]. C5a can be generated by thrombin independent of C3[13]. On the other hand, complements can activate the coagulation cascade directly or indirectly. The MASP2, a component of the MBL pathway, is critically involved in the activation of thrombin and subsequent generation of the fibrin mesh[14]. Sublytic C5b-9 can cause transient membrane depolarization, granule secretion, and induction of platelet-catalyzed thrombin generation and clotting[15, 16].
Although C5a was below detectable limit in 16 out of 20 RVO patients and this does not support systemic complement activation, the higher levels of C4, C4b, C3b/iC3b, CFB, CFI, and CFH in the plasma of RVO patients are indicatives of abnormal systemic complement activities.
Twelve out of 13 complement proteins in the aqueous humor were significantly higher in RVO patients compared to controls. The complement levels in the plasma were 10-1000 times higher than those in the aqueous humor in our study. Theoretically, circulating complement proteins can leak into the retina from the diseased vessels and accumulate in the intraocular compartments. Surprisingly, we detected positive correlations between plasma and aqueous levels of five complement proteins (C4, C4b, C5, CFD, and CFI, r = 0.57 ~ 0.78) in cataract patients, but only one (i.e., CFI, r = 0.5) in RVO patients. Our results suggest that intraocular complement proteins may be somehow, related to their counterparts in the blood circulation under normal physiological conditions, but in RVO, they are independent of their circulating counterparts and may be generated locally within the eye. In other words, intraocular complement production/activation is likely an active response of the retina to ischemic injury in RVO. Retina has a higher level of control over its immune response to injury[17]. The complement system forms an important arm of retinal innate immune protection. Retinal cells, including neurons, microglia[18], and RPE cells[19–23] express various complement genes and their expression is increased under inflammatory conditions[21]. Therefore, it is not surprising to see higher levels of complement activation in the RVO retina. In line with our study, C3, C5 and CFH were detected in the aqueous humor from RVO patients using proteomic analysis by others[11, 24].
The complement system can be activated through the CP, AP, and MBL pathways that the cleavages of C3 and C5 are two key milestone cascades. The intraocular level of C3b/iC3b positively correlated with C1q, CFI, CFH, and MBL, indicating that all three pathways are involved in the cleavage of C3 in RVO retina. On the other hand, the level of C5a positively correlated with C2, C4, CFH and CFB in the aqueous humor of RVO patients. The levels of C2 and C4 were strongly correlated with CFD, CFI, CFH and CFB (Table 4). Our results suggest that intraocular complement activation in RVO is likely mediated through the CP and supported by the AP through amplifying the CP-mediated activation cascade. It is worth noting that C3b/iC3b but not C5a was detected in the aqueous humor of cataract patients and the intraocular level of C5a did not correlate with C3b/iC3b in RVO patients, suggesting that C3 cleavage does not necessarily lead to C5 cleavage inside the eye. Furthermore, C5a can be generated by thrombin independent of C3 [13]. Multiple pathways may be involved in the breakdown of C5 in RVO.
Dysregulated complement activation can lead to pathologies. We found that aqueous levels of CFI and MBL positively correlated with CRT. A positive correlation was also observed between the aqueous level of MBL and the ILM-IPL thickness. Furthermore, the aqueous levels of C2, CFB, CFH, and MBL were negatively correlated with the size of the foveal avascular zone (FAZ). FAZ is the most sensitive central area of vision, and changes in its shape and size can pose a threat to vision. Progressive and irregular expansion of FAZ has been observed in RVO eyes[25], and changes in FAZ is related to capillary remodeling in the macular area[26]. Our results suggest that intraocular complement activation may be involved in retinal oedema and microvascular remodeling in RVO. Since we did not detect any relationship between C3b/iC3b, C5a (indicators of complement activation) and retinal OCTA parameters, intraocular complement activation per se is unlikely a contributor of macular oedema and vascular degeneration in RVO. Exactly how individual complement proteins are involved in retinal pathologies in RVO remains to be elucidated.
The strengths of the study include (1) the simultaneous measurement of complement proteins in the blood and aqueous humor; and (2) comprehensive clinical and laboratory evaluations of the same participants; (3) participants did not receive any medication (systemic or local) prior to the study. The study has several limitations. Firstly, the number of participants enrolled in this study was relatively small. This is because the incidence of RVO is low, and it is extremely difficult to recruit treatment naïve RVO patients to the study. Secondly, the study measured complement proteins but did not test complement activity in the blood and aqueous humor. Third, the study was conducted in a single centre and the results can only reflect the biological features of RVO in the local ethical population. Replication of the study findings with a larger sample size and in multiple ethnic groups is necessary to confirm our results. However, it should be noted that single centre study reduces procedure-related variation and increases the reliability of the results in small sample size studies.