2型糖尿病患者相干光层析血管成像术图像的量化分析
横断面研究纳入糖尿病组105例(129只眼)与健康对照组27名(32只眼),量化分析OCTA图像。糖尿病组血管密度和血管总长度显著低于对照组,黄斑无血管区面积显著大于对照组,各参数可重复性高,轴率对糖尿病视网膜病变诊断的ROC曲线下面积最高。
要点
- 糖尿病组未分层视网膜血管密度49.146%±6.097%,低于对照组52.212%±6.250%(P<0.05)
- 糖尿病组黄斑无血管区面积(0.441±0.167)mm²,大于对照组(0.352±0.109)mm²(P=0.005)
- 各参数ICC值均>0.965,轴率ROC曲线下面积0.737,非圆系数0.724
核心结论
- OCTA图像量化方法可靠性、可重复性及精确性较好。
- 糖尿病患者血管密度、血管长度及无血管区改变可出现在视力受损之前。
- 无血管区形态改变对糖尿病视网膜病变早期诊断具有指导意义。
研究设计与样本
- 横断面研究,糖尿病组为苏州大学附属理想眼科医院糖尿病视网膜病变防治小组2017年9至12月招募的2型糖尿病视网膜病变患者105人(129只眼),男性47例、女性58例,年龄(61.41±8.44)岁。
- 健康对照组为非糖尿病志愿者27名(32只眼),男性14名、女性13名,年龄(64.13±8.10)岁。
- 两组均行OCTA、眼底照相及最佳矫正视力检查,采用Image-pro plus量化分析OCTA图像,并做ICC和RSD检验及ROC曲线分析。
主要发现
- 糖尿病组未分层视网膜和浅层视网膜血管密度分别为49.146%±6.097%、44.038%±5.641%,均显著低于健康对照组52.212%±6.250%、46.698%±5.417%(t=2.534、2.405;P<0.05)。
- 糖尿病组3层血管图血管总长度分别为(19.905±2.285)、(17.596±2.149)、(14.479±2.091)mm,均显著低于健康对照组(21.037±2.185)、(18.739±1.994)、(15.343±2.266)mm(t=2.529、2.731、2.059;P<0.05)。
- 糖尿病组黄斑无血管区面积为(0.441±0.167)mm²,显著大于健康对照组(0.352±0.109)mm²(t=-2.831;P=0.005)。
- 糖尿病组最佳矫正视力与深层视网膜血管密度(r=-0.197,P=0.026)及3层血管图血管总长度均呈负相关(r=-0.225、-0.201、-0.250;P<0.05)。
- 各参数ICC值均>0.965,RSD均<3.049%,轴率ROC曲线下面积0.737,非圆系数0.724。
关键数值与推荐
- 糖尿病组未分层视网膜血管密度49.146%±6.097% vs 健康对照组52.212%±6.250%(P<0.05)。
- 糖尿病组浅层视网膜血管密度44.038%±5.641% vs 健康对照组46.698%±5.417%(P<0.05)。
- 糖尿病组黄斑无血管区面积(0.441±0.167)mm² vs 健康对照组(0.352±0.109)mm²(P=0.005)。
- 轴率ROC曲线下面积0.737,非圆系数ROC曲线下面积0.724。
- 各参数ICC值均>0.965,RSD均<3.049%。
临床怎么用
- 原文结论指出OCTA图像量化方法可靠性、可重复性及精确性较好。
- 原文结论指出糖尿病患者血管密度、血管长度及无血管区改变可出现在视力受损之前。
- 原文结论指出无血管区形态改变对糖尿病视网膜病变早期诊断具有指导意义。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
Objective: To quantify optical coherence tomographic angiography (OCTA) images and to explore potential clinical values of each parameter in diabetic patients. Methods: A case-control and cross-sectional study of diabetic (129 eyes) and healthy (32 eyes) subjects was conducted. 3-mm OCTA scan, fundus photography and best corrected visual acuity measurement were performed. Image-pro plus was used to calculate microvascular and foveal avascular indices in nonsegmented and segmented OCTA images. Intraclass correlation coefficient and relative standard deviation were used to examine the reliability, reproducibility and accuracy. Correlation of each parameter was calculated, and so was the area under the receiver operating characteristic curve. Results: Vessel density (VD) of diabetic patients measured in the nonsegmented retinal layer (RL) and superficial retinal layer (SRL) was 49.146%±6.097%, 44.038%±5.641%, respectively, significantly lower than healthy subjects 52.212%±6.250%, 46.698%±5.417% (t=2.534, 2.405; P=0.012, 0.017), and vessel length (VL) was (19.905±2.285)mm, (17.596±2.149)mm, (14.479±2.091)mm in the three layers, which was significantly shorter than that in the healthy control group (21.037±2.185)mm, (18.739±1.994)mm, (15.343±2.266)mm (t=2.529, 2.731, 2.059; P=0.012, 0.007, 0.041). Compared with non-diabetes, foveal avascular zone (FAZ) in diabetes was (0.441±0.167)mm(2), larger compared to health eyes 0.352±0.109mm(2) (t=-2.831, P=0.005). FAZ demonstrated a negative correlation with VD and VL in RL or SRL(r=-0.227, -0.338; P<0.05) (r=-0.241, -0.332; P<0.05), while best corrected visual acuity showed a negative correlation with VL in each layer (r=-0.225, -0.201, -0.250; P=0.01, 0.022, 0.004), and VD in DRL (r=-0.197, P=0.026). All parameters showed high reproducibility between graders(intraclass correlation coefficient>0.965), while relative standard deviation was greater than 3.049%. AR had the highest area under the receiver operating characteristic curve (0.737), compared with AI (0.724). Conclusions: Our study demonstrates good reliability, reproducibility and accuracy of quantification of OCTA images and reveals changes of VD, VL and FAZ in diabetes, which may occur earlier than vision impairment. AR and AL may have great values in prompting early diabetic retinopathy in diabetic patients. (Chin J Ophthalmol, 2019, 55:273-279). 目的: 探讨2型糖尿病患者相干光层析血管成像术(OCTA)图像量化效果,以及量化后各参数的变化及临床意义。 方法: 横断面研究。糖尿病组受试者纳入苏州大学附属理想眼科医院糖尿病视网膜病变防治小组于2017年9至12月招募的2型糖尿病视网膜病变患者105人(129只眼),其中男性47例,女性58例,年龄(61.41±8.44)岁;健康对照组为非糖尿病志愿者27名(32只眼),其中男性14名,女性13名,年龄(64.13±8.10)岁。对两组进行OCTA、眼底照相及最佳矫正视力(BCVA)检查。采用Image-pro plus对所有OCTA图像进行量化分析,组间各参数比较采用两独立样本t检验,各参数与BCVA的相关性采用Spearman检验,并进行方法的可靠性可重复性(ICC)和精确性(RSD)检验,通过受试者工作特征(ROC)曲线验证各参数对于糖尿病视网膜病变的诊断效率。 结果: 糖尿病组的血管密度在未分层视网膜及浅层视网膜上分别为49.146%±6.097%,44.038%±5.641%,显著低于健康对照组52.212%±6.250%,46.698%±5.417%(t=2.534,2.405;P<0.05)。糖尿病组的血管总长度在3层血管图中分别为(19.905±2.285)、(17.596±2.149)、(14.479±2.091)mm,均显著低于健康对照组的(21.037±2.185)、(18.739±1.994)、(15.343±2.266)mm(t=2.529,2.731,2.059;P<0.05)。糖尿病组的黄斑无血管区面积为(0.441±0.167)mm(2),显著大于健康对照组(0.352±0.109)mm(2)(t=-2.831;P=0.005)。其他的参数两组间差异均无统计学意义(P>0.05)。糖尿病组BCVA与深层视网膜的血管密度(r=-0.197,P=0.026)及3层血管图的血管总长度均呈负相关(r=-0.225,-0.201,-0.250;P<0.05)。与其余的参数均无明显相关(P>0.05)。各个参数的ICC值均>0.965,而RSD均<3.049%。轴率的ROC曲线下面积(AUC)为0.737,非圆系数的AUC为0.724。 结论: 该OCTA检查结果图像量化方法的可靠性、可重复性及精确性较好。糖尿病患者血管密度、血管长度及无血管区的改变可出现在视力受损之前,而无血管区形态改变对于糖尿病视网膜病变的早期诊断具有指导意义。(中华眼科杂志,2019,55:273-279).
引用
- Zhu QJ, Liang J, Xu CL, et al. [Quantification of optical coherence tomographic angiography images in type 2 diabetic patients]. Zhonghua Yan Ke Za Zhi. 2019. PMID: 30982289. doi:10.3760/cma.j.issn.0412-4081.2019.04.009
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。