Arthritis & Rheumatology Vol. 71, No. 6, June 2019, pp 916–924 DOI 10.1002/art.40820 © 2019, American College of Rheumatology

Associations of Symptomatic Knee With Histopathologic Features in Subchondral Bone

Koji Aso,1 S. Mohsen Shahtaheri,2 Roger Hill,3 Deborah Wilson,3 Daniel F. McWilliams,2 and David A. Walsh3

Objective. Subchondral bone and the osteochondral junction are thought to contribute to osteoarthritis (OA) knee pain. We undertook this study to identify osteochondral pathologies specifically associated with symptomatic human knee OA. Methods. Medial tibial plateau samples from 2 groups of subjects (n = 31 per group) were matched for macro- scopic chondropathy scores. The symptomatic chondropathy group had undergone total knee replacement for OA knee pain, at which time specimens of the medial tibial plateau were obtained. The asymptomatic chondropathy group included subjects who died of unrelated illness (specimens were obtained at postmortem examination) and who had not previously sought help for knee pain. OA histopathology, immunoreactivity for nerve growth factor (NGF) and CD68 (macrophages), tartrate-­resistant acid phosphatase–positive subchondral osteoclasts, and synovitis were compared between groups. Results. Mankin scores, subchondral , and subchondral CD68-immunoreactive­ macrophage infil- tration were similar between the 2 groups. NGF-like­ immunoreactivity was found in subchondral mononuclear cells and osteoclasts, as well as in . NGF in osteochondral channels and osteoclast densities in subchondral bone were higher in the symptomatic chondropathy group than in the asymptomatic chondropathy group (P < 0.01 and P = 0.02, respectively), as were synovitis scores (P < 0.01). Osteochondral pathology was not significantly as- sociated with synovitis score. The differences in NGF expression and in osteoclast density remained significant after adjustment for age and synovitis score (P = 0.01 and P = 0.04, respectively). Osteochondral NGF and osteoclast den- sities, together with synovitis scores, explained ~28% of sample allocation to symptomatic or asymptomatic groups. Conclusion. Subchondral pathology was associated with symptomatic knee OA, independent of chondropathy and synovitis. Increased NGF expression in osteochondral channels and increased osteoclast density appear to be key features associated with bone pain in knee OA.

INTRODUCTION netic resonance imaging (MRI) in knee OA are strongly ­associated with pain (1–4,7). Bone attrition, a flattening or depression of the Pain is a major source of disability and the reason for hospital subchondral bone visualized using radiography or MRI, is also visits in patients with knee osteoarthritis (OA). Structural changes associated with the presence of pain (5,6). Microarray analysis of including articular degradation, synovial inflammation, BMLs in OA demonstrated up-regulation­ of genes implicated in osteophytes, and subchondral are characteristic of neurogenesis, osteochondral turnover, and inflammation that might OA, but are not always accompanied by severe pain. Evidence contribute to OA pain (8). In animals, OA caused up-regulation­ of suggests that subchondral bone contributes to knee OA pain (1– nociceptive markers (calcitonin gene-related­ peptide and tropomy- 7). Subchondral bone marrow lesions (BMLs) detected with mag- osin receptor kinase A [TrkA]) in subchondral bone afferents (9).

Supported by Arthritis Research UK (Centre initiative grant 20777) and Nottingham, UK, and Sherwood Forest Hospitals NHS Foundation Trust, a 2016 grant from the Japanese Orthopaedic Society of Knee, Arthroscopy, Sutton-in-Ashfield, UK. and Sports Medicine. Dr. McWilliams has received research support from Pfizer Ltd. Dr. Walsh has 1Koji Aso, MD, PhD: Arthritis Research UK Pain Centre, NIHR received consulting fees, speaking fees, and/or honoraria from GlaxoSmithKline Nottingham Biomedical Research Centre, University of Nottingham, (less than $10,000) and research support from Arthritis Research UK and Pfizer Nottingham, UK, and Kochi Medical School, Kochi University, Nankoku, Ltd. No other disclosures relevant to this article were reported. Japan; 2S. Mohsen Shahtaheri, PhD, Daniel F. McWilliams, PhD: Arthritis Address correspondence to Koji Aso, MD, PhD, Arthritis Research UK Research UK Pain Centre, NIHR Nottingham Biomedical Research Centre, Pain Centre, Clinical Sciences Building, City Hospital, Nottingham NG5 1PB, University of Nottingham, Nottingham, UK; 3Roger Hill, Deborah Wilson, UK. E-mail: [email protected]. RGN, David A. Walsh, FRCP, PhD: Arthritis Research UK Pain Centre, NIHR Submitted for publication June 5, 2018; accepted in revised form Nottingham Biomedical Research Centre, University of Nottingham, December 18, 2018.

916 ASSOCIATIONS OF SYMPTOMATIC KNEE OA WITH SUBCHONDRAL BONE PATHOLOGIES | 917

However, the mechanisms by which subchondral pathology con- were approved by Nottingham 1 Research Ethics Committee tributes to OA pain are incompletely understood. Synovitis has also (05/Q2403/24) and Derby Research Ethics Committee 1 (11/ been associated with OA pain (1,10–13). Synovial­ and subchon- H0405/2). Symptomatic chondropathy samples were from dral pathology can occur together within the same joint, but it is patients fulfilling the American College of Rheumatology classifi- unknown whether these represent discrete painful pathologies that cation criteria for OA (24) at the time of TKR. could be separate targets for therapeutic intervention. Nerve growth factor (NGF) plays a key role in the generation Macroscopic chondropathy score and osteophytes. of acute and chronic pain, especially in inflammation (14,15). NGF Following tissue harvesting, articular surfaces of the medial tibial can bind 2 receptors: the high-­affinity TrkA (16) and the low-­affinity plateau were evaluated by a single assessor (RH) for the extent p75 neurotrophin receptor (17). NGF blockade can be achieved and severity of loss of surface integrity (25). Articular surface using antibodies or TrkA–IgG fusion proteins that bind NGF and defects were graded using the following scale: 0 (normal, smooth prevent its interaction with TrkA and p75 receptors. Recent clinical unbroken surface), 1 (swelling and softening), 2 (superficial fibrilla- trials showed that NGF blockade substantially reduced OA knee tion), 3 (deep fibrillation), or 4 (subchondral bone exposure). The pain (18,19). In human OA, NGF is up-regulated­ in synovium (10) proportion of articular surface area corresponding to each grade and subchondral bone (20), and increased sy­novial NGF expres- was used to calculate a score using the following formula (25): sion was associated with symptomatic knee OA (10), although the macroscopic chondropathy score (0–100) = (grade 1 × 0.14) + relevance of subchondral NGF expression has not been clarified. (grade 2 × 0.34) + (grade 3 × 0.65) + grade 4. Osteophytes were Increased density of tartrate-­resistant acid phosphatase (TRAP)– recorded as present or absent on direct visualization of postmor- positive osteoclasts in subchondral bone has also been shown to tem samples. be associated with OA and knee symptoms (21,22). Inflammatory CD68-positive­ macrophages were also detected in subchondral Radiographic OA severity score. Radiographic OA bone marrow compartments in human OA (23). severity scores were derived using preoperative posteroanterior We hypothesized that structural, cellular, and molecular knee radiographs as previously described (25). An atlas of line changes in subchondral bone are associated with symptomatic drawings of the knee joint was used to grade medial and lat- knee OA. We compared 2 chondropathy groups that had simi- eral joint space narrowing and osteophytes (26). The scores for lar macroscopic chondropathy but differing symptom severities. tibiofemoral joint space narrowing (range 0–6) and osteophytes Patients in the symptomatic chondropathy group had previously (range 0–12) were summed to provide a total radiographic OA seen a clinician for knee pain and undergone total knee replace- severity score (range 0–18) (25). ment (TKR) surgery. Subjects in the asymptomatic chondropathy group had not sought care for knee pain and died of unrelated Sample processing. Midcoronal sections of the middle illness. We hypothesized that NGF expression by cells within third of the medial tibial plateau (an important weight-­bearing area subchondral bone was associated with symptomatic OA. characteristically affected by OA) were fixed in neutral buffered for- malin and then decalcified in 10% EDTA in 10 mM Tris buffer (pH 6.95, 4°C), prior to wax embedding. Synovial tissues were fixed in PATIENTS AND METHODS formalin and wax-embedded­ without decalcification. Samples. The study included 31 consecutive samples from symptomatic chondropathy patients who had donated tibial pla- Histology and grading. Tibial plateau sections (5 μm) were teau at TKR surgery for OA and 31 samples from asymptomatic stained with hematoxylin and eosin or Safranin O–fast green. OA chondropathy subjects obtained at autopsy. All symptomatic articular cartilage changes were graded using the Mankin scale chondropathy patients undergoing TKR had reported severe (27): cartilage surface integrity (0–6, where 0 = normal and 6 = com- knee pain. No asymptomatic chondropathy subjects had sought plete disorganization), tidemark integrity (0–1, where 0 = intact and medical attention for knee pain during the last year, and thus this 1 = crossed by vessels), morphology (0–3, where 0 = group was highly likely to have experienced less pain than the normal and 3 = hypocellular), and proteoglycan loss (0–4, where symptomatic­ chondropathy patients. Asymptomatic chondrop- 0 = normal, no loss of Safranin-­O stain and 4 = complete loss athy samples were selected from 782 consecutive postmortem of stain). Subchondral bone marrow replacement by fibrovascular donors by matching them to each symptomatic chondropathy tissue was assessed as either present or absent. Subchondral sample using both the macroscopic chondropathy score and osteosclerosis was histologically assessed using trabecular bone the percentage of joint surface with grade 4 chondropathy lesion volume/total volume (BV/TV) and subchondral plate area (μm2/ (subchondral bone exposure), with a maximum difference of ±5 μm), which were quantified using computer-assisted­ image anal- for each measure between the matched cases. ysis (Zeiss). Osteochondral channel densities were assessed for Informed consent was obtained from TKR patients and subchondral bone, calcified cartilage, andnoncalcified ­ cartilage, the next of kin consented for postmortem subjects. Protocols separately in each region. Channels passing through one region 918 | ASO ET AL

into another were counted as being in the region occupied by the NGF expression was measured as the proportion of larger part of the channel. Synovial inflammation was assessed osteochondral channels in each sample that displayed NGF-­ using a synovitis histologic score (0–3, where 0 = no synovitis and immunoreactive cells. Subchondral tissues within 400 μm of the 3 = severe synovitis) developed by Haywood et al (28). cement line in the osteochondral junction were classified as bone marrow or fibrovascular tissues. NGF-­like immunoreactivity was Immunohistochemistry. Sections underwent antigen graded in each subchondral tissue type as 0 (none), 1 (focal/sparse retrieval (10 mM citrate buffer, 90°C, 20 minutes) and were blocked distribution), or 2 (high density); it was graded in chondrocytes as 0 with 5% bovine serum albumin containing goat serum, followed (<5% of cells), 1 (5–20% of cells), or 2 (>20% of cells) (20). CD68-­ by incubation with rabbit monoclonal antibody to NGF (EP1320Y; immunoreactive macrophages were graded in subchondral tissues Abcam), and biotinylated goat anti-­rabbit IgG secondary anti- as 0 (none), 1 (focal/sparse distribution), or 2 (high density) (20). body (BA1000; Vector). CD68 immunoreactivity was visualized after citrate buffer antigen retrieval (1 mg/ml pepsin in 0.5M acetic TRAP staining. Differentiated osteoclasts were identified acid, 37°C, 2 hours) and incubation with mouse monoclonal anti-­ by TRAP staining, using a commercially available kit (#386A; human CD68 (MA5-13324;­ ThermoFisher) and biotinylated horse Sigma-­Aldrich) according to the manufacturer’s protocol. TRAP-­ anti-mouse­ IgG secondary antibody (BA2001; Vector). NGF and positive osteoclasts were counted within 400 μm of the cement CD68 immunoreactivities were visualized using avidin–biotin–per- line in the osteochondral junction and divided by the length of oxidase complex (Vector) with nickel-enhanced­ diaminobenzidine the subchondral bone to give an osteoclast density expressed development (29). Sections were counterstained with hematoxylin as TRAP-positive/mm­ (22). One dark-purplish­ or reddish cell so that different regions were more apparent. with ≥3 nuclei was recorded as 1 osteoclast.

Table 1. Patient and sample characteristics* Symptomatic Asymptomatic Postmortem chondropathy chondropathy repository (n = 31 knees) (n = 31 knees) (n = 782 knees) Macroscopic chondropathy 74 (56–80) 76 (56–81)† 33 (24–51) score, 0–100 % of joint surface area with 30 (0–48)‡ 30 (0–50)† 0 (0–0) grade 4 chondropathy Sex, % male 51.6 61.3 54.5 Age, years 67 (55–73)‡ 74 (66–84)§ 69 (60–80) Total radiographic OA severity 13 (10.5–13.5) – – score, 0–18¶ Tibiofemoral JSN score, 0–6# 5 (5–5.8) – – MT JSN score, 0–3 3 (3) – – Osteophyte score, 0–12** 8 (5.5–8) – – MT osteophyte score, 0–3 2 (2–2) – – Samples with osteophytes, no. (%)†† MFC osteophytes – 16 (53.3)† 113 (15.3) LFC osteophytes – 18 (62.1)† 111 (15.0) MT osteophytes – 15 (50.0)† 87 (11.7) LT osteophytes – 13 (43.3)† 82 (11.1) Patellar osteophytes – 10 (50.0)† 41 (11.4) * Except where indicated otherwise, values are the median (interquartile range). OA = osteoarthritis; MT = medial tibial plateau; MFC = medial femoral condyle; LFC = lateral femoral condyle; LT = lateral tibial plateau. † P < 0.01 versus the postmortem repository. ‡ P < 0.01 versus asymptomatic chondropathy. § P = 0.03 versus the postmortem repository. ¶ Summation of tibiofemoral joint space narrowing (JSN) and osteophyte scores. # Summation of medial and lateral tibiofemoral JSN scores. ** Summation of medial and lateral tibial and femoral osteophyte scores. †† Not all samples could be assessed for osteophyte presence; percentages are based on the number of assessments available. ASSOCIATIONS OF SYMPTOMATIC KNEE OA WITH SUBCHONDRAL BONE PATHOLOGIES | 919

Image analysis. All histologic scoring and quantification repository cases are shown in Table 1. Per study protocol, was performed using a Zeiss Axioskop 50 microscope, by a sin- the selected asymptomatic chondropathy group had similar gle observer (KA) who was blinded with regard to details of the macroscopic chondropathy scores and proportion of joint diagnostic group. surface area displaying grade 4 chondropathy to the symp- tomatic group. However, the samples from the asymptomatic Statistical analysis. Analyses were performed with JMP 10 chondropathy group showed more severe OA changes than software (SAS Institute). The significance of differences between were observed overall in the samples from the postmor- groups was assessed by Mann-Whitney­ U test or chi-­square test. tem repository from which they were selected. The asymp- Logistic regression was performed to adjust for age and synovitis tomatic chondropathy group had a higher median age scores and to calculate McFadden’s pseudo R2. The R2 for each than the symptomatic chondropathy group. There were no linear regression model was recorded for each of the individual cases in which medications for were used in histologic measures (NGF alone, osteoclasts alone, or synovitis either group. score alone) and also for the linear regression model where all measures were included together (NGF, osteoclasts, and synovi- Histologic characteristics. Histologic characteris- tis). Associations were assessed by Spearman’s rank correlation. tics of the study groups are shown in Figure 1 and Table 2. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were Osteochondral channels containing inflammatory cells and calculated. P values less than 0.05 were considered significant. blood vessels were observed in subchondral bone plate, cal- cified cartilage, and noncalcified cartilage (Figures 1A and B). Mankin score, proportion of cases with fibrovascular marrow RESULTS replacement, histologic BV/TV, subchondral plate area, and Patient details. Subject demographics and sam- osteochondral channel densities were similar between the ple details of samples selected for this study and of source symptomatic and asymptomatic chondropathy groups. How-

Figure 1. Histopathologic features in subchondral bone. A–E, Nerve growth factor (NGF) immunoreactivity. Sample from a subject in the symptomatic chondropathy group exhibits an NGF-positive­ osteochondral channel (A) (arrowheads), while a sample from a subject in the asymptomatic chondropathy group exhibits an NGF-negative­ osteochondral channel (B) (arrowhead). NGF-immunoreactive­ cells (brown) were found in osteochondral channels (A), fibrovascular tissue C( ), and bone marrow (D). Multinucleated osteoclasts were immunoreactive for NGF (E). F and G, Analysis of macrophage infiltration. CD68-immunoreactive­ macrophages were mainly observed in bone marrow (F) and fibrovascular tissue G( ). H, Tartrate-resistant­ acid phosphatase staining showed multinucleated osteoclasts (purple). Bars = 50 μm. 920 | ASO ET AL

Table 2. Osteochondral histology and synovitis scores* Symptomatic Asymptomatic ­chondropathy ­chondropathy (n = 31 knees) (n = 31 knees) Total Mankin score, 0–14† 9 (7–11) 8 (7–11) Cartilage surface integrity, 0–6 4 (3–6) 4 (3–6) Chondrocyte appearance, 0–3 2 (2–3) 2 (2–2) Tidemark integrity, 0–1 1 (0–1) 0 (0–1) Proteoglycan loss, 0–4 2 (2–3) 2 (2–3) Subchondral bone marrow 11 (35) 14 (45) replacement, no. (%) Histologic BV/TV‡ 50.0 (42.0–61.3) 57.3 (39.0– 63.0) Subchondral plate area, μm2/μm 608.3 (460.0–810.6) 651.5 (431.7–1050.0) Total osteochondral channel 5.4 (3.7–6.4) 4.9 (3.5–7.4) density, mm§ Subchondral bone, mm 4.8 (3.3–6.1) 4.7 (3.4 –7.2) Calcified cartilage, mm 0.24 (0.09–0.57) 0.25 (0–0.46) Noncalcified cartilage, mm 0 (0–0) 0 (0–0) Synovitis histologic score, 0–3 3 (2.75–3)¶ 1 (1–2.5) * Except where indicated otherwise, values are the median (interquartile range). † Summation of cartilage surface integrity, chondrocyte appearance, tidemark integrity, and proteoglycan loss. ‡ Trabecular bone volume per total volume (BV/TV). § Summation of osteochondral channel densities in subchondral bone, calcified cartilage, and noncalcified cartilage. ¶ P < 0.01 versus asymptomatic chondropathy.

ever, synovitis scores were higher in the symptomatic group TRAP-­positive multinucleated osteoclasts were observed than in the asymptomatic group, and this difference remained on the bone surface of subchondral bone (Figure 1). The significant after adjustment for age (adjusted OR [ORadj] 2.75 density of osteoclasts in the subchondral bone observed [95% CI 1.35–6.20]; P = 0.01). in the symptomatic chondropathy group was significantly In samples of medial tibial plateau, NGF immunoreactivity higher than that in the asymptomatic chondropathy group was detected in chondrocytes, subchondral mononuclear cells, (P = 0.02) (Figure 2). This difference remained significant and multinucleate osteoclast-like­ cells adherent to bone (Figure 1).­ after adjustment for age and synovitis score (ORadj 1.19 [95% NGF-immunoreactive­ cells were found in osteochondral­ channels CI 1.01–1.48]; P = 0.04). The percentage of NGF-­positive and in subchondral fibrovascular tissue and bone marrow. CD68-­ osteochondral channels was significantly correlated with the immunoreactive macrophages were observed mainly in subchon- number of TRAP-­positive osteoclasts (rs = 0.34, P = 0.01). dral bone marrow and fibrovascular tissues. A higher proportion of The association between NGF expression in osteochondral osteochondral channels contained NGF-immunoreactive­ cells in channels and symptomatic ­chondropathy remained sig- the symptomatic group than in the asymptomatic chondropathy nificant after adjustment for osteoclast density (ORadj 1.05 group (Figure 2). This difference remained significant after adjust- [95% CI 1.01–1.09]; P < 0.01), but the significant associa- ment for age and synovitis histologic score (ORadj 1.05 [95% CI tion between osteoclast density and symptomatic chondrop- 1.01–1.10]; P = 0.01). Scores for subchondral macrophage infil- athy did not persist after adjustment for NGF expression in tration and NGF immunoreactivity in chondrocytes, subchondral osteochondral channels (ORadj 1.10 [95% CI 0.96–1.32]; fibrovascular tissue, and bone marrow did not differ significantly P = 0.20). Synovitis scores were not significantly associated between groups (see Supplementary Table 1, on the Arthritis with either NGF-­immunoreactive osteochondral channels

& Rheumatology web site at http://onlinelibrary.wiley.com/ (rs = 0.07, P = 0.62) or subchondral TRAP-positive­ ­osteoclasts doi/10.1002/art.40820/abstract). NGF-­immunoreactive osteo- (rs = 0.11, P = 0.44). chondral channels were significantly associated with the Mankin McFadden’s pseudo R2 values for the symptomatic score and its component scores for tidemark integrity and carti- ­chondropathy group versus the asymptomatic chondropathy lage surface integrity (Supplementary Table 2, http://onlinelibrary. group were as follows: synovitis score 0.17, NGF expression wiley.com/doi/10.1002/art.40820/abstract). in osteochondral channels 0.13, and subchondral osteoclast ASSOCIATIONS OF SYMPTOMATIC KNEE OA WITH SUBCHONDRAL BONE PATHOLOGIES | 921

Figure 2. Immunoreactivity for nerve growth factor (NGF) and tartrate-resistant­ acid phosphatase (TRAP)–positive osteoclasts in the subchondral bone from symptomatic and asymptomatic chondropathy samples. Scatterplots illustrate the differences between symptomatic and asymptomatic chondropathy. Each symbol represents an individual sample; bars show the median and interquartile range. * = P < 0.01; # = P = 0.02. density 0.05. The pseudo R2 value for the combination of all 3 synovium, where it was associated with OA pain (10). OA chon- histopathologic features was 0.28. drocytes may also express NGF (10), although we were unable to demonstrate association of chondrocyte-­derived NGF with DISCUSSION symptomatic chondropathy. Increased NGF-immunoreactive­ cells in osteochondral channels could contribute to OA pain by increas- In this study, we demonstrated that components of ing colocalized sensory nerve activity. NGF-immunoreactive­ cells ­subchondral ­pathology are associated with symptomatic chon- were colocalized with sensory nerve fibers within osteochondral dropathy in people undergoing knee for painful OA. channels in human subchondral bone (20). Indeed, most sensory We showed that NGF expression in osteochondral channels and neurons innervating the subchondral bone in rat knee joints were subchondral TRAP-positive­ osteoclast density are individually found to be TrkA-immunoreactive­ (31), and TrkA expression in associated with symptomatic chondropathy. We confirmed pre- subchondral bone afferents was further increased during mono-­ vious findings (10) that suggested symptomatic OA is associated iodoacetate–induced OA in rats (9). with synovitis and showed that associations with subchondral Our findings showed that osteoclast density in subchondral pathology are not dependent on the severity of chondropathy or bone was associated with symptomatic knee OA, and the differ- synovitis. OA can affect all tissues in the joint, and our data support ences between the 2 groups remained significant after adjust- the notion that different joint tissue compartments make discrete ment for age and synovitis histologic score. Osteoclasts might contributions to pain in OA. increase pain either by directly changing the subchondral bio- We found that the proportion of osteochondral channels pos- chemical milieu or by altering subchondral bone structure. Osteo- itive for NGF immunoreactivity was a sensitive measure to dis- clasts release protons that generate a local acidosis, potent acti- tinguish between symptomatic and asymptomatic groups, sup- vators of nociceptors that can increase pain signaling (32). Our porting the hypothesis that osteochondral NGF plays a role in the findings also indicate that osteoclasts are a source of NGF, which generation of OA pain. This association appears to be beyond could then sensitize primary afferents in the subchondral bone. any effect of synovitis or cartilage damage on joint pain. The num- Classification of cases as symptomatic or asymptomatic ber of osteochondral channels penetrating noncalcified cartilage was significantly predicted by NGF immunoreactivity but not by is increased in OA (20), but our findings suggest that this alone subchondral trabecular bone density. Our current results therefore may not be sufficient to explain OA pain. We showed that NGF extend findings from a previous study (22), which demonstrated immunoreactivity in osteochondral channels was correlated with the potential role of increased osteoclast density in subchondral tidemark integrity, suggesting that expression of sensitizing fac- bone in the generation of OA pain. High serum concentrations tors such as NGF mediates the effects of osteochondral channels of TRAP-5b,­ an indicator of osteoclast number, were shown to on OA pain. be associated with subchondral osteoclast density, OA pain, and NGF may directly activate sensory neurons that express worse pain prognosis (22). We now show that the association of TrkA and modulate the expression of TrkA or p75 receptor (30). osteoclast density with symptomatic OA is not explained by asso- Anti-NGF­ antibodies can reduce OA pain (18,19), indicating the ciations with chondropathy, synovitis, or age, suggesting a direct importance of NGF in pain generation, although their anatomic effect of osteoclasts on OA pain. Increased subchondral osteo- site of action remains uncertain. NGF has been localized to human clast number was also associated with pain behavior in rats, and 922 | ASO ET AL

reducing the number of osteoclasts led to decreases in weight-­ contributed to group allocation at similar rates (17% and 13%, bearing pain (33,34). respectively), and both may be important targets for future OA Studies of osteoclast inhibitors such as bisphosphonates, treatments. denosumab, and strontium ranelate show reductions in joint pain in This study has several potential limitations. Some subjects in people with knee OA (35,36). Zoledronic acid, a bisphosphonate, our asymptomatic chondropathy group could have experienced reduced knee pain and BML size in OA patients (36), although knee pain, but relatives may have been unaware of these symp- findings from a meta-­analysis of randomized controlled trials did toms. However, all patients undergoing TKR reported severe knee not support analgesic effects of bisphosphonates in knee OA (37). pain, and it is highly likely that people who did not undergo sur- Our data suggest that OA knee pain has multiple sources, and gery had less pain overall than those who did. The symptomatic targeting osteoclasts will have clinically important benefits only in and asymptomatic chondropathy groups differed in age, although cases in which osteoclast activity is the predominant driver of pain. significant associations with subchondral pathology and synovi- We observed associations between NGF and osteoclast den- tis persisted after adjusting our analyses for age. Samples were sities in subchondral bone. Multinucleated osteoclasts were immu- obtained from the midcoronal section of the medial tibial plateau, noreactive for NGF, and NGF expression in osteochondral chan- a key weight-bearing­ area, but findings might differ for other joint nels was significantly correlated with the number of TRAP-positive­ regions such as femoral condyles. Symptomatic chondropathy osteoclasts detected. NGF expression in osteochondral channels patients had late-stage­ OA and were undergoing arthroplasty, and was associated with symptomatic knee OA after adjustment for different pain mechanisms might be important in cases with less osteoclast density, but the association between osteoclast density severe structural change. and symptoms did not persist after adjustment for NGF. Our data Osteoclast activity itself was not examined in this study, but support the notion that NGF is a more important factor than oste- cells with ≥3 nuclei were counted as 1 osteoclast to estimate oclast density in subchondral bone with regard to the generation active osteoclasts, as resorption activity has been shown under of OA pain. Furthermore, NGF can act as an autocrine or parac- some circumstances to correlate with the number of nuclei pres- rine factor regulating osteoclast activity and bone remodeling. NGF ent (42). However, osteoclast numbers do not necessarily correlate and TrkA are expressed by osteoclasts, and the addition of NGF to with osteoclast activity (i.e., bisphosphonate treatment) (43). More monocyte cultures induces the formation of TRAP-­positive multi- direct measures, such as biomarkers of collagen breakdown, nucleated cells (38). An anti-­NGF antibody reduced subchondral might further clarify whether associations between symptoms and osteoclast numbers in a rat model of OA pain (39). osteoclast number reflect mediation by osteoclast activity. Ours is the first study to evaluate associations between Our models did not explain all of the variance in classifica- symptomatic OA and pathologic changes in discrete tissue tion to symptomatic and asymptomatic groups. Some variation compartments of the human knee. Patients with more severe might be attributable to case ascertainment (e.g., subjects in ­chondropathy have been shown to be more likely to display syno­ the asymptomatic group might have experienced some knee vitis and subchondral bone changes (40). However, in the cur- pain). Factors not explored here, such as other histopathologic rent study, subchondral changes were not significantly associated changes, cytokines/molecules, psychological factors, biome- with synovitis grade, and each compartment might contribute dis- chanical loading, and obesity, are also likely to contribute to OA cretely to OA pain. Our findings support a heterogeneous model of pain. BMLs are associated with knee OA pain, and they have OA pain, resulting from multiple mechanisms in different peripheral also been found to be associated with cartilage surface integ- tissues. The balance between pain mechanisms varies by per- rity and subchondral bone marrow replacement by fibrovas- son. Latent class analysis indicated that synovitis was a defining cular tissue (8), both of which were similar in the symptomatic characteristic for one subgroup of patients with OA (10). Our find- and asymptomatic chondropathy groups in our study. How- ings here suggest that subchondral pathology can define a sub- ever, MRI scans were not available for subjects in our study, group of patients with symptomatic chondropathy, only partially and further investigation is needed to clarify the association of ­overlapping with those whose OA pain is driven by synovitis. MRI BMLs with NGF expression in osteochondral channels and with evidence of cartilage defects (41), bone marrow lesions (7), and TRAP-­positive osteoclast densities. Case-­matching asympto- synovitis (12) can also discretely predict OA pain. We extended matic chondropathy samples (from a total postmortem group of these findings to identify NGF-immunoreactive­ osteochondral 782 samples) to symptomatic ­chondropathy samples allowed channels and subchondral osteoclast densities as key pathologic us to identify histopathologic factors that contribute to OA features that make discrete contributions to OA symptoms. symptoms, but further research is needed to determine their Our results showed that 28% of allocation to symptomatic importance relative to contributions from chondropathy itself. and asymptomatic chondropathy groups can be explained by the In conclusion, we have identified histopathologic features combination of synovitis score, NGF expression in ­osteochondral of subchondral bone that are associated with symptomatic channels, and subchondral osteoclast density. Synovitis score chondropathy. NGF expression in osteochondral channels and NGF expression in osteochondral channels, separately, was associated with symptomatic knee OA independently ASSOCIATIONS OF SYMPTOMATIC KNEE OA WITH SUBCHONDRAL BONE PATHOLOGIES | 923

of any effects of chondropathy, synovitis, or subchondral 8. Kuttapitiya A, Assi L, Laing K, Hing C, Mitchell P, Whitley G, TRAP-­positive osteoclast densities. Increased NGF expres- et al. Microarray analysis of bone marrow lesions in osteoarthritis demonstrates upregulation of genes implicated in ­osteochondral sion appears to be a key feature associated with subchondral turnover, neurogenesis and inflammation. Ann Rheum Dis bone pain in knee OA and could contribute to the previously 2017;76:1764–73. observed association between osteoclasts and OA pain. Our 9. Aso K, Izumi M, Sugimura N, Okanoue Y, Ushida T, Ikeuchi M. findings support a heterogeneous model of OA pain, with dis- ­Nociceptive phenotype alterations of dorsal root ganglia neurons innervating the subchondral bone in osteoarthritic rat knee joints. crete contributions from different compartments in the joint. Osteoarthritis Cartilage 2016;24:1596–603. Different treatment regimens could improve pain due to syno- 10. Stoppiello LA, Mapp PI, Wilson D, Hill R, Scammell BE, Walsh DA. vitis or subchondral pathology, necessitating the development Structural associations of symptomatic knee osteoarthritis. Arthritis of biomarkers to help target therapies to those who will most Rheumatol 2014;66:3018–27. benefit. Other treatments targeting molecular pathways that 11. Sarmanova A, Hall M, Fernandes GS, Bhattacharya A, Valdes AM, Walsh DA, et al. Association between ultrasound-detected­ are shared between tissue compartments will have a greater synovitis and knee pain: a population-­based case-­control study potential for efficacy in unselected OA populations. with both cross-sectional­ and follow-up­ data. Arthritis Res Ther 2017;19:281. 12. Wang X, Jin X, Han W, Cao Y, Halliday A, Blizzard L, et al. Cross-­ ACKNOWLEDGMENTS sectional and longitudinal associations between knee joint effusion synovitis and knee pain in older adults. J Rheumatol 2016;43:121–30. We express our sincere gratitude to all patients who partici- 13. Hill CL, Hunter DJ, Niu J, Clancy M, Guermazi A, Genant H, et al. pated in this study. Synovitis detected on magnetic resonance imaging and its relation to pain and cartilage loss in knee osteoarthritis. Ann Rheum Dis 2007;66:1599–603. AUTHOR CONTRIBUTIONS 14. Watson JJ, Allen SJ, Dawbarn D. Targeting nerve growth factor in All authors were involved in drafting the article or revising it critically pain: what is the therapeutic potential? BioDrugs 2008;22:349–59. for important intellectual content, and all authors approved the final ver- 15. Barthel C, Yeremenko N, Jacobs R, Schmidt RE, Bernateck sion to be published. Dr. Aso had full access to all of the data in the study M, Zeidler H, et al. Nerve growth factor and receptor expression and takes responsibility for the integrity of the data and the accuracy of in rheumatoid arthritis and spondyloarthritis. Arthritis Res Ther the data analysis. 2009;11:R82. Study conception and design. Aso, McWilliams, Walsh. 16. Kaplan DR, Hempstead BL, Martin-Zanca D, Chao MV, Parada LF. Acquisition of data. Aso, Shahtaheri, Hill, Wilson. The trk proto-­oncogene product: a signal transducing receptor for Analysis and interpretation of data. Aso, McWilliams, Walsh. nerve growth factor. Science 1991;252:554–8. 17. Mallett S, Barclay AN. A new superfamily of cell surface pro- REFERENCES teins related to the nerve growth factor receptor. Immunol Today 1. Zhang Y, Nevitt M, Niu J, Lewis C, Torner J, Guermazi A, et al. Fluc- 1991;12:220–3. tuation of knee pain and changes in bone marrow lesions, effusions, 18. Lane NE, Schnitzer TJ, Birbara CA, Mokhtarani M, Shelton DL, and synovitis on magnetic resonance imaging. Arthritis Rheum Smith MD, et al. Tanezumab for the treatment of pain from osteoar- 2011;63:691–9. thritis of the knee. N Engl J Med 2010;363:1521–31. 2. Felson DT, Chaisson CE, Hill CL, Totterman SM, Gale ME, Skinner 19. Sanga P, Katz N, Polverejan E, Wang S, Kelly KM, Haeussler J, et al. KM, et al. The association of bone marrow lesions with pain in knee Efficacy, safety, and tolerability of fulranumab, an anti-nerve­ growth osteoarthritis. Ann Intern Med 2001;134:541–9. factor antibody, in the treatment of patients with moderate to severe 3. Felson DT, Niu J, Guermazi A, Roemer F, Aliabadi P, Clancy M, et osteoarthritis pain. Pain 2013;154:1910–9. al. Correlation of the development of knee pain with enlarging bone 20. Walsh DA, McWilliams DF, Turley MJ, Dixon MR, Franses RE, Mapp marrow lesions on magnetic resonance imaging. Arthritis Rheum PI, et al. Angiogenesis and nerve growth factor at the osteochondral­ 2007;56:2986–92. junction in rheumatoid arthritis and osteoarthritis. Rheumatology 4. Driban JB, Price L, Lo GH, Pang J, Hunter DJ, Miller E, et al. (Oxford) 2010;49:1852–61. Evaluation of bone marrow lesion volume as a knee osteoarthri- 21. Prieto-Potin I, Largo R, Roman-Blas JA, Herrero-Beaumont G, tis biomarker—longitudinal relationships with pain and structural Walsh DA. Characterization of multinucleated giant cells in synovium changes: data from the Osteoarthritis Initiative. Arthritis Res Ther and subchondral bone in knee osteoarthritis and rheumatoid arthri- 2013;15:R112. tis. BMC Musculoskelet Disord 2015;16:226. 5. Hernandez-Molina G, Neogi T, Hunter DJ, Niu J, Guermazi A, 22. Nwosu LN, Allen M, Wyatt L, Huebner JL, Chapman V, Walsh DA, Reichenbach S, et al. The association of bone attrition with knee et al. Pain prediction by serum biomarkers of bone turnover in peo- pain and other MRI features of osteoarthritis. Ann Rheum Dis ple with knee osteoarthritis: an observational study of TRAcP5b and 2008;67:43–7. cathepsin K in OA. Osteoarthritis Cartilage 2017;25:858–65. 6. Reichenbach S, Dieppe PA, Nuesch E, Williams S, Villiger PM, Juni 23. Geurts J, Patel A, Hirschmann MT, Pagenstert GI, Muller-Gerbl M, P. Association of bone attrition with knee pain, stiffness and disabili- Valderrabano V, et al. Elevated marrow inflammatory cells and oste- ty: a cross-­sectional study. Ann Rheum Dis 2011;70:293–8. oclasts in subchondral osteosclerosis in human knee osteoarthritis. 7. Lo GH, McAlindon TE, Niu J, Zhang Y, Beals C, Dabrowski C, J Orthop Res 2016;34:262–9. et al. Bone marrow lesions and joint effusion are strongly and in- 24. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. dependently associated with weight-bearing­ pain in knee osteoar- Development of criteria for the classification and reporting of oste- thritis: data from the Osteoarthritis Initiative. Osteoarthritis Cartilage oarthritis: classification of osteoarthritis of the knee. Arthritis Rheum 2009;17:1562–9. 1986;29:1039–49. 924 | ASO ET AL

25. Walsh DA, Yousef A, McWilliams DF, Hill R, Hargin E, Wilson D. Eval- 35. Reginster JY, Badurski J, Bellamy N, Bensen W, Chapurlat R, uation of a Photographic Chondropathy Score (PCS) for pathologi- Chevalier X, et al. Efficacy and safety of strontium ranelate in the cal samples in a study of inflammation in tibiofemoral osteoarthritis. treatment of knee osteoarthritis: results of a double-­blind, randomised ­Osteoarthritis Cartilage 2009;17:304–12. placebo-­controlled trial. Ann Rheum Dis 2013;72:179–86. 26. Nagaosa Y, Mateus M, Hassan B, Lanyon P, Doherty M. Develop- 36. Laslett LL, Dore DA, Quinn SJ, Boon P, Ryan E, Winzenberg TM, ment of a logically devised line drawing atlas for grading of knee et al. Zoledronic acid reduces knee pain and bone marrow le- osteoarthritis. Ann Rheum Dis 2000;59:587–95. sions over 1 year: a randomised controlled trial. Ann Rheum Dis 27. Mankin HJ, Dorfman H, Lippiello L, Zarins A. Biochemical and met- 2012;71:1322–8. abolic abnormalities in articular cartilage from osteo-arthritic­ human 37. Vaysbrot EE, Osani MC, Musetti MC, McAlindon TE, Bannuru RR. hips. II. Correlation of morphology with biochemical and metabolic Are bisphosphonates efficacious in knee osteoarthritis? A meta-­ data. J Bone Joint Surg Am 1971;53:523–37. analysis of randomized controlled trials. Osteoarthritis Cartilage 28. Haywood L, McWilliams DF, Pearson CI, Gill SE, Ganesan A, Wilson­ 2018;26:154–64. D, et al. Inflammation and angiogenesis in osteoarthritis. Arthritis 38. Hemingway F, Taylor R, Knowles HJ, Athanasou NA. RANKL-­ Rheum 2003;48:2173–7. independent human osteoclast formation with APRIL, BAFF, NGF, 29. Shu SY, Ju G, Fan LZ. The glucose oxidase-­DAB-­nickel method IGF I and IGF II. Bone 2011;48:938–44. in peroxidase histochemistry of the nervous system. Neurosci Lett 39. Xu L, Nwosu LN, Burston JJ, Millns PJ, Sagar DR, Mapp PI, 1988;85:169–71. et al. The anti-­NGF antibody muMab 911 both prevents and 30. Pezet S, McMahon SB. Neurotrophins: mediators and modulators of reverses pain behaviour and subchondral osteoclast numbers in a pain. Annu Rev Neurosci 2006;29:507–38. rat model of osteoarthritis pain. Osteoarthritis Cartilage 2016;24: 31. Aso K, Ikeuchi M, Izumi M, Sugimura N, Kato T, Ushida T, et al. No- 1587–95. ciceptive phenotype of dorsal root ganglia neurons innervating the 40. Niu J, Felson DT, Neogi T, Nevitt MC, Guermazi A, Roemer F, et al. subchondral bone in rat knee joints. Eur J Pain 2014;18:174–81. Patterns of coexisting lesions detected on magnetic resonance im- 32. Jimenez-Andrade JM, Mantyh WG, Bloom AP, Ferng AS, Geffre CP, aging and relationship to incident knee osteoarthritis: the Multicenter Mantyh PW. Bone cancer pain. Ann N Y Acad Sci 2010;1198:173– Osteoarthritis Study. Arthritis Rheumatol 2015;67:3158–65. 81. 41. Moisio K, Eckstein F, Chmiel JS, Guermazi A, Prasad P, Almagor 33. Strassle BW, Mark L, Leventhal L, Piesla MJ, Jian Li X, Kennedy O, et al. Denuded subchondral bone and knee pain in persons with JD, et al. Inhibition of osteoclasts prevents cartilage loss and pain knee osteoarthritis. Arthritis Rheum 2009;60:3703–10. in a rat model of degenerative joint disease. Osteoarthritis Cartilage 42. Boissy P, Saltel F, Bouniol C, Jurdic P, Machuca-Gayet I. Transcrip- 2010;18:1319–28. tional activity of nuclei in multinucleated osteoclasts and its modula- 34. Sagar DR, Ashraf S, Xu L, Burston JJ, Menhinick MR, Poulter CL, tion by calcitonin. Endocrinology 2002;143:1913–21. et al. Osteoprotegerin reduces the development of pain behaviour 43. Weinstein RS, Roberson PK, Manolagas SC. Giant osteoclast for- and joint pathology in a model of osteoarthritis. Ann Rheum Dis mation and long-­term oral bisphosphonate therapy. N Engl J Med 2014;73:1558–65. 2009;360:53–62.