Observer Based Objective Pain Quantification in Sickle Mice Using Grimace Scoring and Body Parameters

Pain often starts in infancy and continues throughout life in sickle cell disease (SCD). Subjects typically quantify pain by themselves, which may lead to overtreatment or under-treatment of pain. Patient's quantification of their own pain may also be influenced by external factors (mood, stress, sleep, etc.). Therefore, we examined the possibility of developing observer-based quantifiable measures of pain using facial features to obtain a mouse grimace scale (MGS) and body dimensions as measures of pain in HbSS-BERK (sickle) and control HbAA-BERK (control) mice.

Methods: Mice aged 6-8 months, were placed in transparent cubicles on a Plexiglas surface at room temperature (RT) or on a cold plate (CP) at 4.5¡C. Mice were acclimated and digitally videotaped at RT for 5 min to capture action units (AU) and then videotaped on the cold plate for 5 min. Individual frames of clear, unobstructed headshots were 'grabbed'. The experiment was repeated after 24h. Four AUs, (orbital tightening, nose bulge, cheek bulge, and ear position) were assessed for each mouse and condition using 4 randomized pictures of each by 3 blinded 'coders'. For MGS, intensity of each AU was scored as follows: '0' = absence; '1'= moderate and '2' = severe. Average AU scores from the 3 coders were averaged for each AU. Cumulative pain calculations were made by averaging all AUs as described (Langford et al., 2010, Nature Methods). For length and body parameter measurements, one camera was placed with a top down view of the mouse with a Plexiglas graph template (1x1cm squares) in the background and another camera was placed in front of the mouse to capture the back curvature. Length was calculated using the graph template. To analyze spinal curvature, best-fit ellipses were constructed from points plotted on the arch of the spine of the mouse. Points were placed between the neck where the arch begins and at the end of the arch, near the tail. Two ellipses were generated for each mouse and difference between maximum and minimum axes were recorded. Smaller differences in max and min axes indicated more pain and a more circular shape and larger differences indicated less pain and a more elliptical shape. To validate facial AU data, sensitivity to 1.0 g Von Frey filament was used.

Results: We observed significantly higher MGS in pain totals for control on CP vs. control at RT (P<0.01), sickle CP vs. sickle RT (P<0.05), and sickle CP vs. control CP (P<0.001), suggesting that both control and sickle mice exhibited a 'pain face' on the cold plate, but sickle mice showed increased intensity as compared to control on the cold plate. Curvature differences were significant in control on CP vs. control at RT (P<0.01), and sickle on CP vs. sickle at RT (P<0.01). Sickle mice showed an average 55% increase in curvature compared to 39% in control mice on a cold plate vs. RT for each (P<0.05), suggesting that sickle mice exhibit more pain in response to cold environment. Complementary to the curvature, sickle mice showed a significant decrease in length on a cold plate vs. RT (24%; P<0.01), but control mice did not show a significant difference in length. The Von Frey test demonstrated increased hyperalgesia in sickle mice as compared to control at RT (P<0.0001). Together, these data show that facial expressions and body dimensions provide quantifiable means to assess pain intensity in sickle mice. Thus facial characteristics, length and curvature can be adapted to assess pain in sickle patients particularly in children, who may not be able to express pain scores accurately in an unbiased way.