Validation of severity assessment and refinement in neuroscience rat models of subarachnoid hemorrhage and cerebrovascular research

急性脑损伤动物模型在严重程度评估方面提出了特殊的挑战。脑损伤本身可能导致行为改变，如感觉和运动缺陷、认知缺陷或冲动不足，这些都可能影响健康测试的结果，这些测试通常用于严重程度评估。此外，病变起始需要进行侵入性手术，这增加了病理生理过程引起的负担。在前两个资助期内，我们使用了两种大鼠蛛网膜下腔出血（SAH）模型（细丝穿孔；大池血注射）来确定最有用的幸福感、痛苦或负担减少的措施，适用于大鼠急性脑损伤模型和一般脑血管研究模型。由于SAH模型中相对较高的死亡率最能反映人类的病理生理，我们额外关注识别有快速达到人道终点或过早死亡风险的动物。在数据科学方法的支持下，我们已经确定了可以在动物的日常护理中实施的核心结果参数（体重和自愿跑轮，补充经典的临床评分）。野外活动和挖洞被认为是进一步有用的参数，它们的评估更耗时，并且在动物之间显示出更多的异质结果。在第三个资助期，基于该数据集，我们将对急性和亚急性期的两种评估的SAH模型进行跨模型分析，旨在对由此产生的SAH的严重程度进行分级，以及在假动物中各自的手术入路。在所有神经科学小组的跨实验室和跨模型分析中，这些核心参数将在研究的时间尺度内验证其信息价值，稳健性和通用性。我们还将在实验室间的方法中具体处理假手术对照组的暴露。除了进一步阐述数据科学方法外，还将应用基于算法的相对严重程度评估（RELSA）多维方案，用于在实验过程中连续定量监测痛苦，以及开发的生物信息学工作流程，用于在研究单元内不同模型之间进行比较严重程度评估的复合测量方案。基于从家庭监测和植入遥测设备获得的行为和重要参数（活动，温度，心率）数据，将通过支持向量机分类和无监督k-means聚类进行严重程度分级和早期人道端点检测的细化。最后，对于基于证据的镇痛改进，卡洛芬联合持久局部麻醉的多模式方法将被评估为一种替代丁丙诺啡的非阿片类药物策略。

Animal models of acute brain injury pose a special challenge regarding severity assessment. The brain lesion itself may cause behavioural alterations like sensory and motor deficits, cognitive deficits, or impulse deficiency, which all may influence the outcome of tests for well-being, which are generally applied for severity assessment. In addition, invasive surgery procedures are needed for lesion initiation, which adds to the burden induced by the pathophysiological processes. Within the first two funding periods we have used two models of subarachnoid hemorrhage (SAH) in rats (filament perforation; cisterna magna blood injection) to identify most useful measures of reduced well-being, distress or burden, applicable in rat acute brain injury models and models for cerebrovascular research in general. Due to the comparably high death rate in the SAH models best reflecting the pathophysiology in humans, we additionally focused on identification of animals at risk for rapidly reaching humane endpoint or premature death. Supported by data science approaches, we have identified core outcome parameters that can be implemented in the everyday care of the animals (body weight and voluntary wheel running, supplementing the classical clinical score). Activity in open field and burrowing figured out as further useful parameters, with their assessment being more time-consuming and showing more heterogenous results between animals. In the third funding period, based on this dataset, we will perform cross-model analysis of the two assessed SAH models in the acute and sub-acute phase aiming at grading of severity of the resultant SAH as well as of the respective surgical approach in sham animals. In a cross-laboratory and cross-model analysis within all neuroscience groups, these core parameters will be validated for their informative value, robustness and generalisability within the time scales of the studies. We will also specifically address the exposure of sham operated control groups in an inter-laboratory approach. In addition to further elaborate data science approaches, the algorithm-based multidimensional scheme of RELative Severity Assessment (RELSA) for continuous quantitative monitoring of distress during the course of the experiment, as well as the developed bioinformatics work flow of composite measure schemes for comparative severity assessment between the different models within the research unit will be applied. Based on data of behavioural and vital parameters (activity, temperature, heart rate) achieved from homecage monitoring and implanted telemetric devices, severity grading and refinement of early humane endpoint detection will be performed by support vector machine classification with unsupervised k-means clustering. Finally, for evidence-based analgesia refinement, the multimodal approach of carprofen in combination with long-lasting local anaesthesia will be assessed as an alternative, non-opioid based strategy to buprenorphine.