Schilling Achim, Krauss Patrick, Gerum Richard, Metzner Claus, Tziridis Konstantin, Schulze Holger
Experimental Otolaryngology, ENT Hospital, Head and Neck Surgery, Friedrich-Alexander University Erlangen-Nürnberg, Erlangen, Germany.
Biophysics Group, Department of Physics, Center for Medical Physics and Technology, Friedrich-Alexander University Erlangen-Nürnberg, Erlangen, Germany.
Front Behav Neurosci. 2017 Oct 18;11:198. doi: 10.3389/fnbeh.2017.00198. eCollection 2017.
An increasingly used behavioral paradigm for the objective assessment of a possible tinnitus percept in animal models has been proposed by Turner and coworkers in 2006. It is based on gap-prepulse inhibition (PPI) of the acoustic startle reflex (ASR) and usually referred to as GPIAS. As it does not require conditioning it became the method of choice to study neuroplastic phenomena associated with the development of tinnitus. It is still controversial if GPIAS is really appropriate for tinnitus screening, as the hypothesis that a tinnitus percept impairs the gap detection ability ("filling-in interpretation" is still questioned. Furthermore, a wide range of criteria for positive tinnitus detection in GPIAS have been used across different laboratories and there still is no consensus on a best practice for statistical evaluation of GPIAS results. Current approaches are often based on simple averaging of measured PPI values and comparisons on a population level without the possibility to perform valid statistics on the level of the single animal. A total number of 32 animals were measured using the standard GPIAS paradigm with varying number of measurement repetitions. Based on this data further statistical considerations were performed. We here present a new statistical approach to overcome the methodological limitations of GPIAS. In a first step we show that ASR amplitudes are not normally distributed. Next we estimate the distribution of the measured PPI values by exploiting the full combinatorial power of all measured ASR amplitudes. We demonstrate that the amplitude ratios (1-PPI) are approximately lognormally distributed, allowing for parametrical testing of the logarithmized values and present a new statistical approach allowing for a valid and reliable statistical assessment of PPI changes in GPIAS. Based on our statistical approach we recommend using a constant criterion, which does not systematically depend on the number of measurement repetitions, in order to divide animals into a tinnitus and a non-tinnitus group. In particular, we recommend using a constant threshold based on the effect size as criterion, as the effect size, in contrast to the -value, does not systematically depend on the number of measurement repetitions.
2006年，特纳及其同事提出了一种越来越常用的行为范式，用于在动物模型中客观评估可能存在的耳鸣感知。它基于听觉惊吓反射（ASR）的间隙前脉冲抑制（PPI），通常被称为GPIAS。由于它不需要条件作用，因此成为研究与耳鸣发展相关的神经可塑性现象的首选方法。GPIAS是否真的适用于耳鸣筛查仍存在争议，因为耳鸣感知会损害间隙检测能力的假设（“填补解释”）仍受到质疑。此外，不同实验室使用了广泛的GPIAS阳性耳鸣检测标准，并且对于GPIAS结果的统计评估的最佳实践仍未达成共识。当前的方法通常基于测量的PPI值的简单平均以及群体水平上的比较，而无法在单个动物水平上进行有效的统计。使用标准GPIAS范式对总共32只动物进行了测量，测量重复次数各不相同。基于这些数据进行了进一步的统计考量。我们在此提出一种新的统计方法，以克服GPIAS的方法学局限性。第一步，我们表明ASR振幅不是正态分布的。接下来，我们通过利用所有测量的ASR振幅的全部组合能力来估计测量的PPI值的分布。我们证明振幅比（1 - PPI）近似对数正态分布，允许对对数化值进行参数检验，并提出一种新的统计方法，允许对GPIAS中PPI变化进行有效且可靠的统计评估。基于我们的统计方法，我们建议使用一个不系统依赖于测量重复次数的恒定标准，以便将动物分为耳鸣组和非耳鸣组。特别是，我们建议使用基于效应大小的恒定阈值作为标准，因为与p值不同，效应大小不会系统地依赖于测量重复次数。
