Method parameters’ impact on mortality and variability in mouse stroke experiments: a meta-analysis

Although hundreds of promising substances have been tested in clinical trials, thrombolysis currently remains the only specific pharmacological treatment for ischemic stroke. Poor quality, e.g. low statistical power, in the preclinical studies has been suggested to play an important role in these failures. Therefore, it would be attractive to use animal models optimized to minimize unnecessary mortality and outcome variability, or at least to be able to power studies more exactly by predicting variability and mortality given a certain experimental setup. The possible combinations of methodological parameters are innumerous, and an experimental comparison of them all is therefore not feasible. As an alternative approach, we extracted data from 334 experimental mouse stroke articles and, using a hypothesis-driven meta-analysis, investigated the method parameters’ impact on infarct size variability and mortality. The use of Swiss and C57BL6 mice as well as permanent occlusion of the middle cerebral artery rendered the lowest variability of the infarct size while the emboli methods increased variability. The use of Swiss mice increased mortality. Our study offers guidance for researchers striving to optimize mouse stroke models.


Category refinement
In the Other strain category, the following were included for hypothesis 1A, 2A and 3A: Patchy fur XO, Patchy fur XX, FVB/N, factor V Leiden, Nestin-Cre, SCID and strain not specified. For hypothesis 4A the category also included C3H mice. In the hypotheses 1B, 2B, 3B and 4B analyses the variants ddY, 129S, Mixed C57BL6/129, BALB/c and Strain not specified constituted the Other strain category. All strains originating from Swiss mice (e.g. Swiss Webster, ICR, CD-1 and Kunming) were included in the Swiss category.
Two separate female categories could be identified in the Sex category, one with ovariectomized females and one with females not explicitly ovariectomized. In the Others category, groups with mixed sexes and not specified sexes were included. In addition, one group with castrated males were added to the Others category for hypotheses 1A, 2A, 3A and 4A while ovariectomized females were added for hypotheses 1B, 2B, 3B and 4B.
The number of anesthesia categories were considerably reduced during the refinement process because of similarity between the different variants. All inhalation anesthesia (isoflurane, halothane, sevoflurane, enflurane) formed one group, chloral hydrate alone or in combination formed the second, ketamine alone or in combination the fourth, different members of the benzodiazepine (midazolam, diazepam) and barbiturate (methohexital, thiopenthal, pentobarbital) groups of substances the fifth, tribromoethanol the sixth and finally a category for groups where the anesthesia method was not specified. For hypothesis 4A no control groups represented the Tribromoethanol category. Further, for hypothesis 1B, 2B, 3B Chloral hydrate and Tribromoethanol formed an Other anesthesia category and no unspecified anesthesia methods were included. Finally, for hypothesis 4B no control groups represented the Tribromoethanol or the Anesthesia not specified categories and these were consequently omitted.
The variable Blood gases/O2 saturation analyzed was initially extracted as three separate categories (Blood pH analyzed, Blood oxygen saturation analyzed and Blood carbon dioxide analyzed) but these were fused since they were so highly correlated.
Regarding the variable Type of middle cerebral artery procedure, all intraluminal filament methods were put together in one category. Note that the effect of different kinds of filaments were analyzed separately in the filament subanalysis. Techniques where the middle cerebral artery was occluded mechanically after craniectomy (clip or electrocauterization) were pooled in a category labelled Direct, mechanical. The last two categories were formed by ischemia induced by photothrombosis and different kinds of emboli or clot techniques. For hypothesis 1B, 2B and 3B only two categories were represented, namely occlusion with intraluminal filament or direct craniectomy technique.
Occlusion duration, which theoretically could be seen as a continuous variable was converted into categories to enable comparison with the permanent occlusion option. The category Short transient was defined as occlusion less than 60 minutes while more than 60 minutes was considered Long transient. Only methods where actions were taken to ensure reperfusion, such as withdrawing the filament, were considered transient. The last category, Permanent, included groups were the middle cerebral artery occlusion was induced but not actively reversed.
Twelve types of lesion evaluation were reduced to four main categories. The most frequently used technique, 2,3,5-triphenyltetrazolium chloride staining, defined the first category, while various acidic/basic staining techniques (such as hematoxylin/eosin, thionine and cresyl violet) formed the second. Silver stain were put in a separate category whereas different immunohistochemical methods, neutral red stain and groups without any specified lesion evaluation technique made up the Other evaluation type category.
Since a number of studies used more than one exclusion criteria, these were divided into four separate binary variables (Hemorrhage, Deficit score, Bad clinical condition and Other exclusion criteria) instead of one variable with four categories. This way one control group could be registered in more than one category. In the Other exclusion criteria category, criteria based on hypothermia, oxygen saturation and size of lesion were put. The use of laser Doppler surveillance could also be considered an exclusion criteria, given it is assumed that in studies where this technique was employed, exclusion based on the extent of cerebral perfusion was practiced.
Since the intraluminal filament method is by far the most common procedure for inducing cerebral ischemia (430 out of the total 500 control groups in the current study employed this technique), a separate analysis was performed for these studies (hypotheses 4A and 4B). For this subanalysis, two additional variables were introduced into the model: Filament coating type and Filament insertion vessel. The variable Filament coating type initially comprised ten categories but they were reduced to five after merging of similar coating types. Silicone formed a separate category, just like Poly-L-lysine did. Different types of glue as well as resin were put in the Glue category while filaments only blunted were considered Uncoated. The Other category contained filaments coated with Teflon or cases where coating (or lack thereof) was not specified. For hypothesis 4B, only three categories were used since both glue and poly-L-lysine coating were added to the Other coating category. This variable was not included in the analyses regarding effect on the mortality outcome measure (hypothesis 1B, 2B, 3B and 4B) since it seems illogical that evaluation type after sacrifice would affect mortality in any way.

Definition of quantitative variables
When data regarding the variable Weight was provided as a range (e.g. 20-25 g) the mean value of that range was calculated. For some control groups, weight was not provided at all and to replace these missing values, the mean value of all other control groups was used. Time after focal ischemia for evaluation of damage was defined as the time from obstruction of the middle cerebral artery until sacrifice. The first outcome variable, Infarct size coefficient of variation was calculated by dividing the standard deviation of the infarct volume by the mean infarct volume. Such definition is strict and applicable irrespective of how the infarct size was presented: as percentage of a hemisphere or as an absolute value in cubic millimeters. The second outcome, Mortality rate, was defined as the unintended mortality, from induction of focal cerebral ischemia until sacrifice, as a percentage of the whole control group. This variable was unfortunately only provided for less than 20% of the control groups, why the predictor variables in the analyses regarding this outcome had to be regrouped in some situations to avoid categories represented by less than 5 control groups (see above).  Variables excluded by the statistical software in the backward step of the analysis, due to low explanatory value: Type of middle cerebral artery occlusion procedure, Occlusion duration, Sex, Exclusion based on other criteria, Exclusion based on bad clinical condition, Time after ischemia for evaluation of damage, Blood gases/O2 saturation analyzed, Blood hemoglobin analyzed, Weight, Blinding of infarct size measurement procedure, Awakening during occlusion.