Abstract
Objective:
To compare the Osteoarthritis Research Society International (OARSI) and Articular Cartilage Structure (ACS) grading schemes applied to multiple and single sections, along with additional histologic measures, in two mouse models of OA.
Methods:
Six coronal histologic stifle joint sections were collected from 40 C57BL/6J mice, including aged mice with spontaneous OA (approximately 18 months of age; n=15) and young (12-week-old) mice that either underwent destabilization of the medial meniscus (DMM) surgery (n=15) or sham surgery (n=10). Sections were evaluated with the standard OARSI (0–6) scheme, a modified OARSI scheme, the ACS (0–12) scheme, histomorphometry of cartilage and bone, and scoring of osteophytes (0–3) and synovial hyperplasia (0–3). Principal components analysis (PCA) was used to determine the features explaining the greatest variability among the sections.
Results:
The grading schemes performed similarly when applied to a single mid-coronal section or six total coronal sections per joint. OARSI grading produced similar results when applied to hematoxylin and eosin or toluidine blue-stained sections. Aged mice had higher severity scores in the LTP than DMM mice (mid-coronal OARSI grade aged=2.3 and DMM= 1.1, p=0.0006; ACS grade aged=4.1 and DMM=1.6, p=0.0024). PCA resulted in retention of four factors that accounted for 78.4% of the total variance. Factor 1 (36.4%) included the OARSI grade, ACS grade, Toluidine blue grade, articular cartilage area and thickness and the osteophyte grade.
Conclusions:
Grading of a single mid-coronal section using either the OARSI or ACS schemes combined with osteophyte and histomorphometric measures can consistently define OA severity in mice.
Keywords: preclinical, osteoarthritis, grading, mouse
Introduction
Mouse models of osteoarthritis (OA) are commonly used to study the role of specific genes that can be manipulated in genetically engineered mice or for pre-clinical studies of new therapeutics. OA severity in these studies is commonly determined histologically using a variety of grading schemes 1–6. However, the results of murine OA studies are often difficult or impossible to reproduce 7–10. As a result, preclinical efficacy of therapeutics for osteoarthritis often fails to translate to clinically effective treatments 7,8,11. One reason for this failure of translation may be the inconsistency with which grading schemes are applied and reported in murine OA studies.
The Osteoarthritis Research Society International (OARSI) published a grading scheme for human OA in 2006 and an updated species-specific set of schemes in 2010 as a part of the OARSI histopathology initiative special issue 2,4,12. The OARSI grading schemes represent more reproducible and species-specific systems than the Mankin score or the Histologic/Histochemical Grading System (HHGS), both of which were designed for humans but were adapted to animal models 2,4,13–15. However, evidence is limited regarding which scheme(s) to use for a particular study and how to apply them to various OA models to provide the best and most consistent determination of OA severity.
The widely used OARSI grading scheme for mice focuses on lesions involving articular cartilage and uses a scale of severity ranging from 0–6 4. This method was developed based on a surgically-induced model of OA and was validated with a set of 10 static images of the medial tibial plateau (MTP) and the medial femoral condyle (MFC). The authors suggested scoring of the entire articular surface using up to 13–16 coronal sections per joint, harvested at 80 um intervals and stained with either safranin-O/fast-green or toluidine blue to identify loss of proteoglycans from the cartilage matrix 4.
The articular cartilage structure (ACS) grading scheme provides a similar method of assessment of the articular cartilage, except that it uses hematoxylin and eosin (H&E) stained sections, with grades ranging from 0–12 depending on the width and depth of the lesions 5,16,17. Notably, the ACS scheme was developed based on assessment of stifle joints with either naturally occurring (aged) or surgically-induced OA using the destabilized medial meniscus (DMM) model 5,18. A goal of the ACS scheme is to use it in conjunction with scoring of osteophytes, synovium, and histomorphometric measures of cartilage and subchondral bone to provide a comprehensive assessment of the joint as a whole 17. In order to limit expense and time in processing multiple sections, a representative mid-coronal section is used (identified based on specific tissue landmarks), rather than the 13 or more sections suggested for the OARSI scheme 8. Importantly, a direct comparison of the OARSI and ACS schemes has not previously been reported.
The primary objective of the present study was to determine the ramifications of applying the ACS or OARSI grading schemes to either single mid-coronal or multiple sections to determine the outcome of a murine OA study using two models: younger mice with OA induced using DMM surgery and older mice with naturally occurring OA. The secondary objective was to compare the performance of the OARSI and ACS grading schemes in their ability to determine murine OA severity. Because the ACS scheme was developed for H&E stained sections, our third objective was to determine the relative importance of a specific stain on the ability to evaluate OA severity, evaluated by applying a modified OARSI scheme to H&E sections. In addition to ACS and OARSI scores applied to the articular cartilage, and in recognition of the fact that OA is a disease of the entire joint, other features of murine OA were also assessed, including osteophyte and synovial hyperplasia grading and histomorphometry of the cartilage and subchondral bone. A principal components analysis (PCA) was performed to assess the contribution of these semi-quantitative and quantitative measures to the OA severity and the association of the included features with one another (OARSI and ACS grades, osteophyte and synovial grades, histomorphometry features).
Methods
Animals
This work was approved by the Institutional Animal Care and Use committee at the University of North Carolina, Chapel Hill, NC, USA. Mouse colonies were maintained in a standard specific pathogen-free facility. Animals were housed in cages holding an average of 3–5 mice per cage with access to water and food ad libitum. The cohort for this study of 40 C57BL/6J background wild-type male mice included aged controls (~18 months of age; n=15) and young (12-week-old) mice that either underwent DMM surgery (n=15) or a sham surgery (n=10) and were sacrificed 8 weeks later (20 weeks of age). Group sizes were powered to detect a significant difference between sham and DMM mice in mean mid-coronal OARSI and ACS grades, using a mean difference of 2 with an estimated standard deviation of 2 for the OARSI scheme and a mean difference of 3 with an estimated standard deviation of 2.9 for the ACS scheme (power=0.8, α=0.05, 2-sided T-test). The DMM and sham surgeries were performed on the right knee in separate groups of mice as previously described 19,20.
Histology
Joints were collected and prepared for histology as previously described (see Supplementary Methods) 18. Joints were serially sectioned with collection of six representative 4-μm-thick sections at approximately 100 μm intervals throughout the joint, based on the size of the joints and the ability of six total sections taken at 80–120 μm intervals to span the extent of the joint where landmarks for orientation are evident (span of approximately 700 μm) and to include the mid-coronal region (see Supplementary Figure 1). Sections were stained with hematoxylin and eosin (H&E). One additional mid-coronal section was stained with toluidine blue.
Grading of cartilage
Historically, lesions in aging and DMM murine models of OA have been most severe at the tibial plateaus 18. Because the mid-coronal grades in the tibial plateaus were equal to or higher than those for the femoral condyles in the majority of individuals (Supplementary Table 1), further evaluation focused on the tibial plateaus. The ACS grading scheme was applied as previously described, except that it was applied to all 6 representative coronal H&E-stained sections through each joint (Figure 1). A modified OARSI grading scheme, using the presence of an area composed of chondrocyte cell death (CCD) within the articular cartilage without other structural changes to determine the grade of 0.5 as a substitute for loss of toluidine blue staining, was applied to the same 6 representative coronal H&E-stained sections (Figure 1). The OARSI grading scheme of 0–6 was also applied as previously published4 to a single toluidine blue stained mid-coronal section from all mice (toluidine blue section not available for one aged mouse). Grading of the MTP and LTP in each section was performed by both an inexperienced grader (trained veterinary student) and an experienced grader (AA; boarded veterinary anatomic pathologist). The interobserver variation is reported in the Supplementary Results. Toluidine-blue sections were also scored from 0–12 based on the extent of loss of staining as described 5.
Figure 1. Histologic comparison of hematoxylin & eosin (H&E) staining and toluidine blue staining of the articular cartilage with different grades of osteoarthritis.
(A) and (B) The medial tibial plateau of a sham mouse with intact articular cartilage and grades based on the ACS scheme and modified OARSI schemes = 0 (A; H&E) and standard OARSI scheme = 0 (B; toluidine blue). (C) and (D) The medial tibial plateau of a DMM mouse with an ACS grade = 2 (C; H&E), modified OARSI grade = 2 (C; H&E), and standard OARSI grade = 2 (D; toluidine blue). (E) and (F) The medial tibial plateau of a mouse after DMM with an ACS grade = 10 (E; H&E), modified OARSI grade = 4 (E; H&E), and standard OARSI grade = 4 (F; toluidine blue). (G) and (H) The medial tibial plateau of an aged mouse with complete loss of the articular cartilage across the full width of the plateau with an ACS grade = 12 (G; H&E), modified OARSI grade = 6 (G; H&E), and standard OARSI grade = 6 (H; toluidine blue). Connected arrows = articular cartilage, arrowheads = tidemark, * = meniscus.
The grading methods were defined as follows: sum grade – summed scores generated from six total sections; average grade – average score generated from six sections; highest grade – highest score selected from six sections; and mid-coronal grade – single score generated from a mid-coronal section chosen based on known landmarks 8.
Additional histologic and histomorphometric analysis
Abaxial osteophytes (0–3) and synovial hyperplasia (0–3) were graded as previously described (see Supplementary Methods, Supplementary Figure 2) 5,20. The details of the histomorphometry have been previously published and are included in the Supplementary Methods 5.
Statistical analysis
All statistical tests were performed in R (R version 3.5.1, 2018, Boston, MA) unless noted otherwise, with figures generated in Prism (Prism version 8.3.0, La Jolla, CA). A bimodal distribution was observed in both the OARSI and ACS grades when assessed for normality visually using a QQ plot and statistically with the Shapiro-Wilk test of normality, indicating that the grades were not normally distributed, except for the sum OARSI grades and the MTP+LTP mid-coronal OARSI grades (used for linear regression). Simple linear regression and determination of the Pearson (OARSI) and Spearman (ACS) correlation coefficients to determine the effect of the grading method (single mid-coronal section vs. sum of multiple sections) on the determination of OA severity were used to compare the relationship of a summed score to the mid-coronal score for all mice. Additional comparisons within each grading scheme were made by comparing the grading methods (average vs. mid-coronal, highest vs. mid-coronal) against another using a Wilcoxon matched pairs signed rank test (non-parametric test), which was also used to compare between the OARSI (toluidine blue) and modified OARSI (H&E) grades. The histomorphometry data were continuously and normally distributed and were compared using Students T-tests between the two OA models (DMM and naturally occurring age-related OA) and between the sham-operated and aged mice.
Details of the PCA and interobserver variation analysis are included within the Supplementary Methods.
Results
Comparison of Grading Results Using Single Versus Multiple Histologic Sections
A key point of interest in this study was determination of the effect of evaluating multiple (six) equally spaced coronal sections compared to a single mid-coronal section. In the mice from all experimental groups (n=40), simple linear regression indicated a strong linear correlation between the sum grades from multiple sections and the individual mid-coronal grades, regardless of grading scheme used (OARSI R2 = 0.8413, p<0.0001; ACS R2 = 0.8455, p<0.0001) with a strong positive correlation based on Pearson correlation coefficients (OARSI r = 0.917, 95% CI: 0.848, 0.956; ACS r = 0.919, 95% CI: 0.852, 0.957; Figure 2). Given the lack of normality of the sum ACS grades and MTP+LTP ACS mid-coronal grades, the Spearman correlation was also calculated and was 0.917. This indicates that the mid-coronal grade from a single section per joint consistently generated a similar result for cartilage lesion severity to that determined by a sum grade from multiple sections.
Figure 2. Comparison of OARSI and ACS grades from multiple sections (summed) to those from a single section.
Simple linear regression showing the strong positive correlation between the summed grade (sum of MTP + LTP grades from six total coronal sections) to the mid-coronal grades from a single section (MTP+LTP) for n=40 mice (n=10 sham mice, n=15 young DMM mice, and n=15 aged mice). (A) The OARSI sum grades were strongly positively correlated to the OARSI mid-coronal grades; R2=0.8413, linear regression, slope p <0.0001. (B) The ACS sum grades were strongly positively correlated to the ACS mid-coronal grades; R2=0.8455; linear regression, slope p <0.0001. Dotted lines = 95%CI.
The location of the most severe grade was mid-coronal or within one section of mid-coronal (within 120 um) in 77.5% of OARSI-graded MTP sections and 80% of ACS-graded MTP sections (Supplementary Table 2). For both grading schemes, the average grade of six sections (average MTP + average LTP grade) produced a lower mean grade compared to the MTP+LTP mid-coronal grade (Table 1, Figure 3). As expected, the highest MTP grade from six sections produced a significantly higher mean grade compared to a single mid-coronal grade (OARSI p<0.0001, ACS p<0.0001), given that the highest grade will always be equal to or greater than the mid-coronal grade. The grades from young sham mice were compared with those of aged animals. Both grading schemes detected a significant difference in mean OA severity, regardless of the number of sections evaluated or the method of calculating the overall grade (Table 1). Overall, the OARSI and ACS grading schemes performed similarly across the variety of grading methods that were applied (highest grade, average grade, mid-coronal grade), with small but statistically significant differences between the means generated from the various methods. A bimodal distribution of grades was particularly apparent in ACS grading; OARSI grades tended to group in the lower range of the 0–6 scale (Figure 3).
Table 1.
Summary statistics for the ACS and OARSI grades by group.
| p-value for difference between aged and DMM mice (Wilcoxon test) | p-value for difference between aged and sham mice (Wilcoxon test) | |||||
|---|---|---|---|---|---|---|
| Sum Grade (MTP+LTP) | ACS±SD | OARSI±SD | ACS | OARSI | ACS | OARSI |
| Aged mice | 3 9.1±27.3 | 21.2±9.6 | 0.6629 | 0.3294 | <0.0001* | <0.0001* |
| DMM mice | 36.7±27.2 | 17.5±9.1 | ||||
| Sham mice | 5.1±3.1 | 6.2±4.0 | ||||
| Sum Grade (MTP) | ||||||
| Aged mice | 16.1±20.0 | 9.7±8.4 | 0.0707 | 0.0439* | 0.0003* | 0.0001* |
| DMM mice | 30.5±20.5 | 12.7±6.5 | ||||
| Sham mice | 2.1±2.0 | 2.5±1.7 | ||||
| Sum Grade (LTP) | ||||||
| Aged mice | 22.9±16.5 | 11.5±4.4 | <0.0001* | <0.0001* | <0.0001* | <0.0001* |
| DMM mice | 6.3±10.9 | 4.7±4.6 | ||||
| Sham mice | 3.0±2.0 | 3.7±2.5 | ||||
| Average Grade (MTP+LTP) | ||||||
| Aged mice | 6.5±4.5 | 3.5±1.6 | 0.6782 | 0.2671 | <0.0001* | <0.0001* |
| DMM mice | 6.1 ±4.5 | 2.8±1.5 | ||||
| Sham mice | 0.9±0.5 | 1.0±0.7 | ||||
| Average Grade (MTP) | ||||||
| Aged mice | 2.7±3.3 | 1.6±1.4 | 0.0646 | 0.05358 | 0.0004* | 0.0001* |
| DMM mice | 5.1 ±3. 4 | 2.1±1.1 | ||||
| Sham mice | 0.4±0.3 | 0.4±0.3 | ||||
| Average Grade (LTP) | ||||||
| Aged mice | 3.8±2.8 | 1.9±0.7 | <0.0001* | <0.0001* | <0.0001* | <0.0001* |
| mice | 1. 0± 1.8 | 0.8±0.8 | ||||
| Sham mice | 0.5±0.3 | 0.6±0.4 | ||||
| Highest Grade (MTP+LTP) | ||||||
| Aged mice | 11.9±6.1 | 5.5±2.1 | 0.1695 | 0.2123 | 0.0001* | 0.0001* |
| DMM mice | 9.6±5.7 | 4.6±1.9 | ||||
| Sham mice | 2.2±1.0 | 2.2±1.3 | ||||
| Highest Grade (MTP) | ||||||
| Aged mice | 4.9±4.1 | 2.6±1.7 | 0.203 | 0.178 | 0.0002* | 0.0019* |
| DMM mice | 7.7±4.2 | 3.1±1.2 | ||||
| Sham mice | 1.0±0.7 | 1.0±0.6 | ||||
| Highest Grade (LTP) | ||||||
| Aged mice | 6.9±3.9 | 2.9±1.0 | <0.0001* | 0.0002* | 0.0005* | 0.0002* |
| DMM mice | 1.9±2.7 | 1.5± 1.1 | ||||
| Sham mice | 1.2±0.6 | 1.2±0.8 | ||||
| Mid-Coronal Grade (MTP+LTP) | ||||||
| Aged mice | 7.8±6.5 | 4.2±1.5 | 0.7861 | 0.5375 | <0.0001* | 0.0013* |
| DMM mice | 8.2±6.1 | 3.8±1.7 | ||||
| Sham mice | 1.1±0.7 | 1.5±1.5 | ||||
| Mid-Coronal Grade (MTP) | ||||||
| Aged mice | 3.7±3.8 | 1.9±1.4 | 0.3598 | 0.0704 | 0.0001* | 0.0046* |
| DMM mice | 6.6±4.8 | 2.7±1.3 | ||||
| Sham mice | 0.3±0.5 | 0.6±0.8 | ||||
| Mid-Coronal Grade (LTP) | ||||||
| Aged mice | 4.1±3.9 | 2.3±0.9 | 0.0024* | 0.0006* | 0.0002* | 0.0003* |
| DMM mice | 1.6±2.5 | 1.1±0.9 | ||||
| Sham mice | 0.8±0.4 | 0.9±0.7 | ||||
Sham n=10, DMM n=15, aged n=15.
indicates p<0.05. Average and sum grades were generated from 6 scored sections (MTP or LTP) or from 12 total scores (MTP+LTP).
Figure 3. Comparison of different ways to present the results for OARSI and ACS scores.
(A) Comparison of the OARSI grade by method (average, mid-coronal, highest grade) for the MTP+LTP (n=40). The OARSI grades tend to cluster lower on the scale, with fewer high-scoring individuals relative to the ACS grades. (B) Comparison of the OARSI grade by method (average, mid-coronal, highest grade) for the MTP only (n=40). (C) Comparison of the ACS grade by method (average, mid-coronal, highest grade) for the MTP+LTP (n=40). The frequently bimodal distribution of ACS grades is particularly evident when assessing the mid-coronal and highest grade. (D) Comparison of the ACS grade by method (average, mid-coronal, highest grade) for the MTP alone (n=40). The bimodal distribution of ACS grades is particularly evident when assessing the mid-coronal and highest grade within this group of 40 mice. Wilcoxon signed rank test, paired. Boxplot with whiskers = minimum, maximum values.
Comparison of Hematoxylin & Eosin Staining and Toluidine Blue Staining
Both safranin O and toluidine blue stains allow quantification of proteoglycan content within articular cartilage. Although widely used, safranin O staining is prone to variability and is affected by fixation method, length of decalcification, and staining time 21–23. Toluidine blue staining is often suggested as an alternative stain, although it can also vary in intensity based on the application methods 24. A comparison of the standard OARSI scheme applied to toluidine blue-stained sections revealed similar results to the modified OARSI scheme using the presence of CCD on H&E to assign a 0.5 score (Figure 4). The modified OARSI grading scheme resulted in the same score in all mice except one, which was graded 0 with the modified OARSI scheme on H&E and 0.5 with the standard OARSI scheme on toluidine blue.
Figure 4. Comparison of modified OARSI grading to standard OARSI grading.
The modified OARSI scheme uses presence of articular chondrocyte cell death (CCD) on H&E as a substitute for loss of toluidine blue staining (proteoglycan loss; n=40) to assign a grade of 0.5. Boxplot with whiskers = minimum, maximum values.
Comparing the DMM Model and the Aging Model
The ACS and OARSI schemes identified a similar severity of MTP disease for the aged mice (18 months old) and DMM mice (evaluated 8 weeks after surgery), with the exception of a significant difference in the MTP OARSI sum score, with the DMM mice having a mean sum OARSI grade of 12.73 (95%CI = 9.165, 16.30) and the aged group having a mean sum OARSI grade of 9.67 (95%CI = 5.017, 14.32) (Mann Whitney U test, p=0.044; Table 1). Given that the DMM procedure is performed on the medial side of the joint, this model is expected to induce more severe cartilage lesions at the MTP. Both schemes detected an increased severity of LTP cartilage lesions in the aged mice as compared to the LTP lesions in DMM mice using the sum grade (OARSI and ACS p<0.0001; data not shown), the average grade (OARSI and ACS p<0.0001), the mid-coronal grade (OARSI p=0.0006; ACS p=0.0024), and the highest grade (OARSI p=0.0002, ACS p=0.0006).
Based on histomorphometry, the aged mice had significantly increased area and average thickness of subchondral bone compared with DMM mice (area p=0.0016; thickness p=0.0023; Table 2) but had a similar area and thickness of articular and calcified cartilage. The percentage of CCD was much higher in aged mice compared to the DMM mice (aged mean = 56.2% vs. DMM mean = 12.58%; p<0.0001). There was no significant difference in the MTP osteophyte or synovial hyperplasia grades (Figure 5; Wilcoxon signed rank test; osteophytes p=0.1838, synovial hyperplasia p=0.0792).
Table 2.
Histomorphometry values by group.
| PARAMETER | SHAM MEAN (SD) | DMM MEAN (SD) | AGED MEAN (SD) | SHAM VS DMM P VAL | SHAM VS AGED P VAL | DMM VS AGED P VAL |
|---|---|---|---|---|---|---|
| Art Cart Area (μm2) | 38,880 (8,011) | 29,466 (10,855) | 30,030 (8,585) | 0.0215* | 0.0168* | 0.876 |
| Art Cart Thickness (μm) | 48.43 (8.7) | 36.93 (10.5) | 40.45 (7.2) | 0.0074* | 0.0281* | 0.304 |
| Subchondral Bone Area (μm2) | 65,600 (20,892) | 85,067 (21,409) | 149,283 (63,334) | 0.0353* | 0.0001* | 0.0016* |
| Subchondral Bone Thickness (μm) | 64.76 (21.7) | 84.85 (23.3) | 138.4 (54.5) | 0.0391* | 0.0001* | 0.0023* |
| Calcified cartilage area (μm2) | 51,300 (8,642) | 46,200 (7,757) | 42,283 (7,469) | 0.1497 | 0.0151* | 0.169 |
| Calcified cartilage thickness (μm) | 49.87 (9.5) | 46.67 (9.0) | 42.11 (7.0) | 0.4111 | 0.0427* | 0.132 |
| % Chondrocyte cell death (area cell death/art cart area*100) | 1.66 (1.5) | 12.58 (6.3) | 56.2 (28.4) | <0.0001 * | <0.0001 * | <0.0001* |
Articular cartilage (Art Cart), subchondral bone, and calcified cartilage measurements were collected from a mid-coronal section from the medial tibial plateau with analysis performed by Osteomeasure Histomorphometry System (OsteoMetrics®). Sham: n=10, DMM: n=15, aged: n=15.
indicates p-values <0.05 based on a Student’s T-test (unpaired).
Figure 5. Osteophyte and synovial hyperplasia grades by group.
(A) Osteophyte grades included the full range from 0–3 for all groups, with no significant difference between DMM and aged mice. (B) Synovial hyperplasia grades were limited to either 0 or 1 in all groups. Box plots with whiskers indicating minimum and maximum values.
Principal Components Analysis (PCA) of Histologic Features of Murine Osteoarthritis
PCA resulted in retention of four factors (principal components) that together accounted for 78.4% of the total variance of the data (Factor 1, 36.4%; Factor 2, 20.8%: Factor 3, 11.3%; Factor 4, 9.9%) (Table 3). Factor 1 was driven predominantly by measures of the articular cartilage damage, which included the OARSI grade, ACS grade, toluidine blue grade, articular cartilage area, and articular cartilage thickness, with the osteophyte grade also contributing significantly to this factor. The second factor was largely driven by the loss of calcified cartilage area and thickness, despite these variables not playing a significant role in distinguishing between the groups (sham, DMM, aged) based on histomorphometry results alone (Table 2). The subchondral bone thickness also contributed significantly to variation in Factor 2, with synovial hyperplasia being inversely related to subchondral bone thickness. Variation explained by Factor 3 was largely driven by %CCD, followed by subchondral bone thickness, ACS grade, articular cartilage thickness, and OARSI grade.
Table 3.
Principal components analysis.
| Factor 1: Responsible for 36.4% of variation | Factor 2: Responsible for 20.8% of variation | Factor 3: Responsible for 11.3% of variation | Factor 4: Responsible for 9.9% of variation |
|---|---|---|---|
| ACS grade (0.41) | CC thickness (−0.50) | %CCD (0.55) | SCB area (0.62) |
| Tblue grade (0.40) | CC area (−0.50) | SCB thickness (0.38) | CC thickness (0.41) |
| OARSI grade (0.40) | SCB thickness (0.36) | ACS grade (−0.33) | AC area (0.34) |
| Osteophytes (0.36) | Synovial hyperplasia (−0.33) | AC thickness (−0.31) | SCB thickness (0.33) |
| AC area (−0.35) | %CCD (0.28) | OARSI grade (−0.31) | AC thickness (0.28) |
| AC thickness (−0.34) | AC thickness (0.26) | AC area (−0.29) | CC area (0.25) |
| %CCD (0.21) | AC area (0.19) | CC area (0.25) | Osteophytes (0.23) |
| SCB thickness (0.18) | SCB area (0.17) | Tblue grade (−0.23) | Synovial hyperplasia (−0.09) |
| CC area (−0.15) | Tblue grade (−0.16) | CC thickness (0.20) | %CCD (−0.08) |
| SCB area (0.14) | OARSI grade (−0.09) | Osteophytes (0.07) | ACS grade (0.08) |
| CC thickness (−0.09) | ACS grade (−0.08) | Synovial hyperplasia (0.06) | Tblue grade (0.08) |
| Synovial hyperplasia (−0.09) | Osteophytes (−0.05) | SCB area (0.00) | OARSI grade (0.02) |
The first four factors accounted for 78.4% of the variation within the data set (n=40). Features with a factor loading of > 0.3 or <−0.3 are highlighted in bold. Articular cartilage (AC), calcified cartilage (CC), and subchondral bone (SCB) area and average thickness. %CCD indicates chondrocyte cell death within the articular cartilage (area of articular cartilage chondrocyte cell death/total area of articular cartilage *100).
Discussion
The ACS and OARSI grading schemes showed a strong correlation between grades generated from multiple sections (sum grade) and the mid-coronal grade from a single section. This suggests that a single mid-coronal section is sufficient for producing a representative cartilage lesion severity grade across a wide range of OA severities, regardless of whether the OARSI or ACS grading scheme is used or whether the model was an aging or surgical model. This contradicts the original recommendation that the mouse OARSI scheme be applied to multiple sections per mouse.4 Use of a single mid-coronal section reduces the time and expense of preparing and grading multiple slides, while allowing time for more detailed assessment, such as histomorphometry, to complement cartilage grading data.
While the mid-coronal section represented the most severe lesions in a majority of mice, there was variation in the location of the most severe lesions by model, with some aged mice having more severe lesions posterior to the mid-coronal point and a portion of DMM mice having more severe lesions anterior to the mid-coronal point (Supplementary Table 2). Additional caveats to focusing on the mid-coronal region are that the individual doing the sectioning should be well trained in the appropriate landmarks that allow identification of a truly mid-coronal plane of section and the model used does not have a propensity for developing lesions at a location that is not mid-coronal, as in the ACL tear model that develops posterior compartment lesions 8,25,26.
Both the OARSI and the ACS scheme produced different mean grades for the overall OA severity when analyzed using the average or highest grade from multiple sections as compared to the grade from a single mid-coronal section. This indicates that the method of reporting cartilage lesions (average, mid-coronal, highest) may impact the overall severity of cartilage lesions that in some cases might explain differences in cartilage lesion severity within or between studies that use different reporting methods. The use of an average grade from multiple sections may lead to a decreased standard deviation, affecting the power calculation and potentially reducing the necessary group size. However, the use of an average grade may also reduce the overall effect size (difference between group means), reducing or eliminating any perceived advantage in using an average grade based on a power calculation. An additional point of interest is that the variation in these mean grades on the ACS and OARSI scales was small (ACS average = 3.0, mid-coronal = 3.9, highest = 5.0; OARSI average = 1.5, mid-coronal = 1.9, highest = 2.4). The biological significance of such small differences is not clear, suggesting any of these reporting measures would lead to similar conclusions about the results. Importantly, murine OA studies should be appropriately powered to detect biologically significant differences (effect size chosen by the investigator when designing the study) between groups, with experimental design fully taking into consideration the range of the chosen histologic grading scheme 27.
The observed bimodal distribution of the ACS grades is attributed to the tendency for murine articular cartilage to form tidemark clefts, with subsequent full-thickness loss of the articular cartilage across the tibial articular surface rather than loss of articular cartilage from the surface down, leading to a higher (10–12) ACS grade in severely affected animals 5,18. The addition of evaluation of articular chondrocyte cell death (CCD) may provide a useful parameter that is more sensitive to early pathologic changes preceding articular cartilage loss.
The presence of focal chondrocyte cell death on H&E to give a score of 0.5 was an effective alternative for the standard 0.5 OARSI score measured by loss of Toluidine blue staining. The standard OARSI scores did not generate a significantly different mean OA severity compared to the modified OARSI grades using sham mice, young DMM mice, or aged mice. Toluidine blue or safranin O staining have commonly been utilized in murine OA studies due to their ability to detect loss of proteoglycans from the articular cartilage, thought to represent an early change in the progression of OA 4,23. However, staining intensity can vary significantly between individuals, between staining batches, and between laboratories, necessitating internal positive controls and careful assessment of sections to validate the success of the staining procedure 23, 28. Decalcification can affect the proteoglycan content of tissue, and loss of proteoglycan content during processing can be significant 9,29,30. H&E staining is the gold standard for evaluation of cell and tissue morphology, is widely used by laboratories performing routine histopathologic assessment, and produces highly consistent results. An added advantage of H&E staining is the ability to evaluate viability of articular cartilage chondrocytes, allowing determination of the %CCD. Interestingly, the grading scheme outlined by Pritzker et al. incorporated chondrocyte cell death, but the OARSI scheme for murine OA studies dropped this feature 2,4. Given our findings, H&E is considered a reasonable and effective method for evaluation of murine OA using either the OARSI or ACS grading scheme. Toluidine blue or safranin O stains may be used to provide supplementary information, as both the OARSI and ACS publications include supplemental schemes using these techniques. This allows more detailed evaluation of the articular cartilage when a significant portion of samples lack articular cartilage structure changes (e.g., fibrillation or cartilage loss) to distinguish severity of OA using the OARSI or ACS scheme.
DMM mice and aged mice had a similar severity of OA in the MTP by the majority of the methods used, while aged mice had significantly greater severity of OA at the LTP regardless of the grading scheme (ACS or OARSI) or method of analyzing the data (sum, average, mid-coronal, highest grade). These results indicate the importance of scoring the LTP in addition to the MTP when evaluating age-related OA. While the OARSI recommendations advocate for evaluation of all four quadrants (MTP, LTP, MFC, and LFC), the femoral condylar articular cartilage is thinner and can be more difficult to evaluate than that of the tibial plateau. In our study and past work, the femoral lesions have been less severe than those of the medial tibial plateau 5,18. However, given the fact that 40% of the aged mice in this study had more severe LFC OA than LTP OA, with differences greater than 2 grades (ACS) and greater than 1 grade (OARSI) in 2 mice (13.3%), evaluation of the LFC in addition to the LTP may be informative in aging studies. Embedding and sectioning with the focus on alignment of the tibia to obtain optimal mid-coronal sections with maximal consistency also favors evaluation of the tibial compartment 9. While these recommendations are based on assessment of established models (aging and DMM), new models may require a separate assessment to determine the location of greatest OA severity. While this study was limited to one time-point for the DMM model (eight weeks post-surgery), we would not expect the location of the lesions to vary at other time points.
Several additional histological features showed variation between the models, including the subchondral bone area and width (increased in aged mice) and the % CCD (increased in aged mice). Evaluation of these features in murine OA studies could be useful in capturing the effects of successful interventions and in aligning murine OA findings more closely with those of human OA phenotypes. For example, subchondral bone has been recognized as having an important role in some subsets of human OA and using a model such as the aging model that recapitulates these changes and reporting histomorphometry findings regarding these features are important 2,31. Other commonly used murine models, such as the ACL tear model, may have similar subtle changes in disease measures that may be missed if histomorphometry is not included as a study endpoint.
PCA allows for the combination of semiquantitative assessments (grades) with continuous measurements (histomorphometry) into a single assessment that simplifies complex data while correlating features that behave similarly in terms of outcome (OA severity) with one another. Four factors were retained by PCA, accounting for 78% of the variance in the MTP data. Both OARSI and ACS grades grouped together in Factor 1, along with the articular cartilage area and width and Toluidine blue staining. Interestingly, calcified cartilage area and thickness were responsible for much of the variation explained by Factor 2. CCD has been reported in past murine OA studies, with limited studies suggesting apoptosis as the causative mechanism, but remains to be investigated in detail 18. The % CCD was the primary driver of Factor 3, along with subchondral bone width, ACS grade, articular cartilage with, and the OARSI grade. The % CCD was significantly higher in aged mice compared to DMM and sham mice, differentiating these models.
Overall, this study establishes that the OARSI and ACS schemes perform similarly when applied to a murine aging or DMM model and suggests a single H&E stained mid-coronal section is effective in determination of OA severity in these models. The results also demonstrate the utility of H&E stained sections with evaluation of cell death as a reliable replacement for proteoglycan-based special stains such as toluidine blue in early disease. The use of histomorphometry to supplement either grading scheme may be useful in capturing morphologic changes beyond fibrillation/loss of articular cartilage, such as changes to the subchondral bone and %CCD, that may contribute substantially to variation in severity between treatment groups or models.
Supplementary Material
Acknowledgements
We would like to acknowledge Katalin Kovacs and Paula Overn from the University of Minnesota Masonic Cancer Center Comparative Pathology Shared Resources Laboratory for their meticulous preparations of the histologic sections utilized in this study, as well as the veterinary students including Katie McDermott, Lindsey Harper, and Sery Johnson who were involved in grading and collection of histomorphometry data for this study.
Role of the funding source
This project was supported by grants from the National Institute of Arthritis, Musculoskeletal, and Skin Disease (R37-AR049003 and T32-AR050938) and the National Institute on Aging (R01-AG044034). The study sponsors had no role in the study design, collection, analysis and interpretation of data; the writing of the manuscript; or the decision to submit the manuscript for publication.
Footnotes
Competing Interest Statement
The authors declare that they have no competing interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References:
- 1.Chambers MG, Bayliss MT, Mason RM. Chondrocyte cytokine and growth factor expression in murine osteoarthritis. Osteoarthritis and Cartilage 1997;5:301–308. [DOI] [PubMed] [Google Scholar]
- 2.Pritzker KPH, Gay S, Jimenez SA, Ostergaard K, Pelletier J-P, Revell PA, Salter D, and van den Berg WB. Osteoarthritis cartilage histopathology: grading and staging. Osteoarthritis and Cartilage 2006;14:13–29. [DOI] [PubMed] [Google Scholar]
- 3.Glasson SS, Blanchet TJ, Morris EA. The surgical destabilization of the medial meniscus (DMM) model of osteoarthritis in the 129/SvEv mouse. Osteoarthritis and Cartilage 2007;15:1061–1069. [DOI] [PubMed] [Google Scholar]
- 4.Glasson SS, Chambers MG, Van Den Berg WB, and Little CB. The OARSI histopathology initiative – recommendations for histological assessments of osteoarthritis in the mouse. Osteoarthritis and Cartilage 2010;18:S17–S23. [DOI] [PubMed] [Google Scholar]
- 5.McNulty MA, Loeser RF, Davey C, Callahan MF, Ferguson CM, and Carlson CS. A comprehensive histological assessment of osteoarthritis lesions in mice. Cartilage 2011;2(4):354–363. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Nagira K, Ikuta Y, Shinohara M, Sanada Y, Omoto T, Kanaya H et al. Histological scoring system for subchondral bone changes in murine models of joint aging and osteoarthritis. Scientific Reports 2020;10:10077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Malfait A, Little CB. On the predictive utility of animal models of osteoarthritis. Arthritis Res Ther 2015;17(1): 225. DOI 10.1186/s13075-015-0747-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.McCoy AM. Animal models of osteoarthritis: comparisons and key considerations. Veterinary Pathology 2015;52(5):803–818. [DOI] [PubMed] [Google Scholar]
- 9.Pastoureau PC, Hunziker EB, Pelletier J-P. Cartilage, bone and synovial histomorphometry in animal models of osteoarthritis. Osteoarthritis and Cartilage 2010;18:S106–S112. [DOI] [PubMed] [Google Scholar]
- 10.Kuyinu EL, Narayanan G, Nair LS, Laurencin CT. Animal models of osteoarthritis: classification, update, and measurement of outcomes. J Ortho Surg Res 2016;11(19):1–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Wendler A, Wehling M. The translatability of animal models for clinical development: biomarkers and disease models. Curr Opin Pharmacol 2010;10(5):601–606. [DOI] [PubMed] [Google Scholar]
- 12.Percie du Sert N. Maximising the output of osteoarthritis research: the ARRIVE guidelines. Osteoarthritis and Cartilage 2012;20(4):253–255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Mankin HJ, Dorfman H, Lippiello L, Zarins A. Biochemical and metabolic abnormalities in articular cartilage from osteo-arthritic human hips. J Bone Joint Surgery 1971;53-A(3):523–537. [PubMed] [Google Scholar]
- 14.Ostergaard K, Petersen J, Andersen CB, Bendtzen K, Salter DM. Histologic/histochemical grading system for osteoarthritic articular cartilage. Arthritis Rheum 1997;40(10):1766–1771. [DOI] [PubMed] [Google Scholar]
- 15.Ostergaard K, Andersen CB, Petersen J, Bendtzen K, Salter DM. Validity of histopathological grading of articular cartilage from osteoarthritic knee joints. Ann Rheum Dis 1999;58:208–213 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Aigner T, Cook JL, Gerwin N, Glasson SS, Laverty S, Little CB, et al. Histopathology atlas of animal model systems – overview of guiding principles. Osteoarthritis and Cartilage 2010;18:S2–S6. [DOI] [PubMed] [Google Scholar]
- 17.Loeser RF, Goldring SR, Scanzello CR, Goldring MB. Osteoarthritis: a disease of the joint as an organ. Arthritis Rheum 2012;64(6):1697–1707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.McNulty MA, Loeser RF, Davey C, Callahan MF, Ferguson CM, and Carlson CS. Histopathology of naturally occurring and surgically induced osteoarthritis in mice. Osteoarthritis and Cartilage 2012;20:949–956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Loeser RF, Olex A, McNulty MA, Carlson CS, Callahan M, Ferguson C et al. Microarray analysis reveals age-related differences in gene expression during the development of osteoarthritis in mice. Arthritis Rheum 2012;64(3):705–717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rowe MA, Harper LR, McNulty MA, Lau AG, Carlson CS, Lin Leng, et al. Reduced osteoarthritis severity in aged mice with deletion of macrophage migration inhibitory factor. Arthritis Rheumatol 2017;69:352–361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Camplejohn KL, Allard SA. Limitations of safranin ‘O’ staining in proteoglycan-depleted cartilage demonstrated with monoclonal antibodies. Histochemistry 1988;89:185–188. [DOI] [PubMed] [Google Scholar]
- 22.Király K, Lammi M, Arokoski J, Lapveteläinen T, Tammi M, Helminen H et al. Safranin O reduces loss of glycosaminoglycans from bovine articular cartilage during histological specimen preparation. Histochemical Journal 1996;28:99–107. [DOI] [PubMed] [Google Scholar]
- 23.Schmitz N, Laverty S, Kraus VB, Aigner T. Basic methods in histopathology of joint tissues. Osteoarthritis and Cartilage 2010;18:S113–S116. [DOI] [PubMed] [Google Scholar]
- 24.Bergholt NL, Lysdahl H, Lind M, Foldager CB. A standardized method of applying toluidine blue metachromatic staining for assessment of chondrogenesis. Cartilage 2019;10(3):370–374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kamekura S, Hoshi K, Shimoaka T, Chung U, Chikuda H, Yamada T et al. Osteoarthritis development in novel experimental mouse models induced by knee joint instability. Osteoarthritis and Cartilage 2005;13(7):632–641. [DOI] [PubMed] [Google Scholar]
- 26.Farooq Rai M, Duan X, Quirk JD, Holguin N, Schmidt EJ, Chinzei N et al. Post-traumatic osteoarthritis in mice following mechanical injury to the synovial joint. Sci Reports 2017;7:45223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Pearce GL, Frisbie DD. Statistical evaluation of biomedical studies. Osteoarthritis Cartilage 2010;18:S117–S122. [DOI] [PubMed] [Google Scholar]
- 28.Laverty S, Girard CA, Williams JM, Hunziker EB, Pritzker KPH. The OARSI histopathology initative – recommendations for histological assessments of osteoarthritis in the rabbit. Osteoarthritis and Cartilage 2010;18(S3):S53–265. [DOI] [PubMed] [Google Scholar]
- 29.Hunziker EB, Graber W. Differential extraction of proteoglycans from cartilage tissue matrix compartments in isotonic buffer salt solutions and commercial tissue-culture media. J Histochem Cytochem 1986;34(9):1149–53. [DOI] [PubMed] [Google Scholar]
- 30.Melrose J, Smith S, Ghosh P. Histological and immunohistological studies on cartilage. Cartilage and Osteoarthritis, Vol. 2, Totowa, NJ: Humana Press; 2004. [DOI] [PubMed] [Google Scholar]
- 31.Pelletier J-P, Boileau C, Brunet J, Boily M, Lajeunesse D, Reboul P, et al. The inhibition of subchondral bone resorption in the early phase of experimental dog osteoarthritis by licofelone is associated with a reduction in the synthesis of MMP-13 and cathepsin K. Bone 2004;34(3):527–538. [DOI] [PubMed] [Google Scholar]
- 32.Little CB, Barai A, Burkhardt D, Smith SM, Fosang AJ, Werb Z, et al. Matrix metalloproteinase-13 deficient mice are resistant to osteoarthritis cartilage erosion but not chondrocyte hypertrophy or osteophyte development. Arthritis Rheumatol 2009;60(12):3723–3733. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.