比较扫频源光学相干断层扫描(OCT)和光学相干断层扫描血管成像(OCTA)获取的视盘周围与黄斑区结构及血流参数在高度近视眼(HM)合并开角型青光眼(OAG)中的诊断效能，并构建精简联合诊断模型。

选取2022年6月至2024年12月在北京大学第三医院眼科就诊并完成相关检查的HM患者239例(239只眼)作为研究对象。其中，男性97例(97只眼)，女性142例(142只眼)；年龄32~68岁，平均(53.1±8.6)岁。依据是否合并OAG，分为HM组与HM合并OAG组(HMG组)。记录年龄、性别、等效球镜屈光度、眼轴(AL)长度及视野指标。使用OCT和OCTA获取视盘周围视网膜神经纤维层(RNFL)厚度、放射状盘周毛细血管(RPC)密度、黄斑神经节细胞复合体(GCC)厚度、GCC损失体积指标、浅层毛细血管丛(SCP)血管密度以及深层毛细血管丛(DCP)血管密度等参数。年龄、等效球镜屈光度、AL、视野指标、RNFL厚度、RPC密度、黄斑GCC厚度、GCC损失体积指标、SCP毛细血管密度(CVD)及DCP的CVD等经Shapiro-Wilk正态检验符合正态分布，以±s表示，两组比较采用独立样本t检验，分区结构指标和血流指标采用Benjamini-Hochberg法进行假发现率校正；性别变量以例数和百分比表示，采用χ²检验。采用受试者工作特征曲线分析计算曲线下面积(AUC)及95%置信区间(CI)，AUC差异采用DeLong检验。选取AUC具有显著统计学意义的指标进入多因素逻辑回归模型，采用多因素Logistic回归分析构建精简联合诊断模型。采用方差膨胀因子评估多重共线性，方差膨胀因子<5为可接受。

HM组117例(117只眼)和HMG组122例(122只眼)在年龄、AL长度、等效球镜屈光度及中央角膜厚度方面比较的差异均无统计学意义(t＝-1.63，-1.50，0.97，0.8；P>0.05)，HMG组和HM组视野平均偏差和模式标准差组间比较差异有统计学意义(t＝15.3，-11.9；P<0.05)。HMG组和HM组患者的全局视盘周围RNFL厚度分别为(83.5±11.6)μm和(94.2±10.2)μm，组间比较差异有统计学意义(t＝7.58，P_校正<0.05)，其AUC为0.76(95%CI：0.70~0.82)。分象限分析显示，HMG组患者的上方和下方RNFL厚度差异均有统计学意义(t＝7.61，12.85；P_校正<0.05)。HMG组和HM组患者的全局视盘周围RPC密度分别为(47.1±5.1)%和(49.2±3.4)%，组间比较差异有统计学意义(t＝3.76，P_校正<0.05)，AUC为0.63(95%CI：0.56~0.70)。其中，下方象限RPC密度AUC最高(AUC＝0.79，95%CI：0.73~0.84)，HMG组和HM组患者的颞侧象限RPC密度分别为(44.8±4.5)%和(45.2±3.8)%，比较的差异无统计学意义(t＝0.74，P_校正>0.05)。HMG组和HM组患者的GCC全局平均厚度分别为(85.1±10.5)μm和(92.4±8.2)μm，组间比较差异有统计学意义(t＝6.00，P_校正<0.05)，AUC为0.71(95%CI：0.64~0.77)。HMG组和HM组患者的总体损失体积(GLV)分别为(14.2±6.8)%和(4.2±3.2)%，组间比较差异有统计学意义(t＝-14.65，P_校正<0.05)，AUC为0.88(95%CI：0.85~0.91)。HMG组和HM组患者的内环下方SCP血管密度分别为(44.5±5.2)%和(47.8±3.9)%，组间比较差异有统计学意义(t＝5.57，P_校正<0.05)，在SCP各分区中鉴别效能最佳，AUC为0.69(95%CI：0.62~0.76)。两组内环和外环各分区DCP层比较的差异均无统计学意义(t＝1.11，1.93，0.69，0.99，1.73，2.20，0.64，0.93；P_校正>0.05)。基于单指标筛选结果构建递进式多因素Logistic回归模型，模型1纳入视盘周围下方RNFL厚度，AUC为0.850(95%CI：0.810~0.890)；模型2较模型1加入GLV，AUC提高至0.928(95%CI：0.895~0.961)，较模型1差异有统计学意义(DeLong检验P<0.05)；模型3较模型2加入黄斑SCP内环下方毛细血管密度，AUC提高至0.946(95%CI：0.918~0.974)，差异有统计学意义(DeLong检验P<0.05)。

HM患者视盘周围下方RNFL厚度、黄斑GLV及黄斑SCP内环下方血管密度均具有较好的青光眼鉴别价值。多模态结构与血流参数联合可显著提高诊断效能，为HM合并OAG的辅助诊断提供依据。

The aim of this study is to compare the diagnostic performance of peripapillary and macular structural and perfusion parameters obtained by swept-source optical coherence tomography (OCT) and optical coherence tomography angiography (OCTA) in high myopia with open-angle glaucoma (OAG), and to develop a parsimonious combined diagnostic model.

A total of 239 patients (239 eyes) with HM who attended the Department of Ophthalmology, Peking University Third Hospital, from June 2022 to December 2024 and completed the relevant examinations were enrolled in the study. Among them, 97 were male (97 eyes) and 142 were female (142 eyes) with a mean of (53.1±8.6) years (ranging from 32 to 68 years). According to the presence or absence of OAG, the participants were divided into the HM group and the HM combined with OAG group (HMG group). Age, gender, spherical equivalent, axial length (AL), and visual field indices were recorded. OCT and OCTA were used to obtain parameters including peripapillary retinal nerve fiber layer (RNFL) thickness and radial peripapillary capillary (RPC) density, macular ganglion cell complex (GCC) thickness and loss volume indices, superficial capillary plexus (SCP) vessel density, and deep capillary plexus (DCP) vessel density. After Shapiro-Wilk normality testing, data were expressed as ±s, and comparisons between the two groups were performed using the independent-samples t test. The Benjamini-Hochberg method was used for false discovery rate correction of sectoral structural and blood flow parameters. Gender was expressed as number of cases and percentage and compared using the χ² test. Receiver operating characteristic curve analysis was used to calculate the area under the curve (AUC) and 95% confidence interval (CI), and differences in AUC were compared using the DeLong test. The indicators with significant statistical significance of AUC were selected to enter the multivariate Logistic regression model, and the simplified joint diagnosis model was constructed by multivariate Logistic regression analysis. Multicollinearity was evaluated by variance expansion factor, and variance expansion factor<5 was acceptable.

There were no statistically significant differences between the HM group 117 patients (117 eyes) and the HMG group 122 patients(122 eyes) in age, AL, spherical equivalent refraction, or central corneal thickness (t=-1.63, -1.50, 0.97, 0.8; P>0.05). However, the differences in mean deviation and pattern standard deviation of the visual field between the HMG and HM groups were statistically significant (t=15.3, -11.9; P<0.05). The global peripapillary RNFL thickness in the HMG group was (83.5±11.6) μm, which was lower than that in the HM group (94.2±10.2) μm, and the difference between groups was statistically significant (t=7.58, adjusted P<0.05), with an AUC of 0.76 (95%CI: 0.70 to 0.82). Quadrant analysis showed that the differences in superior and inferior RNFL thickness in the HMG group were both statistically significant (t=7.61 and 12.85; adjusted P<0.05). The global peripapillary RPC density in the HMG group was (47.1±5.1) %, which was lower than that in the HM group (49.2±3.4)%, and the difference between groups was statistically significant (t=3.76, adjusted P<0.05), with an AUC of 0.63 (95%CI: 0.56 to 0.70). Among these, inferior quadrant RPC density had the highest AUC (AUC=0.79, 95%CI: 0.73 to 0.84). The temporal quadrant RPC density in the HMG group was (44.8±4.5) %, which was close to that in the HM group (45.2±3.8)%, with no statistically significant difference (t=0.74, adjusted P>0.05). Among macular structural parameters, the global average GCC thickness in the HMG group was (85.1±10.5) μm, lower than that in the HM group (92.4±8.2) μm, and the difference between groups was statistically significant (t=6.00, adjusted P<0.05), with an AUC of 0.71 (95%CI: 0.64 to 0.77). The GLV in the HMG group was (14.2±6.8) %, higher than that in the HM group (4.2±3.2) %, and the difference between groups was statistically significant (t=-14.65, adjusted P<0.05), with an AUC of 0.88 (95%CI: 0.85 to 0.91). Analysis of macular blood flow parameters showed that, in the SCP, the inferior inner-ring vessel density in the HMG group was (44.5±5.2)%, lower than that in the HM group (47.8±3.9)%, and the difference between groups was statistically significant (t=5.57, adjusted P<0.05); its discriminatory performance was the best among all SCP sectors, with an AUC of 0.69 (95%CI: 0.62 to 0.76). At the DCP level, there were no statistically significant differences between the two groups in any inner-ring or outer-ring sector (t=1.11, 1.93, 0.69, 0.99, 1.73, 2.20, 0.64, 0.93; adjusted P>0.05). Based on the screening results of single indicators, stepwise multivariable logistic regression models were constructed. Model 1 included inferior peripapillary RNFL thickness and had an AUC of 0.850 (95%CI: 0.810 to 0.890). Compared with Model 1, Model 2 additionally included GLV, and the AUC increased to 0.928 (95%CI: 0.895 to 0.961), with a statistically significant difference compared with Model 1 (DeLong test P<0.05). Compared with Model 2, Model 3 additionally included inner inferior macular SCP capillary density, and the AUC increased to 0.946 (95%CI: 0.918 to 0.974), with a statistically significant difference (DeLong test P<0.05).

Inferior peripapillary RNFL thickness, macular GLV, and inferior inner-ring macular SCP vessel density all have good value for glaucoma discrimination in HM patients. The combination of multimodal structural and blood flow parameters can significantly improve diagnostic performance and provide a basis for the auxiliary diagnosis of HM combined with OAG.