兽研引言
既往研究表明，内皮糖萼可因容量超负荷而受损。本研究采用侧流暗场视频显微镜和Glycocheck™软件，评估短暂性血管内容量超负荷对健康猫估计内皮糖萼厚度的影响。
方法
本研究为一项前瞻性、盲法、交叉、单中心实验研究，纳入二十只健康猫。猫被随机分配接受30 mL/kg/10 min的Hartman溶液（容量超负荷组）或不接受液体（对照组），洗脱期为7天。在丙泊酚麻醉下，于T0（基线）、T1（干预后立即）、T2（干预结束后3小时）和T3（干预结束后24小时）获取舌下黏膜视频显微镜图像，每个时间点进行三次测量。随后使用Glycocheck™软件自动计算直径分别为5–25、5–9、10–19和20–25 μm血管的灌注边界区域（PBR），其为内皮糖萼厚度的反向估计值。在每个时间点测量红细胞压积（PCV）。使用Shapiro–Wilk检验和视觉检查评估数据正态性。采用以猫为随机效应的线性混合效应回归模型分析GlycoCheck™与其他参数之间的交互作用。
结果
PBR 5–25 μm、5–9 μm和10–19 μm随时间增加，在T2时达到显著性（分别为p = 0.001、p = 0.017和p = 0.008）。PBR 5–9发现显著的组×时间点交互作用，容量超负荷组在T2（p = 0.028）和T3（p = 0.010）时数值降低。PCV降低与PBR 5–25 μm（p = 0.003）和PBR 10–19 μm（p = 0.001）增加显著相关。
结论
本研究未证明健康猫在短暂性容量超负荷后估计内皮糖萼厚度降低。PBR确实随时间增加，提示可能存在麻醉效应。PCV变化影响GlycoCheck™参数，这一发现应进一步探讨。需要设计基于猫容量动力学认识的研究，以探究静脉输液对猫内皮糖萼的影响。
引言
血管内容量超负荷（VO）及由此导致的液体超负荷是有害的，因为它们促进间质性水肿、损害细胞氧输送，并最终促进多器官功能障碍和死亡（1）。容量超负荷和液体超负荷通常为医源性，由过度积极的静脉（IV）液体输注引起，尤其是在适应能力降低的危重患者中（2, 3）。容量超负荷可被一类称为利钠肽（NP）的分子所抵消，这些分子主要由心肌分泌（4, 5）。其作用导致利钠、抑制肾素-血管紧张素-醛固酮系统、血管舒张和血管通透性增加（5）。NP诱导的内皮糖萼（EG）损伤可增加血管通透性，从而使液体转移至间质间隙（4）。虽然这些机制在健康个体中具有保护作用，但它们可能变得适应不良，尤其是在危重患者中，可能加剧已有的EG损伤（2, 6）。
大量实验和临床研究探讨了VO对EG的影响，结果相互矛盾，主要通过测量脱落的EG损伤生物标志物。一些研究报告VO和积极IV液体输注导致NP释放和EG损伤（7–13），而其他研究（通常较新）未能证明此类效应（14–18）。这些发现主要见于稳定患者还是已有EG损伤的危重患者仍存在争议（3, 6）。此外，大多数研究关注VO的即时效应，缺乏中长期后果的信息。
猫可能特别易发生VO，因为与犬和人相比，其血容量相对较小（19–21），无症状心脏病发病率高（22–24），液体清除率较低（25, 26）。因此，猫为探索IV液体诱导的VO对EG的影响提供了有说服力的模型。
有两种研究EG损伤的方法适用于临床环境。第一种是测量推测在损伤后脱落的EG生物标志物。然而，结果解读仍然具有挑战性——特别是在静脉输液治疗的背景下，因为升高往往不一致、幅度较低，并受研究设计的影响（3, 6, 27）。黏膜血管的实时侧流暗视野视频显微镜检查（SDFV）（6）结合专有的GlycoCheck™软件，提供了另一种临床上可用的评估EG厚度的方法。该软件分析SDFV获取的图像并计算灌注边界区域（PBR），这是对直径5至25 μm血管中EG厚度的反向估计。较高的PBR值表示较薄（受损）的EG，而较低的值表示较厚（健康）的层。GlycoCheck™参数的容许区间已在猫中建立（28），并且在一项针对自然发病猫的探索性研究中调查了其可行性及与疾病严重程度的关联（29）。此外，我们课题组此前曾使用GlycoCheck™评估液体负荷对健康麻醉猫EG的即时影响，但结果尚无定论（30）。
本研究旨在使用SDFV和GlycoCheck™软件（31）研究一过性静脉输液诱导的容量超负荷（VO）对健康麻醉猫EG厚度的影响。我们假设，与对照组相比，快速给予30 mL/kg缓冲平衡等渗晶体液（10分钟内输注）所诱导的VO会导致直径为5–25 μm及5–9、10–19和20–25 μm亚组血管的PBR一过性升高，表明EG厚度降低。此外，PCV下降（由于血液稀释和/或麻醉效应）可能影响所评估血管中红细胞的边缘化，从而增加上述PBR。
材料与方法
研究群体
这是一项前瞻性、随机、盲法、交叉实验研究，经梅西大学动物伦理委员会批准（MUAEC方案22/81），并遵循动物体内实验研究报告指南（32）。共纳入20只健康的研究用猫。纳入标准为：根据病史健康、无体重减轻、体格检查发现、全血细胞计数、血清生化和T4浓度（在研究第一周干预前进行）以及研究开始前两周内进行的两分钟聚焦超声心动图检查筛查隐匿性心脏病（33）均正常。没有动物符合预定义的排除标准。预定义的人道终点和停止标准包括临床相关的麻醉并发症（如心血管不稳定、呼吸障碍）、操作限制（如静脉导管脱落、对低应激操作的抵抗）或任何被认为损害动物福利的不良事件。在此类情况下，动物按适当方式复苏，且不在后续时间点收集进一步数据。
干预方案
猫被放置于单独笼中，每次干预开始前禁食12小时，可自由饮水。所有猫在静脉导管置入左或右头静脉前，先给予加巴喷丁（Neurontin™，Viatris Specialty LLC）100 mg口服（90分钟前）和布托啡诺（Torbugesic，Zoetis Inc.）0.3 mg/kg皮下注射（30分钟前）进行镇静。随后静脉给予丙泊酚（Provive 1%，Baxter Healthcare Ltd.）至起效（下颌张力松弛）。在图像采集期间，通过丙泊酚持续速率输注（0.05–0.2 mg/kg/min）维持麻醉。猫自主呼吸，未插管。麻醉由经验丰富的兽医技术员监督，并频繁监测麻醉变量——心率（HR）、脉搏质量、黏膜颜色、毛细血管再充盈时间、体温、呼吸频率、脉搏血氧饱和度（SpO2，Radical 7，Masimo Corporation，Irvine，CA，USA）和示波法血压（PetTrust，BioCare Corporation，Taoyuan，Taiwan）（临床参数每5分钟监测一次，血压每10分钟一次，脉搏血氧饱和度连续监测）。干预时间线见图1。
[图：涉及20只健康麻醉猫的交叉研究干预方案时间线，随机接受以下任一]
每只猫使用在线随机化工具（https://www.randomizer.org/）随机分配至以下两组：一组通过液体输注泵（Niki V4 Volumetric Infusion Pump, Caesarea Medical Electronics Ltd., Israel）在室温下于10分钟内给予30 mL/kg的缓冲平衡等渗晶体液（Hartman's solution, Baxter Healthcare Ltd.）——容量超负荷（VO）组；另一组不给液体——对照组。所有猫在初始图像采集（T0）、静脉输液（或对照）及第二次图像采集（T1）期间均维持丙泊酚麻醉，之后待其复苏并置于各自笼中，在T2（3小时）和T3（24小时）时间点短暂静脉注射丙泊酚重新麻醉以进行图像采集。T2图像采集后，将猫置于各自笼中，保留静脉导管在位，自由采食，并在次日T3图像采集前12小时禁食，自由饮水。24小时结束时，拔除静脉导管，猫返回其常规群养环境。在每个时间点于麻醉状态下经颈静脉穿刺采集血液用于测定红细胞压积（PCV）和总固体（TS）。随后每只猫经过7天的洗脱期，之后接受相反的干预。
图像采集
所有图像采集均由同一操作者（IY）完成，该操作者对受试对象的分组分配设盲，仅在场进行图像采集，并在给予静脉液体（或不给予）前离开干预区域，在给药完成后返回。将猫置于侧卧位，暴露舌根部以使舌下黏膜可供视频显微镜检查，方法如前所述（28）。操作者将连接至装有GlycoCheck™软件（MicroVascular Health Solutions, USA）（34）的笔记本电脑的手持式SDFV摄像头放置与舌下黏膜接触。在四个不同时间点——T0（基线）、T1（容量超负荷结束时或对照组基线后10分钟）、T2（干预结束后3小时）和T3（干预结束后24小时）——分别从不同的舌下黏膜部位采集三次视频记录。
GlycoCheck™分析
测量和计算GlycoCheck™参数的方法已如前所述（28）。简言之，SDFV摄像头发射波长为540 nm的绿光，该光被循环红细胞（RBC）中的血红蛋白吸收。因此，血管在明亮背景下呈现暗色。
视频记录实时显示，以便操作者评估图像质量，包括对焦、压力和运动伪影（35）。该软件还提供运动、强度和对焦的线性质量指示条，绿色区段表示可接受范围，并且仅在满足预定义质量标准时才记录图像。自动应用运动、对焦和强度调整以优化图像采集。
每次测量从约10个视频记录中进行，每个记录包含40帧（31）。软件通过分析RBC与背景之间的对比度，自动识别直径小于30 μm的血管。然后软件将血管按长度细分为主要（10 μm）和随后的次要（0.5 μm）血管段，并进一步确定主要血管段内具有足够对比度可供分析的次要血管段的百分比（图2）。每次测量期间共记录约3,000个主要血管段，包括直径约3–30 μm的血管。此外还记录每个分析直径的主要血管段计数。次要血管段中具有足够对比度的比例>60%的主要血管段被视为有效血管段（VVS），用于分析。有效血管密度（VVD）/mm²则通过将10 μm乘以所识别的VVS数量来表示。
[图：猫舌下毛细血管的侧流暗场视频显微镜和GlycoCheck™图像，无（A）和有（B）m]
其次，评估RBC柱宽度的最小柱宽度、柱位置和信噪比。根据所测量的柱宽度，软件随后基于特定VVS在每个部位所有记录中携带RBC的时间比例计算RBC充盈百分比（%RBC充盈）。中位RBC柱宽度（Median P50）利用所有测量柱宽度中RBC动态侧向位置的强度分布曲线计算。
最后，对获得的Median P50曲线拟合线性回归线，以推导血管段的灌注直径。由此，PBR（灌注边界区域）作为EG厚度的反向估计值，计算为Median P50与灌注直径之间的距离（[灌注直径 – 中位RBC柱宽度]/2）。因此，所报告的GlycoCheck™参数包括直径5至25 μm血管以及直径5–9 μm、10–19 μm和20–25 μm血管亚组的VVD、RBC充盈百分比、Median P50和PBR。因此，较高的PBR表明EG较薄，较低的PBR表明EG较厚。
红细胞压积和总固体
每只猫在每个时间点分别使用离心毛细管微量血细胞比容管和手持折光仪测量红细胞压积和总固体（TS），由同一操作者完成。
统计分析
样本量计算基于一项接受高渗盐水扩容的麻醉兔研究（36）。所报告的PBR 5–25 μm的平均差异为0.19 μm，报告的基线和干预后标准差（SD）分别为0.30和0.18。由于该研究未提供配对差异数据，本交叉设计的受试者内差异SD使用这些报告的SD和假设的受试者内相关性进行估计。使用双侧α为0.05和80%把握度，得出估计样本量约为20只猫。
根据既往在健康和危重病人中的报告推荐（37, 38），对测得的GlycoCheck™参数的三次重复取平均值用于进一步分析。使用Shapiro–Wilk检验并通过直方图和Q-Q图目视检查检验数据正态性（39–41）。正态分布数据以平均值±标准差（SD）表示，非正态分布数据以中位数（四分位距，IQR）表示（41）。
全程使用线性混合效应回归模型评估治疗效应，以猫作为随机效应（42）。GlycoCheck™参数作为因变量分别对组别、时间点及其交互作用进行评估（模型1.1–1.7），并单独对PCV进行评估（模型2.1–2.7）。红细胞压积和TS同样作为因变量分别对组别、时间点及其交互作用进行评估（分别为模型3和4）。作为敏感性分析，PBR的组别×时间点模型在纳入PCV作为额外自变量后重复进行（模型5.1–5.4）。
PBR（5–25 μm及5–9 μm、10–19 μm和20–25 μm亚组）与HR、平均动脉压（MAP）和体温之间的关联在单独模型中评估（分别为模型6.1–6.4、7.1–7.4和8.1–8.4）。最后，HR、收缩动脉压、MAP、舒张动脉压和体温分别对组别、时间点及其交互作用进行评估（分别为模型9.1–9.5）。每个模型的详细信息包括所纳入的因变量和自变量，汇总于补充表1。
通过使用Shapiro–Wilk检验评估正态性和异方差性以及目视检查残差直方图来评估模型假设。线性混合效应模型的固定效应估计值以近似95%置信区间呈现，置信区间使用Wald方法基于估计系数±1.96×标准误得出（42）。未进行多重比较的正式校正；模型得出的估计值在探索性背景下报告和解释。使用四分位距方法调查离群值，下界设定为第1四分位数减去1.5倍IQR，上界为第3四分位数加上1.5倍IQR。在纳入和排除离群值的情况下均进行建模，以比较模型的准确性和拟合度。
PCV与GlycoCheck™参数之间的相关性对正态分布数据使用Pearson检验计算，对非正态分布数据使用Kendall's tau计算。所有统计分析均在R中进行（Team 2023）（43）。完整数据集，包括每个时间点所有变量的个体值，见补充材料1。
结果
共有20只猫纳入分析：9只雄性去势、7只雌性未绝育和4只雌性绝育。中位年龄为3.21岁（IQR：1.68），平均体重为3.77±0.76 kg。共分析了156个时间点的图像。三只猫的四次图像采集在第二周因短期并发症未完成。一只猫在T1图像采集结束时麻醉期间出现短暂氧饱和度下降。麻醉终止后该猫顺利恢复，未再出现临床症状。两只猫在最终数据点采集前发生IV导管脱落，且IV导管无法重新置入。这导致总体成功率为97%。
麻醉变量结果和统计分析见补充材料2。组别和时间点对麻醉变量有不同影响，包括T1时MAP较基线下降。然而，数值大多保持在临床监测可接受范围内。一只猫在VO组时于T1出现低血压（定义为MAP < 60 mmHg）（44）。该猫在各组和各时间点血压均持续略低，因此被认为是生物学离群值。28%的不同时间点记录到低体温（定义为T° < 36.7 °C）（45），记录的最低体温为34.8 °C。心率持续略低于该物种的报道值（46）。然而，这对于该群体研究猫来说是典型的，不被认为异常。所有个体麻醉变量结果可查阅补充材料1。心率、MAP和体温对PBR无显著影响（补充材料2）。
两套探究GlycoCheck™参数的模型（含与不含离群值）得出相似结果，并呈现一致的总体趋势。报告结果采用包含离群值的模型。在模型1中纳入PCV并未实质性改变任何PBR参数的组×时间点效应。GlycoCheck™参数的汇总统计见表1。统计报告见补充材料3。所分析的所有GlycoCheck™参数在基线时均无显著差异。
Table 1. GlycoCheck™ parameters in 20 healthy anesthetized cats subjected to transient volume overload (30 mL/kg Hartman’s solution over 10 min) or control (no intravenous fluids), measured at four timepoints over 24 h.
| Volume overload group | Control group | |||||||
|---|---|---|---|---|---|---|---|---|
| Baseline | 10 min | 3 h | 24 h | Baseline | 10 min | 3 h | 24 h | |
| Parameter | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD |
| VVD (mm 2 /mm 2 ) | 452.96 ± 98.08 | 405.98 ± 106.42 | 429.56 ± 100.86 | 444.63 ± 109.12 | 466.93 ± 134.28 | 448.82 ± 87.40 | 442.03 ± 81.30 | 463.09 ± 88.81 |
| %RBC filling (%) | 0.60 ± 0.07 | 0.59 ± 0.04 | 0.59 ± 0.06 | 0.62 ± 0.06 | 0.62 ± 0.06 | 0.59 ± 0.04 | 0.57 ± 0.06 | 0.62 ± 0.05 |
| Median P50 (μm) | 9.11 ± 1.52 | 8.95 ± 1.28 | 8.49 ± 1.07 | 8.41 ± 1.06 | 8.69 ± 1.53 | 8.91 ± 1.38 | 8.73 ± 1.39 | 8.24 ± 1.33 |
| PBR 5–25 (μm) | 2.58 ± 0.24 | 2.62 ± 0.11 | 2.64 ± 0.17 | 2.46 ± 0.18 | 2.48 ± 0.21 | 2.55 ± 0.19 | 2.63* ± 0.13 | 2.48 ± 0.20 |
| PBR 5–9 (μm) | 1.40 ± 0.13 | 1.38 ± 0.12 | 1.38 § ± 0.09 | 1.29 § ± 0.14 | 1.34 ± 0.15 | 1.38 ± 0.10 | 1.42* ± 0.10 | 1.36 ± 0.09 |
| PBR 10–19 (μm) | 2.93 ± 0.28 | 2.99 ± 0.13 | 3.00 ± 0.17 | 2.79 ± 0.20 | 2.83 ± 0.23 | 2.92 ± 0.21 | 3.00 ± 0.15 | 2.83 ± 0.23 |
灌注边界区随时间呈上升趋势，表观峰值出现在T2（干预后3小时）。混合效应模型发现时间对PBR 5–25 μm、PBR 5–9 μm和PBR 10–19 μm有显著影响。对于PBR 5–25 μm和PBR 10–19 μm，T2时的模型估计值相对基线更高（估计值分别为0.150，95% CI 0.037至0.264，p=0.001；估计值0.175，95% CI 0.047至0.303，p=0.008）。
对于PBR 5–9 μm，也观察到时间的显著影响，T2时数值相对基线升高（估计值0.0887，95% CI 0.016至0.161，p=0.017）。然而，在T2（估计值−0.116，95% CI −0.219至−0.013，p=0.028）和T3（估计值−0.138，95% CI −0.0242至−0.033，p=0.010）发现显著的组×时间点交互作用，表明这种时间依赖性升高在VO组中被减弱，导致这些时间点VO组的PBR 5–9 μm低于对照组。PBR的组与时间点交互作用见图3。各组的每个时间点的个体PBR值见补充材料1。PBR 5–25 μm的个体轨迹见图4，PBR 5–9、10–19和20–25 μm的个体轨迹见补充材料4。
[图：直径为5–25 μm（A）、5–9 μm（B）、10–19 μm（C）的血管中灌注边界区随时间的变化，]
[图：20只健康麻醉猫中直径为5–25 μm的血管灌注边界区随时间的个体变化]
直径为5–25 μm的主要（10 μm）血管段计数的中位数（IQR）为1520.05（356.50）。在此范围内，5–9 μm、10–19 μm和20–25 μm亚组的计数中位数分别为910.32（348.75）、566.91（172.83）和42.82（33.83）（图5）。
[图：健康麻醉猫按血管直径（5–25 μm）划分的主要血管段计数分布。柱形表示]
红细胞压积随时间显著波动，混合效应模型发现时间点有显著影响。相对基线，模型估计的PCV在T1较低（估计值−3.350，95% CI −4.863至−1.837，p<0.001），在T2较高（估计值1.950，95% CI 0.437至3.463，p=0.012），在T3也较高（估计值2.234，95% CI 0.699至3.767，p=0.005）。在T1（估计值−4.200，95% CI −6.339至−2.061，p<0.001）和T2（估计值−2.554，95% CI −4.709至−0.399，p=0.021）发现显著的组×时间点交互作用，表明这些时间点VO组的PCV低于对照组。
同样，TS随时间变化，混合效应模型发现时间点有显著影响。相对基线，模型估计的TS在T1较低（估计值−2.200，95% CI −4.292至−0.108，p=0.04），在T2较高（估计值2.600，95% CI 0.508至4.692，p=0.016）。在T1（估计值−12.450，95% CI −15.409至−9.491，p<0.001）和T2（估计值−5.812，95% CI −8.793至−2.831，p<0.001）发现显著的组×时间点交互作用，表明这些时间点VO组的TS值低于对照组。
红细胞压积与VVD呈正相关（估计值5.747，95% CI 1.956至9.537，p=0.003），与%RBC充盈率呈正相关（估计值0.004，95% CI 0.002至0.006，p<0.001），与PBR 5–25 μm呈负相关（估计值−0.011，95% CI –0.018至−0.004，p=0.003），与PBR 10–19 μm呈负相关（估计值−0.013，95% CI −0.021至−0.005，p=0.001）。PCV与GlycoCheck™参数的相关系数见表2，尽管具有统计学显著性，但总体较弱。所有线性混合效应模型的完整固定效应结果见补充材料5。
Table 2. Correlation between packed cell volume and GlycoCheckTM parameters using Pearson’s test (except for perfused boundary region 20–25 μm).
| Parameter | r | 95% CI | p – value |
|---|---|---|---|
| VVD | 0.235 | 0.081—0.378 | 0.003 |
| %RBC filling | 0.277 | 0.126—0.416 | < 0.001 |
| Median P50 | −0.071 | −0.225 — 0.088 | 0.382 |
| PBR 5–25 μm | −0.238 | −0.381 — −0.0841 | 0.003 |
| PBR 5–9 μm | −0.102 | −0.255 — 0.056 | 0.203 |
| PBR 10–19 μm | −0.255 | −0.396 — −0.102 | 0.001 |
| PBR 20–25 μm | τ = −0.111 | - | 0.048 |
讨论
本研究结果表明，快速静脉给予30 mL/kg缓冲平衡晶体液不会降低健康猫的估计EG厚度。然而，PBR 5–25 μm、PBR 5–9 μm和PBR 10–19 μm在干预后3小时增加，提示EG厚度短暂降低，可能继发于麻醉效应。有趣的是，小血管（5–9 μm）的PBR在VO组后期时间点随时间推移较低，提示这些效应可能因静脉液体的给予而减轻。
短暂性容量超负荷未导致该组健康猫PBR增加。这与我们团队先前在健康猫中的研究结果一致，该研究显示，使用Hartman溶液或6%羟乙基淀粉130/0.4进行液体负荷（20分钟内10 mL/kg）后PBR未增加（30）。在麻醉兔中也报道了类似结果，仅高渗盐水推注（5 mL/kg的10% NaCl）增加PBR，而使用乳酸林格氏液进行容量负荷（30 mL/kg/h）无影响（17, 36）。
在犬失血性休克模型中，无静脉液体时，PBR在失血和自体输血期间通常保持不变（47）。然而，一项研究显示，与天然或合成血液制品相比，使用30 mL/kg LRS或7.5 mL/kg羟乙基淀粉复苏后PBR增加（13）。同样，在脓毒症儿童中，与非平衡晶体液相比，平衡晶体液推注导致干预后6小时PBR增加（48）。综上所述，这些研究提示，单独静脉液体可能不会改变健康动物的PBR，而更严重的干预（如失血）或基础疾病状态（如脓毒症）可能易导致可检测的EG改变。然而，研究设计、物种和方法学的差异限制了直接比较或转化结论。
在本研究中，对照组PBR 5–25 μm、5–9 μm和10–19 μm随时间推移增加，并在24小时恢复至基线。这些变化不能归因于静脉液体相关的VO。因此，应考虑其他解释。麻醉相关的血容量变化在文献中已有充分记录，主要归因于麻醉诱导的低血压引起液体从间质向血管内间隙转移（18, 49, 50）。此外，大多数麻醉药，包括丙泊酚，在犬和猫中引起脾脏松弛，导致红细胞隔离和PCV短暂降低（20, 51–55）。在本研究中，两组PCV和TS均短暂降低，支持麻醉相关的血容量变化，尤其是在初始较长的麻醉期间（T0和T1合并）。
这种血容量变化是否与EG损伤相关仍不确定。最近一项研究检查了低血压与血压正常的人类手术患者的血浆容量变化，显示低血压组血浆容量增加十倍。然而，这与NP产生或EG脱落增加无关（18），提示低血压诱导的容量扩张不会损伤EG。更重要的是，尽管本研究中T1时MAP降低，但仅一只猫出现轻度低血压。因此，麻醉相关低血压及其相关的血浆容量扩张不太可能解释观察到的PBR变化。
更广泛地说，麻醉血流动力学变量（HR、MAP和温度）与PBR无关，这与先前使用类似麻醉方案的健
另一个值得考虑的麻醉相关因素是麻醉药物本身的选择，特别是丙泊酚，它可能对内皮及其糖萼有直接影响。使用SDFV的研究报告了丙泊酚对人体微循环的影响，通常显示随时间推移或与其他镇静剂和麻醉剂相比，灌注血管密度及其他微循环功能指标呈下降趋势（56–60）。因此，丙泊酚也可能影响猫的微循环，从而影响估计的EG厚度。然而，这些研究纳入的是异质性队列，主要是危重患者，且研究设计差异很大，无法与本研究进行直接比较。此外，他们未评估估计的EG厚度，因为所使用的方法不支持PBR计算。
在微循环成像之外，体外和体内实验研究均表明，丙泊酚过量后可导致内皮细胞损伤和EG脱落（61–63）。相比之下，比较丙泊酚与挥发性麻醉剂对人体影响的临床研究并未一致显示丙泊酚给药后EG脱落增加（64–66）。这些效应似乎呈剂量依赖性和时间相关性，且诸如潜在的EG损伤性疾病等混杂因素限制了从临床数据中得出的结论。
既往评估猫PBR的研究中使用了丙泊酚（28, 30），尽管这些研究未在3小时内重复给予丙泊酚。具体而言，在我们团队先前一项评估静脉输液对健康猫GlycoCheck™参数影响的研究中，发现输液结束时与基线相比PBR 10–19 μm增加（30），提示效应相当。考虑到猫对丙泊酚相关氧化应激的相对“敏感性”（67–69）、猫体内酚类化合物半衰期较长且代谢较慢（51），丙泊酚可能导致了EG厚度的短暂降低，尤其是考虑到T0至T2之间的重复给药。需要进一步研究比较重复给药和使用不同麻醉剂的效果，以探讨这种可能性。
令人惊讶的是，VO组在较晚时间点（T2和T3）PBR 5–9 μm较低。静脉输液可能减轻了一些负面影响，特别是在分析的最小血管中。我们团队先前一项研究使用Hartman溶液或羟乙基淀粉对健康猫进行静脉输液挑战的即时效应，同样报告了羟乙基淀粉组PBR降低，尽管是在较大的血管中（10–19 μm）（30）。这可能是由于麻醉剂稀释或EG的“再水化”所致。有必要专门设计研究来探究麻醉剂对EG的影响，以支持这一假设。
本研究表明PCV与GlycoCheck™参数之间存在弱关联。先前在成人及新生儿人类患者中已发现PCV与PBR之间的关联（70, 71）。这些发现与我们团队先前的一项研究基本一致，该研究评估了液体挑战对猫GlycoCheck™参数的影响，发现随着PCV降低，VVD和%RBC充盈度降低，PBR增加（30）。
尽管这些关联较弱，但鉴于所有GlycoCheck™参数均依赖于在对比背景下识别红细胞，且PBR计算依赖于存在符合“有效”密度标准和足够%RBC充盈度的血管（31），这些关联值得进一步研究。如果专门设计的研究证实PCV对PBR存在一致影响，则在解读PBR测量值时，尤其是在临床研究中，可能需要对PCV进行调整。
有趣的是，大多数GlycoCheck™参数在24小时时恢复至基线，提示任何EG紊乱要么是短暂的，要么是迅速消退的。然而，这与目前对EG脱落、周转和修复的理解相矛盾（72）。虽然EG脱落可在显著刺激后迅速发生（数分钟至数小时内），但新成分的生物合成通常需要数小时至数天，取决于所涉及的分子和机制（72, 73）。因此，24小时时恢复至基线可能代表尺寸恢复或功能性再水化，而非完全的结构恢复。可以想象，静脉输液和麻醉刺激导致了短暂的EG“干燥”或“收缩”，而非EG成分的大量丢失。尽管不完全的结构恢复理论上可能质疑7天洗脱期的充分性，但GlycoCheck™测量值在24小时内正常化使得有意义的残留效应不太可能。遗憾的是，本研究未评估EG脱落生物标志物。鉴于其在检测液体诱导的EG变化方面的局限性，此类额外分析的可靠性仍不确定。
在考虑内皮糖萼（EG）变化的潜在刺激因素时，所选用的静脉输液剂量值得进一步思考。除了代表经验性“高”静脉输液速率外，该静脉负荷（VO）模型所选用的剂量是基于一项先前的研究，该研究评估了诱导VO的健康猫的心脏超声和尾腔静脉参数（74）。在参考研究中，快速输注30 mL/kg的Hartman's溶液导致心腔和尾腔静脉尺寸出现轻度但具有统计学意义的增大，提示影像学上存在心脏牵张。然而，这种牵张是否足以触发利钠肽（NP）产生仍不清楚。相比之下，一项犬类研究在更长时间内使用了更高的总剂量（1小时内输注90 mL/kg等渗晶体液），显示NP显著升高（75），证实静脉输液在某些条件下可刺激NP释放。考虑到猫对容量过负荷的易感性，研究使用了在伦理和生理上可接受的最高输液速率，以期在不造成伤害的情况下潜在影响EG。然而，本研究未测量NP浓度，因此无法确认NP是否参与EG调节。进一步研究应纳入物种特异性容积动力学数据，以指导猫微循环相关研究或其他静脉输液研究中的液体给药策略（25, 76）。
尽管无统计学意义，但本研究中PBR测量存在显著变异性，尤其是在基线时，尽管采用了交叉设计并按照目前人类推荐的方法对三次测量取平均值（37, 38）。这与先前在大型健康猫队列中的发现一致（28）。这种变异性可能既反映了生物学变异，也反映了将GlycoCheck™系统应用于猫微循环相关的局限性。该软件为人类使用而开发，尚未专门针对猫进行调整，因此猫微循环的物种特异性特征可能影响其衍生参数的某些方面。已有报道显示不同哺乳动物物种间毛细血管床密度存在差异（77），包括使用SDFV评估猫、犬和马灌注血管密度的研究（78–81）。主观上，使用SDFV检查时猫的毛细血管床似乎密度也较低（个人观察）。重要的是，已有报道显示猫的VVD显著低于犬和人（28, 31, 82）。由于VVD是用于计算PBR的源参数，较低的血管密度可能减少可用于分析的数据点数量，从而可能增加测量“噪声”并导致PBR变异性。
GlycoCheck™测量的个体内和个体间变异性并非猫所固有。如上所述，变异性也已在健康和危重病人中专门研究过（37, 38, 83）。尽管对多次测量取平均值可提高可重复性，但并不能消除变异性。操作者间变异性也有报道，尽管一些研究已证明操作者之间具有可接受的可转移性（83, 84）。
除测量变异性外，GlycoCheck™还存在若干方法学局限性。尽管图像分析是自动化的，但图像采集仍依赖于操作者，并易受压力伪影等因素影响。测量还可能受到生理和检查条件的影响（38），凸显了标准化采集方案的重要性。软件版本之间的差异可能限制研究间的直接比较（38）。重要的是，PBR是EG厚度的替代指标，而非EG结构完整性的直接测量。因此，它不能直接量化结构成分的丢失，更重要的是不能评估EG功能。其中一些局限性并非GlycoCheck™所特有，而是使用SDFV进行舌下微循环成像所固有的，在解释和比较使用该方法获得的结果时应予以考虑（85–87）。
总体而言，这些局限性表明，使用当前版本的软件评估猫患者PBR的个体纵向变化可能具有挑战性。尽管如此，只要承认这些局限性，GlycoCheck™仍可用于研究估计EG厚度的组水平变化，特别是考虑到缺乏在猫中无创研究EG的替代方法。
本研究的其他局限性包括样本量相对较小以及时间点略显随意。值得注意的是，T0和T1测量是在同一次麻醉过程中获得的，而T2和T3使用了单独的、较短的麻醉，可能导致各次测量之间的生理背景存在差异。T0–T1麻醉是研究中最长的。然而，其持续时间仍相对较短（约20分钟）。此外，在T1——最可能受较长麻醉持续时间影响的时间点——未观察到PBR的相关变化。
布托啡诺和加巴喷丁因其温和的镇静和心血管作用而被选用，主要用于便于静脉导管置入，因此在T2和T3时未重复给药，尽管加巴喷丁的残余效应可能在T2时仍然存在。由于它们对猫微循环的影响尚不清楚，不同时间点之间给药差异构成了无法排除的潜在混杂因素。
最后，这是一项小规模实验研究，是在该物种先前数据有限的背景下进行的，现有证据大部分来自同一研究团队。因此，研究结果应被视为探索性的，为进一步研究和独立重复提供基础，而非确定性结论或直接临床应用。
结论
本研究未证明短暂、轻度静脉阻塞后估计EG厚度降低。然而，PBR随时间的变化可能提示潜在的麻醉相关效应。由于小动物SDFV成像不可避免地需要短时间麻醉，专门设计以优化麻醉方案的研究至关重要。PCV的变化与GlycoCheck™参数改变相关，这些关联的临床意义值得进一步研究。鉴于猫已知易发生静脉阻塞，应继续探索静脉输液对猫微循环的影响，理想情况下应通过基于对该物种容量动力学更详细了解的研究设计来进行。
查看英文原文▾
Introduction
Previous studies suggest that the endothelial glycocalyx is damaged by volume overload. This study assessed the effects of transient intravascular volume overload on estimated endothelial glycocalyx thickness in healthy cats, using sidestream dark field videomicroscopy and the Glycocheck™ software.
Methods
This was a prospective blinded cross-over single-center experimental study on twenty healthy cats. Cats were randomized to either receive 30 mL/kg/10 min of Hartman’s solution (volume overload group) or no fluids (control) with a washout of 7 days. Sublingual mucosa videomicroscopy images were acquired at T0 (baseline), T1 (immediately after intervention), T2 (3 h after end of intervention) and T3 (24 h after end of intervention) with three measurements performed per timepoint under propofol anesthesia. The perfused boundary region (PBR), an inverse estimate for endothelial glycocalyx thickness, for vessels with diameter of 5–25, 5–9, 10–19 and 20–25 μm, respectively, was then automatically calculated with the Glycocheck™ software. Packed cell volume (PCV) was measured at each timepoint. Data normality was assessed using the Shapiro–Wilk test and visual inspection. Linear mixed-effects regression modelling, with cat as a random effect, analyzed interactions between GlycoCheck™ and other parameters.
Results
PBR 5–25 μm, 5–9 μm and 10–19 μm increased over time, reaching significance at T2 (p = 0.001, p = 0.017 and p = 0.008, respectively). A significant group x timepoint interaction was identified for PBR 5–9 with decreased values at T2 (p = 0.028) and T3 (p = 0.010) in the volume overload group. Decreases in PCV were significantly associated with increases in PBR 5–25 μm (p = 0.003) and PBR 10–19 μm (p = 0.001).
Conclusion
This study did not demonstrate decreases in estimated endothelial glycocalyx thickness after transient volume overload in healthy cats. The PBR did increase over time, suggesting potential anesthetic effects. Changes in PCV affected GlycoCheck™ parameters and this finding should be further explored. Further studies designed with understanding of feline volume kinetics are required to investigate the effects of IV fluids on the feline endothelial glycocalyx.
Introduction
Intravascular volume overload (VO) and the resulting fluid overload are harmful, as they contribute to interstitial oedema, impair oxygen delivery to cells, and ultimately promote multi-organ dysfunction and mortality (1). Volume overload and fluid overload are commonly iatrogenic, resulting from overzealous intravenous (IV) fluid administration, particularly in critically ill patients with reduced adaptive capacity (2, 3). Volume overload is counteracted by a group of molecules called natriuretic peptides (NP), secreted primarily by cardiac muscle (4, 5). Their action results in natriuresis, inhibition of the renin-angiotensin-aldosterone system, vasorelaxation and increased vascular permeability (5). Vascular permeability increases by NP-induced damage to the endothelial glycocalyx (EG), thus allowing fluid to shift into the interstitial space (4). While these mechanisms are protective in healthy individuals, they can become maladaptive, especially in critically ill patients, potentially exacerbating pre-existing EG damage (2, 6).
Numerous experimental and clinical studies have investigated the effects of VO on the EG with conflicting results, primarily by measuring shed EG damage biomarkers. While some report NP release and EG damage with VO and aggressive IV fluid administration (7–13), others (typically more recent) fail to demonstrate such effects (14–18). Whether these findings are seen mostly in stable patients or critically ill patients with pre-existing EG damage remains controversial (3, 6). Additionally, most studies have focused on the immediate effects of VO, and information on medium- and long-term consequences is lacking.
Cats may be particularly predisposed to VO due to their relatively smaller blood volume compared to dogs and people (19–21), a high incidence of asymptomatic heart disease (22–24) and lower fluid elimination rates (25, 26). For these reasons, cats offer a compelling model to explore effects of IV fluid-induced VO on the EG.
Two methodologies for investigating EG damage are applicable to the clinical setting. The first consists of measuring EG biomarkers presumably shed after damage. However, result interpretation remains challenging - particularly in the context of IV fluid administration, as elevations are often inconsistent, of low magnitude, and influenced by study design (3, 6, 27). Real-time sidestream dark field videomicroscopy (SDFV) of mucosal blood vessels (6) combined with the proprietary GlycoCheck™ software, offers another clinically applicable measure for estimating EG thickness. The software analyses SDFV-acquired images and calculates the perfused boundary region (PBR) an inverse estimate of EG thickness in vessels ranging from 5 to 25 μm in diameter. Higher PBR values indicate thinner (damaged) EG while lower value indicates a thicker (healthy) layer. Tolerance intervals for GlycoCheck™ parameters have been established in cats (28) and feasibility and associations with illness severity have been investigated in an exploratory study in cats with naturally occurring disease (29). Moreover, our group has previously evaluated the immediate effects of a fluid challenge on the EG in heathy anesthetized cats using the GlycoCheck™, though findings have been inconclusive (30).
The aim of this study was to investigate the effects of a transient IV fluid-induced VO on the EG thickness in healthy anesthetized cats using SDFV and the GlycoCheck™ software (31). We hypothesized that VO induced by rapid administration of 30 mL/kg of buffered balanced isotonic crystalloid solution given over 10 minutes would lead to transient increase in PBR of vessels with diameters 5–25 μm and 5–9, 10–19 and 20–25 μm subgroups, indicating decreased EG thickness, compared to a control group. Additionally, decreases in PCV (due to hemodilution and/or anesthetic effects) may affect RBC lateralization in assessed vessels which would increase the above-mentioned PBR.
Materials and methods
Study population
This was a prospective, randomized, blinded, cross-over experimental study, approved by Massey University’s Animal Ethics Committee (MUAEC Protocol 22/81) and adhering to the Animal Research Reporting of In Vivo Experiments guidelines (32). Twenty healthy research-purposed cats were enrolled. Cats met inclusion criteria if they were healthy based on history, no weight loss, physical examination findings, complete blood count, serum biochemistry and T4 concentrations (performed during the first week of the study prior to intervention) and the two-minute focused echocardiography exam to screen for occult heart disease (33) performed up to two weeks prior to the start of the study. No animals met predefined exclusion criteria for enrolment. Predefined humane endpoints and stopping criteria included clinically relevant anesthetic complications (e.g., cardiovascular instability, respiratory compromise), procedural limitations (e.g., IV catheter dislodgement, resistance to low-stress handling) or any adverse event deemed to compromise animal welfare. In such cases, animals were recovered as appropriate, and no further data were collected at subsequent timepoints.
Intervention protocol
The cats were placed in individual cages and food withheld 12 h prior to the beginning of each intervention with ad libitum access to water. All cats were sedated with gabapentin (Neurontin™, Viatris Specialty LLC) 100 mg PO 90 min prior and butorphanol (Torbugesic, Zoetis Inc.) 0.3 mg/kg SC 30 min prior to intravenous catheter placement in the left or right cephalic vein. Propofol (Provive 1%, Baxter Healthcare Ltd.) was then administered IV to effect (relaxation of jaw tone). Anesthesia was maintained using a continuous rate infusion of propofol (0.05–0.2 mg/kg/min) for the duration of image acquisitions. The cats were spontaneously breathing and not intubated. Anesthesia was overseen by an experienced veterinary technician and anesthetic variables - heart rate (HR), pulse quality, mucous membrane color, capillary refill time, temperature, respiratory rate, pulse oximetry (SpO2, Radical 7, Masimo Corporation, Irvine, CA, USA), and oscillometric blood pressure (PetTrust, BioCare Corporation, Taoyuan, Taiwan) were frequently monitored (every 5 min for clinical parameters, 10 min for blood pressure and continuous for pulse oximetry). The timeline of the intervention is presented in Figure 1.
[图：Intervention protocol timeline for a crossover study involving 20 healthy anesthetized cats, randomized to receive eithe]
Each cat was randomized using an online randomization tool (https://www.randomizer.org/) to either receive 30 mL/kg over 10 min of a buffered balanced isotonic crystalloid solution (Hartman’s solution, Baxter Healthcare Ltd.) via a fluid infusion pump (Niki V4 Volumetric Infusion Pump, Caesarea Medical Electronics Ltd., Israel) at room temperature – volume overload (VO) group; or no fluids - control group. All cats remained under propofol anesthesia for the initial image acquisition (T0), IV fluid administration (or control) and second image acquisition (T1) after which they were allowed to recover and placed in their individual cages to be re-anesthetized briefly with propofol IV for image acquisitions at T2 (3 h) and T3 (24 h) timepoints. After the T2 image acquisition, cats were placed in their individual cages with the IV catheters maintained in place, fed ad libitum and food removed 12 h prior to the T3 image acquisition the following day with at libitum access to water. At the end of the 24 h period, the IV catheters were removed and the cats returned to their usual group housing. Blood for packed cell volume (PCV) and total solids (TS) measurements was collected under anesthesia via jugular venipuncture at each timepoint. Each cat then underwent a washout period of 7 days after which it received the opposite intervention.
Image acquisition
All image acquisitions were performed by the same operator (IY), who was blinded to subject allocation by being present only for image acquisitions and leaving the intervention area prior to IV fluids being administered (or not) and returning after completion of the administration. Cats were placed in lateral recumbency, and the base of the tongue exposed such as the sublingual mucosa was accessible for the videomicroscope as previously described (28). A hand-held SDFV camera connected to a laptop with GlycoCheck™ software (MicroVascular Health Solutions, USA) (34) was placed in contact with the sublingual mucosa by the operator. Three video recordings were taken from different sublingual mucosal sites at each of four different time points – T0 (baseline), T1 (end of volume overload or 10 min after baseline for the control group), T2 (3 h after end of intervention) and T3 (24 h after end of intervention).
GlycoCheck™ analysis
The method for measuring and calculating GlycoCheck™ parameters has been previously described (28). Briefly, the SDFV camera emits green light at 540 nm wavelength which is absorbed by hemoglobin in circulating red blood cells (RBC). Therefore, vessels appear dark on a bright background.
Video recordings are displayed in real-time to allow the operator to assess image quality, including focus, pressure and motion artifacts (35). The software additionally provides linear quality indicator bars for motion, intensity and focus with green segments denoting acceptable ranges and records images only when predefined quality criteria are met. Automatic adjustments for motion, focus and intensity are applied to optimize image acquisition.
Each measurement is performed from approximately 10 video recordings containing 40 frames each (31). The software automatically identifies vessels that are less than 30 μm in diameter by analysing the contrast between the RBC and the background. The software then subdivides vessels into major (10 μm) and subsequently minor (0.5 μm) vascular segments in length and further determines the percentage of minor segments within a major segment with sufficient contrast for analysis (Figure 2). A total of approximately 3,000 major vascular segments including vessels ranging from ~ 3–30 μm diameter are recorded during each measurement. The counts for major vascular segments of each analysed diameter are additionally recorded. Major segments with > 60% of minor segments with sufficient contrast are considered valid vascular segments (VVS) and are used for analysis. Valid vessel density (VVD) per mm2 is then expressed by multiplying 10 μm by the number of VVS identified.
[图：Sidestream dark field videomicroscopy and GlycoCheck™ images of feline sublingual capillaries without (A) and with (B) m]
Secondly, RBC column widths are assessed for minimal column width, column position, and signal-to-noise ratio. From the measured column widths, the software then calculates the percentage of RBC filling (%RBC filling) based on the proportion of time a particular VVS contains RBC throughout all recordings per site. The median RBC column width (Median P50) is calculated using the intensity profiles of the dynamic lateral positions of RBCs in all measured column widths.
Finally, a linear regression line is fitted through the obtained Median P50 curve to derive the perfused diameter of vascular segments. From that, the PBR (perfused boundary region), an inverse estimate of EG thickness, is calculated as the distance between the Median P50 and the perfused diameter ([Perfused diameter – median RBC column width]/2). Reported GlycoCheck™ parameters, therefore, include VVD, RBC filling %, Median P50 and PBR for vessels with diameter from 5 to 25 μm and for subgroups of vessels with diameters from 5–9 μm, 10–19 μm and 20–25 μm. Higher PBR is, therefore, an indication for a thinner EG and a lower PBR indicates a thicker EG.
Packed cell volume and total solids
Packed cell volume and total solids (TS) were measured at each timepoint for each cat using centrifuged capillary microhematocrit tubes and a hand-held refractometer, respectively, by the same operator.
Statistical analysis
Sample size calculation was based on a study in anesthetized rabbits receiving volume expansion with hypertonic saline (36). The reported mean difference for PBR 5–25 μm was 0.19 μm with reported baseline and post-intervention standard deviation (SD) of 0.30 and 0.18, respectively. As paired-difference data were not available from that study, the SD of the within-subject differences for the present crossover design was estimated using these reported SDs and an assumed within-subject correlation. Using a two-sided α of 0.05 and 80% power, this yielded an estimated sample size of approximately 20 cats.
Triplicates from measured GlycoCheck™ parameters were averaged for further analysis as recommended in previous reports in healthy and critically ill people (37, 38). Data normality was tested using Shapiro–Wilk test and visually inspected using histograms and Q-Q plots (39–41). Normally distributed data is presented as mean ± standard deviation (SD), not normally distributed data is presented as median (Interquartile Range, IQR) (41).
Linear mixed-effects regression modelling was used throughout to evaluate treatment with cat as a random effect (42). GlycoCheck™ parameters were evaluated as dependent variables against group, timepoint and their interaction (model 1.1–1.7), and separately against PCV (model 2.1–2.7). Packed cell volume and TS were similarly evaluated as dependent variables against group, timepoint and their interaction (model 3 and 4, respectively). As a sensitivity analysis, the group x timepoint models for PBR were repeated with PCV included as an additional independent variable (model 5.1–5.4).
Associations between PBR (5–25 μm and the 5–9 μm, 10–19 μm, and 20–25 μm subgroups) and HR, mean arterial pressure (MAP), and temperature were evaluated in separate models (Models 6.1–6.4, 7.1–7.4 and 8.1–8.4, respectively). Finally, HR, systolic arterial pressure, MAP, diastolic arterial pressure, and temperature were evaluated separately against group, timepoint, and their interaction (model 9.1–9.5, respectively). Details for each model including the dependent and independent variables included are summarized in Supplementary Table 1.
Model assumptions were assessed by evaluating normality and heteroscedasticity using the Shapiro–Wilk test and visual inspection of residual histograms. Fixed-effect estimates from the linear mixed-effects models are presented with approximate 95% confidence intervals derived using the Wald method, based on the estimated coefficient ± 1.96 × standard error (42). No formal adjustment for multiple comparisons was applied; model-derived estimates are reported and interpreted in an exploratory context. Outliers were investigated using the interquartile range method, with lower boundary set as Quartile 1 minus 1.5 times IQR and upper boundary – Quartile 3 plus 1.5 times IQR. Modelling was performed with both outliers included and excluded to compare model accuracy and fit.
Correlation between PCV and GlycoCheck™ parameters was calculated using Pearson’s test for normally distributed data and Kendall’s tau for non-normally distributed data. All statistical analyses were conducted in R (Team 2023) (43). The complete dataset, including individual values for all variables at each timepoint, is provided in Supplementary material 1.
Results
Twenty cats were included in the analysis: nine male neutered, seven female entire, and four female spayed. The median age was 3.21 years (IQR: 1.68), and the mean body weight was 3.77 ± 0.76 kg. Images from a total of 156 timepoints were analyzed. Four image acquisitions in three cats were not completed during the second week due to short-term complications. One cat developed a transient oxygen desaturation during anesthesia at the end of T1 image acquisition. The anesthesia was discontinued and the cat recovered uneventfully, showing no further clinical signs. Two cats experienced IV catheter dislodgment prior to final datapoint collection, and IV catheters could not be replaced. This resulted in an overall success rate of 97%.
Anesthetic variable results and statistical analysis are presented in Supplementary material 2. Group and timepoint had varying effects on anesthetic variables, including a decrease in MAP from baseline at T1. However, values largely remained within acceptable limits for clinical monitoring. One cat experienced hypotension (defined as MAP < 60 mmHg) (44) at T1 while in the VO group. This cat demonstrated consistently slightly lower blood pressure across groups and timepoints and therefore was considered a biological outlier. Hypothermia (defined as T° < 36.7 °C) (45) was recorded in 28% of various timepoints, with the lowest recorded temperature being 34.8 °C. Heart rate was consistently slightly lower than the reported for the species (46). However, this was typical for research cats from this colony and was not considered abnormal. All individual anesthetic variable results can be consulted in Supplementary material 1. Heart rate, MAP and temperature had no significant effect on PBR (Supplementary material 2).
Both sets of models investigating GlycoCheck™ parameters (with and without outliers) yielded similar findings and exhibited consistent overall trends. Models with outliers included were used for reporting results. Inclusion of PCV in Model 1 did not materially alter the group x timepoint effects for any PBR parameter. Summary statistics for GlycoCheck™ parameters are presented in Table 1. A statistical report is presented as Supplementary material 3. None of the analyzed GlycoCheck™ parameters were significantly different at baseline.
Table 1. GlycoCheck™ parameters in 20 healthy anesthetized cats subjected to transient volume overload (30 mL/kg Hartman’s solution over 10 min) or control (no intravenous fluids), measured at four timepoints over 24 h.
| Volume overload group | Control group | |||||||
|---|---|---|---|---|---|---|---|---|
| Baseline | 10 min | 3 h | 24 h | Baseline | 10 min | 3 h | 24 h | |
| Parameter | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD | Mean ± SD |
| VVD (mm 2 /mm 2 ) | 452.96 ± 98.08 | 405.98 ± 106.42 | 429.56 ± 100.86 | 444.63 ± 109.12 | 466.93 ± 134.28 | 448.82 ± 87.40 | 442.03 ± 81.30 | 463.09 ± 88.81 |
| %RBC filling (%) | 0.60 ± 0.07 | 0.59 ± 0.04 | 0.59 ± 0.06 | 0.62 ± 0.06 | 0.62 ± 0.06 | 0.59 ± 0.04 | 0.57 ± 0.06 | 0.62 ± 0.05 |
| Median P50 (μm) | 9.11 ± 1.52 | 8.95 ± 1.28 | 8.49 ± 1.07 | 8.41 ± 1.06 | 8.69 ± 1.53 | 8.91 ± 1.38 | 8.73 ± 1.39 | 8.24 ± 1.33 |
| PBR 5–25 (μm) | 2.58 ± 0.24 | 2.62 ± 0.11 | 2.64 ± 0.17 | 2.46 ± 0.18 | 2.48 ± 0.21 | 2.55 ± 0.19 | 2.63* ± 0.13 | 2.48 ± 0.20 |
| PBR 5–9 (μm) | 1.40 ± 0.13 | 1.38 ± 0.12 | 1.38 § ± 0.09 | 1.29 § ± 0.14 | 1.34 ± 0.15 | 1.38 ± 0.10 | 1.42* ± 0.10 | 1.36 ± 0.09 |
| PBR 10–19 (μm) | 2.93 ± 0.28 | 2.99 ± 0.13 | 3.00 ± 0.17 | 2.79 ± 0.20 | 2.83 ± 0.23 | 2.92 ± 0.21 | 3.00 ± 0.15 | 2.83 ± 0.23 |
Perfused boundary region trended upward over time, with apparent peak values at T2 (3 h post-intervention). Mixed-effects modeling identified a significant effect of time for PBR 5–25 μm, PBR 5–9 μm and PBR 10–19 μm. For PBR 5–25 μm and PBR 10–19 μm model estimate values at T2 were higher relative to baseline (estimate 0.150, 95% CI 0.037 to 0.264, p = 0.001 and estimate 0.175, 95% CI 0.047 to 0.303, p = 0.008, respectively).
For PBR 5–9 μm, a significant effect of time was also observed, with increased values at T2 relative to baseline (estimate 0.0887, 95% CI 0.016 to 0.161, p = 0.017). However, significant group x timepoint interactions were identified at T2 (estimate −0.116, 95% CI −0.219 to −0.013, p = 0.028) and T3 (estimate −0.138, 95% CI −0.0242 to −0.033, p = 0.010), indicating that this time-dependent increase was attenuated in the VO group, resulting in lower PBR 5–9 μm compared to the control group at these timepoints. Group and timepoint interactions for PBR are illustrated in Figure 3. Individual PBR values for each group and timepoint are presented as Supplementary material 1. Individual trajectories for PBR 5–25 μm are presented in Figure 4 and for PBR 5–9, 10–19 and 20–25 μm as Supplementary material 4.
[图：Changes in perfused boundary regions over time in vessels with diameters of 5–25 μm (A) , 5–9 μm (B) , 10–19 μm (C) , an]
[图：Individual changes in perfused boundary regions over time in vessels with diameters of 5–25 μm in 20 healthy anesthetize]
The median (IQR) count for major (10 μm) vascular segments with diameters 5–25 μm was 1520.05 (356.50). Within this range, median counts for the 5–9 μm, 10–19 μm, and 20–25 μm subgroups were 910.32 (348.75), 566.91 (172.83), and 42.82 (33.83), respectively (Figure 5).
[图：Distribution of major vascular segment counts by vessel diameter (5–25 μm) in healthy anesthetized cats. Bars represent ]
Packed cell volume fluctuated significantly over time, with mixed effects modeling identifying a significant effect of timepoint. Relative to baseline, model-estimated PCV was lower at T1 (estimate −3.350, 95% CI −4.863 to −1.837, p < 0.001), and higher at T2 (estimate 1.950, 95% CI 0.437 to 3.463, p = 0.012) and T3 (estimate 2.234, 95% CI 0.699 to 3.767, p = 0.005). Significant group x timepoint interactions were identified at T1 (estimate −4.200, 95% CI −6.339 to −2.061, p < 0.001) and T2 (estimate −2.554, 95% CI −4.709 to −0.399, p = 0.021), indicating lower PCV in the VO group compared to the control group at these timepoints.
Similarly, TS varied over time, with mixed-effects modelling identifying a significant effect of timepoint. Relative to baseline, model-estimated TS was lower at T1 (estimate −2.200, 95% CI −4.292 to −0.108, p = 0.04) and higher at T2 (estimate 2.600, 95% CI 0.508 to 4.692, p = 0.016). Significant group x timepoint interactions were identified at T1 (estimate −12.450 95% CI −15.409 to −9.491, p < 0.001) and T2 (estimate −5.812, 95% CI −8.793 to −2.831, p < 0.001), indicating lower TS values in the VO group compared to the control group at these timepoints.
Packed cell volume showed positive associations with VVD (estimate 5.747, 95% CI 1.956 to 9.537, p = 0.003) and %RBC filling (estimate 0.004, 95% CI 0.002 to 0.006, p < 0.001), and negative associations with PBR 5–25 μm (estimate −0.011, 95% CI –0.018 to −0.004, p = 0.003) and PBR 10–19 μm (estimate −0.013, 95% CI −0.021 to −0.005, p = 0.001). Correlation coefficients for PCV and GlycoCheck™ parameters are presented in Table 2 and, although statistically significant, were overall weak. Full fixed-effect results from all linear mixed-effects models are presented as Supplementary material 5.
Table 2. Correlation between packed cell volume and GlycoCheckTM parameters using Pearson’s test (except for perfused boundary region 20–25 μm).
| Parameter | r | 95% CI | p – value |
|---|---|---|---|
| VVD | 0.235 | 0.081—0.378 | 0.003 |
| %RBC filling | 0.277 | 0.126—0.416 | < 0.001 |
| Median P50 | −0.071 | −0.225 — 0.088 | 0.382 |
| PBR 5–25 μm | −0.238 | −0.381 — −0.0841 | 0.003 |
| PBR 5–9 μm | −0.102 | −0.255 — 0.056 | 0.203 |
| PBR 10–19 μm | −0.255 | −0.396 — −0.102 | 0.001 |
| PBR 20–25 μm | τ = −0.111 | - | 0.048 |
Discussion
Results from this study suggest that rapid IV administration of 30 mL/kg of buffered balanced crystalloid solution does not decrease estimated EG thickness in healthy cats. However, the PBR 5–25 μm, PBR 5–9 μm and PBR 10–19 μm, increased 3 h after intervention, suggesting a transient decrease in EG thickness potentially secondary to anesthetic effects. Interestingly, the PBR of small vessels (5–9 μm) was lower over time in the VO group at later timepoints, suggesting these effects might have been mitigated by the administration of IV fluids.
Transient volume overload did not result in increases in PBR in this group of healthy cats. This aligns with findings from a previous study in healthy cats from our group, which showed no increase in PBR following a fluid challenge (10 mL/kg over 20 min) using either Hartman’s solution or 6% hydroxyethyl starch 130/0.4 (30). Similar results have been reported in anesthetized rabbits, where only a hypertonic saline bolus (5 mL/kg of 10% NaCl) increased PBR, while volume loading with Lactated Ringer’s solution (30 mL/kg/h) had no effect (17, 36).
In canine haemorrhagic shock models, PBR generally remained unchanged during haemorrhage and autotransfusion in the absence of IV fluids (47). However, one study showed increased PBR following resuscitation with 30 mL/kg of LRS or 7.5 mL/kg of hydroxyethyl starch compared to natural or synthetic blood products (13). Likewise, in septic children, fluid boluses with unbalanced versus balanced crystalloids led to PBR increases 6 h post-intervention (48). Taken together, these studies suggest that IV fluids alone may not alter PBR in healthy animals, while more severe interventions (e.g., haemorrhage) or underlying disease states (e.g., sepsis) may predispose to detectable EG alterations. Nevertheless, differences in study design, species, and methodology limit direct comparison or translational conclusions.
The PBR 5–25 μm, 5–9 μm and 10–19 μm increased over time in the control group and returned to baseline by 24 h in this study. These changes cannot be attributed to IV fluid-associated VO. Therefore, alternative explanations should be considered. Anesthesia-related changes in blood volume are well documented in the literature and primarily attributed to fluid shifts from the interstitial into the IV space in response to anesthesia-induced hypotension (18, 49, 50). In addition, most anesthetics, including propofol, cause splenic relaxation in dogs and cats, leading to sequestration of RBC and transient reductions in PCV (20, 51–55). In the present study both PCV and TS decreased transiently across both groups, supporting anesthesia-related changes in blood volume, particularly during the initial longer anesthetic period (T0 and T1 combined).
Whether such changes in blood volume could be associated with EG damage remains uncertain. A recent study examining plasma volume changes in hypotensive versus normotensive human surgery patients demonstrated a tenfold increase in plasma volume in the hypotensive group. However, this was not associated with increased NP production or EG shedding (18), suggesting that hypotension-induced volume expansion does not damage the EG. More importantly, although MAP decreased at T1 in the present study, only one cat experienced mild hypotension. Therefore, anesthesia-related hypotension and its associated plasma volume expansion are unlikely to explain the observed changes in PBR.
More broadly, hemodynamic anesthetic variables (HR, MAP and temperature) were not associated with PBR, consistent with previous studies in healthy cats using similar anesthetic protocols (28, 30). Nonetheless, neither study was designed to investigate anesthesia-related effects. Blood pressure was monitored non-invasively primarily for anesthetic safety. Proper exploration of hemodynamic anesthetic effects on the EG would require a more targeted setup, including direct blood pressure measurement, cardiovascular point-of-care ultrasound, and collection of additional hemodynamic parameters.
Another anesthesia-related factor worth considering is the choice of anesthetic agent itself, particularly propofol, which may have direct effects on the endothelium and its glycocalyx. Studies using SDFV have reported effects of propofol on the microcirculation in people, generally showing trends toward decreased perfused vascular density and other measures of microcirculatory function over time or compared with other sedatives and anesthetics (56–60). It is therefore possible that propofol may also affect the feline microcirculation and consequently influence estimated EG thickness. However, these studies included heterogeneous cohorts, predominantly critically ill patients, and varied substantially in study design, precluding direct comparison with the present study. Furthermore, they did not assess estimated EG thickness, as the methodologies used did not support a PBR calculation.
Outside of microcirculatory imaging, both in vitro and in vivo experimental studies have demonstrated endothelial cell damage and EG shedding after propofol overdose (61–63). In contrast, clinical studies comparing the effects of propofol and volatile anesthetics in people have not consistently demonstrated increased EG shedding following propofol administration (64–66). These effects appear to be dose-dependent and time-related, and confounding factors such as underlying EG-damaging illnesses limit the conclusions that can be drawn from clinical data.
Propofol has been used in previous studies evaluating PBR in cats (28, 30), though these studies did not include repeated propofol administration within 3 h. Namely, in a previous study from our group evaluating effects of an IV fluid challenge on GlycoCheck™ parameters in healthy cats, increases in PBR 10–19 μm were found at the end of infusion, compared to baseline (30), suggesting comparable effects. Considering the relative feline “sensitivity” to propofol-related oxidative stress (67–69), longer half-life and slower metabolism of phenolic compounds in cats (51), it is possible that propofol may have contributed to a transient reduction in EG thickness, especially considering the repeated administrations between T0 and T2. Further studies comparing effects of repeated administration and use of different anesthetics are necessary to investigate this possibility.
Surprisingly, the PBR 5–9 μm was lower at later timepoints (T2 and T3) in the VO group. It is possible that IV fluid administration mitigated some of the negative effects, specifically in the smallest analyzed vessels. A previous study from our group, investigating the immediate effects of an IV fluid challenge using Hartman’s solution or hydroxyethyl starch in healthy cats similarly reported decreases in PBR in the hydroxyethyl starch group, though in larger vessels (10–19 μm) (30). This may have resulted from dilution of anesthetic agents or potentially from “rehydrating” the EG. Further studies specifically designed to investigate anesthetic influences on the EG are warranted to support this hypothesis.
This study demonstrated weak associations between PCV and GlycoCheck™ parameters. Associations between PCV and PBR have been previously found in adult and neonate human patients (70, 71). These findings are broadly consistent with a previous study from our group, which evaluated the effects of a fluid challenge on GlycoCheck™ parameters in cats and found that VVD and %RBC filling decreased and PBR increased with decreasing PCV (30).
Although these associations are weak, they warrant further investigation given that all GlycoCheck™ parameters rely on identification of RBCs against a contrasting background, with PBR calculations dependent on the presence of vessels meeting criteria for “valid” density and adequate % RBC filling (31). If a consistent influence of PCV on PBR is confirmed with specifically designed studies, adjustment for PCV may be warranted when interpreting PBR measurements, particularly in clinical research.
Interestingly, most GlycoCheck™ parameters returned to baseline by 24 h, suggesting that any EG disturbance was either transient or rapidly resolved. However, this contrasts with current understanding of EG shedding, turnover, and repair (72). While EG shedding can occur rapidly (within minutes to hours) following a significant stimulus, the biosynthesis of new components typically requires several hours to days, depending on the molecule and mechanism involved (72, 73). Therefore, the return to baseline at 24 h may represent dimensional recuperation or functional rehydration, rather than full structural restoration. It is conceivable that IV fluids and anesthetic stimuli caused a transient EG “desiccation” or “contraction,” rather than substantial loss of EG components. Although incomplete structural recovery could theoretically challenge the adequacy of the 7-day washout period, normalization of GlycoCheck™ measurements within 24 h makes a meaningful carryover effect unlikely. Unfortunately, EG shedding biomarkers were not assessed in this study. Given their limitations in detecting fluid-induced changes to the EG, the reliability of such additional analyses remains uncertain.
In considering the potential stimulus for EG change, the selected IV fluid dose warrants further reflection. Beyond representing empirically “high” IV fluid rate, the dose chosen for this VO model was based on a previous study evaluating cardiac ultrasound and caudal vena cava parameters in healthy cats with induced VO (74). In the referenced study, rapid infusion of 30 mL/kg of Hartman’s solution resulted in a mild but statistically significant increase in cardiac chamber and caudal vena cava size, suggesting cardiac stretch on imaging. However, whether this stretch was sufficient to trigger NP production remains unclear. By comparison, a canine study using a much higher overall dose over a longer time (90 mL/kg of isotonic crystalloids over 1 h) demonstrated marked increases in NP (75), confirming that IV fluids can stimulate NP release under certain conditions. Considering the feline predisposition to volume overload, the highest fluid rate considered ethically and physiologically acceptable was used to potentially affect the EG without causing harm. However, NP concentrations were not measured in this study, so any involvement of NP in the EG modulation could not be confirmed. Further investigations should incorporate species-specific volume kinetic data to guide fluid dosing strategies in cats for studies related to microcirculation or other IV fluid research (25, 76).
Although not statistically significant, there was notable variability in PBR measurements in this study, especially at baseline, despite the cross over design and averaging three measurements as per current recommendations in people (37, 38). This is consistent with previous findings in a large cohort of healthy cats (28). Such variability may reflect both biological variation and limitations associated with applying the GlycoCheck™ system to feline microcirculation. The software was developed for human use and has not been specifically adapted for use in cats, therefore species-specific characteristics of the feline microcirculation may influence some aspects of its derived parameters. Differences in capillary bed density have been reported across mammalian species (77) including studies using SDFV to assess perfused vessel density in cats, dogs, and horses (78–81). Subjectively, cats also appear to have less dense capillary beds when examined using SDFV (personal observation). Importantly, VVD has been reported to be considerably lower in cats compared to dogs and people (28, 31, 82). As VVD is a source parameter used in the calculation of PBR, lower vessel density may reduce the number of datapoints available for analysis, potentially increasing measurement ‘noise’ and contributing to PBR variability.
Intra- and inter-individual variability in GlycoCheck™ measurements is not inherent to cats. As discussed above, variability has also been specifically investigated in healthy and critically ill people (37, 38, 83). Although averaging multiple measurements improves reproducibility, it does not eliminate variability. Inter-operator variability has also been reported, although some studies have demonstrated acceptable transferability between operators (83, 84).
Beyond measurement variability, GlycoCheck™ has several methodological limitations. Although image analysis is automated, image acquisition remains operator dependent and susceptible to factors such as pressure artefacts. Measurements may also be influenced by physiological and examination conditions (38), highlighting the importance of standardized acquisition protocols. Differences between software versions may limit direct comparison between studies (38). Importantly, PBR is a surrogate of EG thickness rather than a direct measurement of EG structural integrity. It therefore does not directly quantify loss of structural components or more importantly assess EG function. Some of these limitations are not specific to GlycoCheck™ but are inherent to sublingual microcirculatory imaging using SDFV and should be considered when interpreting and comparing findings obtained using this methodology (85–87).
Collectively, these limitations suggest that evaluating individual longitudinal changes in PBR may be challenging in feline patients using the current version of the software. Nevertheless, the GlycoCheck™ may remain useful for investigating group-level changes in estimated EG thickness, provided these limitations are acknowledged, particularly given the lack of alternative methods for non-invasive investigation of the EG in cats.
Additional limitations in this study include the relatively small sample size and somewhat arbitrary timepoints. Notably, T0 and T1 measurements were obtained during the same anesthetic episode, whereas separate, shorter anesthesia was used at T2 and T3, potentially resulting in differences in the physiological context between measurements. The T0–T1 anesthesia was the longest of the study. However, its duration remained relatively short (~20 min). Furthermore, no relevant changes in PBR were observed at T1, the timepoint most likely to have been affected by the longer anesthetic duration.
Butorphanol and gabapentin, selected for their mild sedative and cardiovascular effects were administered primarily to facilitate IV catheter placement, and therefore not repeated at T2 and T3, although residual gabapentin effects may have persisted at T2. As their effects on the feline microcirculation are unknown, differences in their administration between timepoints represent a potential confounding factor that cannot be excluded.
Lastly, this was a small-scale experimental study conducted in the context of limited prior data in this species, with much of the existing evidence originating from the same research group. The findings should therefore be interpreted as exploratory, providing a basis for further investigation and independent replication rather than definitive conclusions or direct clinical application.
Conclusion
This study did not demonstrate decreases in estimated EG thickness after transient, mild VO. However, changes in PBR over time may suggest potential anesthesia-related effects. As short anesthesia is inevitably required for SDFV imaging in small animals, further studies specifically designed to optimize anesthetic protocols are essential. Changes in PCV were associated with alterations in GlycoCheck™ parameters, and the clinical significance of these associations warrants further investigation. Given the known predisposition of cats to VO, the effects of IV fluids on the feline microcirculation should continue to be explored, ideally through study designs informed by a more detailed understanding of volume kinetics in this species.