An ultrasonographic study of osteoarthritis of the hand: Synovitis and its relationship to structural pathology and symptoms

Authors


Abstract

Objective

Few studies have examined hand osteoarthritis (OA) pathology using sensitive imaging techniques. The aim of this study was to determine the extent of ultrasound (US)–detected pathology and investigate its relationship with symptoms in hand OA.

Methods

Subjects with symptomatic hand OA and controls were recruited. All underwent clinical and US examination of the small joints of both hands and completed a range of measures of hand pain, stiffness, and function.

Results

Thirty-six subjects with symptomatic OA and 19 control subjects with similar demographics were recruited. US-detected pathology (osteophytes, joint space narrowing, gray-scale synovitis, and power Doppler signal) occurred frequently in symptomatic hand OA (41%, 40%, 46%, and 7% of joints, respectively), and significantly less often in controls (P < 0.001 for all comparisons). Symptomatic joints were more likely to demonstrate US-detected changes of gray-scale synovitis, power Doppler signal, or osteophytes (P < 0.001, P = 0.002, and P < 0.001, respectively). Neither the number of affected joints per individual nor the summative semiquantitative scores for synovitis per individual correlated with symptoms (pain visual analog scale [VAS], global VAS, or Australian/Canadian Osteoarthritis Hand Index).

Conclusion

This study demonstrated extensive synovitis changes as well as the traditional structural radiographic findings of hand OA. Symptomatic joints were significantly more likely to demonstrate US-detected structural changes or inflammation in symptomatic hand OA; however, the extent of changes in individual joints or in individuals did not correlate with the degree of symptoms, which may relate to both the assessment tools and the complex nature of pain.

INTRODUCTION

Osteoarthritis (OA) is not only a common cause of pain and morbidity but has a major impact on health economics in the western world (1). Despite this, little is understood about peripheral sources of pain in OA, which limits targeted therapy (2, 3). Hand OA is studied less than large joint disease, and epidemiologic studies of hand OA usually investigate radiographic rather than symptomatic OA (4–8), which may be considered more clinically relevant. Additionally, few studies investigate symptoms such as pain, which is the most important manifestation of the disease (9).

Although conventional radiographs have traditionally been used to diagnose and assess OA, there is discordance between radiographic changes and symptoms in OA (6, 10). In addition, articular cartilage is aneural and unlikely to be the primary source of pain in OA. Given this, investigation into the relationship between other potential peripheral sources of pain, such as the synovium, is warranted (2, 3, 11), although this has been methodologically difficult until the advent of modern imaging techniques.

Ultrasonography (US) has been found to be a valid and reliable method of imaging synovial inflammation in small joints in rheumatoid arthritis (12). In addition, we have recently demonstrated that US detects more osteophytosis than conventional radiography in the small joints of the hands (13). However, the prevalence and importance of US-detectable pathology, particularly synovitis, in hand OA is not known.

The aim of this study was to use US to evaluate the hands of people with and without symptomatic OA to examine the extent and severity of US-detected pathology and the relationship between symptoms and structural abnormalities in this common condition.

PATIENTS AND METHODS

Approval for this study was obtained from the local research ethics committee. Patients attending the musculoskeletal clinics of the Leeds Teaching Hospitals National Health Service Trust were recruited and gave informed consent. Two cohorts were examined: subjects with symptomatic hand OA (SOA) and control subjects. The SOA group included subjects with a clinical diagnosis of symptomatic hand OA who either met American College of Rheumatology (ACR) criteria (14) or had radiographic changes consistent with hand OA who reported hand pain on direct questioning. Several definitions of hand OA currently exist; however, it is recognized that diagnosis and classification of hand OA remain problematic (15). For example, the ACR criteria have recently been criticized for requiring involvement of multiple joints for diagnosis (16); consistent with several recent studies (17, 18), our criteria allowed inclusion of subjects with less extensive disease, for example, subjects with primarily base-of-thumb disease. Subjects were excluded if an inflammatory arthritis was present or suspected, or if they had an unexplained elevated C-reactive protein level or rheumatoid factor prior to baseline.

Controls subjects were subjects recruited from the musculoskeletal clinics who answered negatively to direct verbal questioning concerning whether they had any pain in their hands. The typical patient may have been attending the clinics with OA of the knee, greater trochanteric bursitis, or mechanical foot pain. Radiographs were not performed on these subjects because we were interested in comparing the relationship between subjects with symptomatic hand OA and an age-matched cohort who did not consider themselves to have significant hand pain.

A single investigator performed painful, tender, and enlarged joint counts. Subjects were asked to identify the single most painful joint and completed the following: a 10-cm visual analog scale (VAS) for pain in the most painful joint (painful joint VAS), overall pain in both hands (pain VAS), overall score for all the ways the hands affect the individual (global VAS), and the Australian/Canadian Osteoarthritis Hand Index (AUSCAN) (19).

All patients underwent US using a Philips HDI 5000 sonoCT scanner (Phillips, Eindhoven, The Netherlands) with a multilinear 15–7-MHz hockey stick probe by a single physician ultrasonographer (HIK). Power Doppler was assessed with a pulse repetition frequency of 750 Hz and medium wall filter. Gain was adjusted until background signal was removed. Joints were assessed globally, scanning on both the dorsal and palmar aspects of the hand in both the longitudinal and transverse planes. Each subject had 30 joints imaged: bilateral first carpometacarpal joints (CMC), metacarpophalangeal joints 1–5 (MCP), proximal interphalangeal joints 1–5 (PIP), and distal interphalangeal joints 2–5 (DIP). Four pathologies were evaluated with US.

Gray-scale synovitis.

Gray-scale (GS) synovitis was assessed, defined as a composite of synovial hypertrophy and effusion according to the Outcome Measures in Rheumatology Clinical Trials (OMERACT) definition (20). Synovitis was graded on a scale of 0–3 according to a semiquantitative scale (21) (Figure 1).

Figure 1.

a, Radiograph of the 3rd proximal interphalangeal joint demonstrating severe erosive osteoarthritis. b, Ultrasound of the same joint, dorsal longitudinal view demonstrating gray-scale synovitis (arrows). c, Ultrasound of the same joint, dorsal longitudinal view demonstrating power Doppler signal within the gray-scale synovitis. PP = proximal phalanx; MP = middle phalanx; O = osteophyte.

Power Doppler signal.

Power Doppler signal (PDS) was assessed with areas of color signal within the joint capsule, when the gain was adjusted to exclude only background noise. Power Doppler was graded on a scale of 0–3 according to a semiquantitative scale (21) (Figure 1).

Osteophytes.

Osteophytes were defined as cortical protrusions seen in 2 planes (21) as defined and illustrated in a previous report (13). Osteophytes were scored by counting the number of osteophytes at each joint (Figure 1).

Joint space narrowing.

A surrogate of radiographic joint space narrowing was used. Intrinsic to the physical properties of sound, sound waves are reflected as echoes on meeting bone, and only the superficial cortical surface of the closest joint is able to be visualized. This means that the joint space in the most central portion of the joint is unable to be visualized. As a result of the same principle, osteophytes protruding over the joint space at times prevented even the superficial cortical surfaces of the joint from being visualized. The joint space was considered reduced when the space between the superficial cortical margins appeared reduced, or was assumed to be reduced because osteophytes prevented visualization as defined and illustrated in a previous report (13).

Statistical analysis.

Data were summarized using the mean (95% confidence interval) for normally distributed, continuous variables, and the median (interquartile range) for non-normally distributed or ordinal variables. The study was powered on the relationship between the imaging changes (presence of synovitis, power Doppler, osteophytes, and joint space narrowing) and joint pain (most painful VAS), which was to be assessed using forward stepwise multiple linear regression. With 4 potential predictors, testing at the 5% significance level with 80% power, and assuming a minimum R2 of 0.3, 33 patients were needed to demonstrate that the structural predictors explain a statistically significant amount of the variation in joint pain.

In the secondary analyses, the statistical test used was indicated next to the P value, i.e., Spearman's rank correlation, Student's t-test, chi-square, Mann-Whitney U test. All hypotheses were tested at the 5% level of significance, with family-wise corrections for multiple comparisons made according to the Holm modification of the Bonferroni correction (22).

Intrareader reliability for osteophytes and joint space narrowing was determined by reimaging a subgroup of 10 subjects 4 weeks after the initial US assessment. Because the stability of synovitis (GS) and PDS in hand OA is unknown, intrareader reliability for synovitis (GS) and PDS was determined by reading stored images at a minimum of 12 weeks after acquisition. Kappa (κ) and quadratic-weighted kappa (κw) values were presented. Weighted kappa has been demonstrated to approximate the intraclass correlation coefficient in both formula and interpretation when quadratic weights are used (23). Levels of agreement measured by both κ and κw were assessed according to the rules of thumb proposed by Landis and Koch (<0: poor; 0.00–0.20: slight; 0.21–0.40: fair; 0.41–0.60: moderate; 0.61–0.80: substantial; 0.81–1.00: almost perfect) (24). Analyses were undertaken using SPSS 15.0 software (SPSS, Chicago, IL).

RESULTS

Demographic and symptom characteristics.

Thirty-six subjects were recruited into the SOA group, and 31 (86.1%) met the ACR criteria for hand OA; of the remaining 5 subjects, 4 had base-of-thumb OA and 1 had predominant symptoms in the right first MCP. In 1 subject, 3 joints were not able to be examined due to previous finger amputation, therefore, 1,077 joints were examined in total. Twenty subjects were recruited into the control group, although 1 was subsequently excluded prior to analysis of results because wrist pain was believed to confound clinical outcome measures. Therefore, in the control group, 570 joints were examined. Demographics and symptom levels are presented in Table 1 and there were no significant differences between groups, although there was a trend toward more men in the control group. As expected, the control group had significantly less symptoms than the OA group, with the exception of early morning stiffness. On clinical examination, some control subjects who reported no hand pain on questioning did report some joints to be painful or tender, often reportedly the result of recent low-grade trauma.

Table 1. Baseline variables in the symptomatic OA hand group and control group*
 OA (n = 36)Controls (n = 19)P
  • *

    Values are the median (interquartile range) or mean ± SD unless otherwise indicated. OA = osteoarthritis; VAS = visual analog scale; NA = not assessed; AUSCAN = Australian/Canadian Osteoarthritis Hand Index; JSN = joint space narrowing; GS = gray scale.

  • Mann-Whitney U test (not corrected for ties).

  • Pearson's chi-square.

  • §

    Student's t-test.

  • Significant values; Mann-Whitney U test (not corrected for ties).

Age, years58 (53.2–66.5)58 (51–71)0.838
Female sex, no. (%)31 (86.1)13 (68.4)0.119
Years of education11.8 ± 3.311.4 ± 4.00.68§
Most painful VAS76.5 (46.25–91.5)NANA
Pain VAS65 (41.0–83.0)2.5 (1.75–9.5)< 0.001
Global VAS65.5 (45.0–84.25)2.0 (0.75–9.5)< 0.001
AUSCAN pain299.0 (199.0–354.25)59.0 (10.5–139.5)< 0.001
AUSCAN stiffness61.0 (32.0–80.75)10.0 (2.0–31.25)< 0.001
AUSCAN function523.0 (433.25–713.75)90.0 (19.75–216.2)< 0.001
Painful joint count8.0 ± 6.63.1 ± 4.60.008§
Tender joint count5.47 ± 5.41.6 ± 3.80.011§
No. of joints in an individual with osteophytes11.5 (7.25–17.0)6.0 (3.0–15.0)0.073
No. of joints in an individual with JSN11.5 (7.0–17.0)7.0 (2.0–12.0)0.014
No. of joints in an individual with synovitis (GS)12.0 (10.0–18.75)11.0 (3.0–15.0)0.092
No. of joints in an individual with power Doppler signal1.5 (0.0–3.0)0 (0.0–3.0)0.02

The intrareader reliability for the presence of US-detected osteophytosis at a joint was almost perfect with a κ value of 0.83 (percentage exact agreement [PEA] 91.8%), for the presence of US-detected joint space narrowing was substantial with a κ of 0.64 (PEA 82.2%), for semiquantitative synovitis (GS) was substantial with a κw of 0.62 (PEA 54.0%, percentage close agreement [PCA] 90.0%), and for semiquantitative PDS was almost perfect with a κw of 0.87 (PEA 92.0%, PCA 98.0%). We have previously published the construct validity of US in detecting joint space narrowing, and discussed the difficulties in accurately measuring joint space narrowing with current US technology in OA of the small joints of the hands (13).

Distribution of US-detected pathology in the SOA group.

US-detected pathology was common. The number of joints in an individual with US-detected pathology is presented in Table 1. The distribution of US-detected pathology is presented by joint in Table 2.

Table 2. The distribution of ultrasound-detected pathology by joint*
 DIPPIPMCPCMCTotal% of total
543254321543211
  • *

    DIP = distal interphalangeal joint; PIP = proximal interphalangeal joint; MCP = metacarpophalangeal joint; CMC = carpometacarpal joint; SOA = subjects with symptomatic hand osteoarthritis; GS = gray scale; JSN = joint space narrowing.

Left                 
 No. of joints imaged                 
  SOA363636363636363636363636363636540100
  Control191919191919191919191919191919285100
 No. of joints with synovitis (GS)                 
  SOA20232824211317152148712152325146.4
  Control7687544386558768932.9
 No. of joints with power  Doppler synovitis                 
  SOA013612420000328325.9
  Control00000010001101262.2
 No. of joints with JSN                 
  SOA291725231311141614000362819936.8
  Control136109455660001197527.8
 No. of joints with osteophytes                 
  SOA231926261713151920012562221439.6
  Control126810465581014378029.6
Right                 
 No. of joints imaged                 
  SOA353636353636363536363636363636537100
  Control191919191919191919191919191919285100
 No. of joints with synovitis (GS)                 
  SOA152023221614161423910814132624345.2
  Control667968771036798610538.9
 No. of joints with power  Doppler synovitis                 
  SOA111634342006307417.6
  Control011300110001211124.4
 No. of joints with JSN                 
  SOA232530301817151519103363223744.1
  Control14799457471012198029.6
 No. of joints with osteophytes                 
  SOA1918293016161617231028102022541.9
  Control105910488681016388732.2

The pattern of US joint involvement was symmetric with a predominance of pathology in the DIP joints and CMC 1, followed by PIP involvement, with less involvement of the MCP joints (Table 2). Both GS and power Doppler, when present, were more often scored toward the lower end of the semiquantitative scale. The distribution and severity of US-detected pathology was different depending on which row (i.e., CMC, MCP, PIP, or DIP) was examined. Kruskal-Wallis tests indicated there were overall differences between the joint groups for each of the US parameters (all P < 0.001). Bonferroni-corrected post hoc tests were then performed to identify significant differences between pairs of joint groups, and the P values are presented in Table 3.

Table 3. Relationship between ultrasound-detected pathology and joint row*
 CMCMCPPIPDIP
MWU ZPMWU ZPMWU ZPMWU ZP
  • *

    MWU = Mann-Whitney U; PDS = power Doppler signal; see Table 2 for additional definitions.

  • Significant after applying the Holm modification of the Bonferroni correction.

MCP        
 No. of osteophytes−10.279< 0.001--−11.812< 0.001−15.466< 0.001
 JSN−15.215< 0.001--−11.314< 0.001−17.08< 0.001
 Synovitis (GS)−6.751< 0.001--−6.784< 0.001−8.773< 0.001
 PDS−5.214< 0.001--−1.8110.07−1.5110.131
PIP        
 No. of osteophytes−1.2510.211----−4.24< 0.001
 JSN−6.343< 0.001----−7.167< 0.001
 Synovitis (GS)−1.5300.126----−1.0040.284
 PDS−3.674< 0.001----−0.2460.805
DIP        
 No. of osteophytes−1.4080.159------
 JSN−2.1720.030------
 Synovitis (GS)−0.8910.373------
 PDS−3.785< 0.001---- -

In individual joints, there were strong correlations between US-detected pathologic domains. After correction for multiple comparisons, joint space narrowing correlated with synovitis (GS) (Spearman's ρ = 0.465, P < 0.001) and osteophytes (ρ = 0.765, P < 0.001), synovitis (GS) correlated with osteophytes (ρ = 0.534, P < 0.001) and PDS (ρ = 0.307, P < 0.001), and osteophytes correlated with PDS (ρ = 0.244, P < 0.001).

Distribution of US-detected pathology in the control group.

US-detected pathology was less common in the control group than in the SOA group. The number of joints in an individual with US-detected pathology is presented in Table 1. The distribution of US-detected pathology is presented by joint in Table 2. The distribution and relative frequency of US-detected pathology by joint occurred in a similar distribution to that seen in the symptomatic OA group. Low-grade synovitis (GS) and PDS were present in several joints, particularly the MCP joints, where grade 1 synovitis (GS) occurred in 53 (27.9%) of 190 joints and grade 1 PDS occurred in 8 (4.2%) of 190, in the absence of pain, tenderness, osteophytosis, or joint space narrowing.

US findings and relationship with symptoms in the SOA group.

The most painful joint had more osteophytes (χ2 = 90.36, 5 df, P < 0.001), joint space narrowing (χ2 = 40.43, 1 df, P < 0.001), synovitis (GS) (χ2 = 28.49, 3 df, P < 0.001), and PDS (χ2 = 98.07, 3 df, P < 0.001). However, when the US scores in the most painful joint were dichotomized (each absent/present), no substantive associations were found with pain VAS (osteophytes: Pearson's r = 0.175, P = 0.308; joint space narrowing: r = −0.040, P = 0.818; synovitis [GS]: r = −0.245, P = 0.15; and PDS r = −0.18, P = 0.293) (Figure 2). The lack of association between US findings and VAS prevented the multivariate model from being constructed; guidelines for linear regression recommend that the independent variables show at least some evidence of correlation with the dependent variable on a univariate basis before they are put forward into the multivariate model.

Figure 2.

Visual analog scale (VAS) scores for the most painful joint against the ultrasound pathology score. A, VAS scores for joints with osteophyte counts of 0–5. B, VAS scores for joints without (score 0) and with (score 1) joint space narrowing. C, VAS scores for joints with semiquantitative gray-scale (GS) synovitis scores of 0, 1, 2, and 3. D, VAS scores for joints with semiquantitative power Doppler (PD) scores of 0, 1, 2, and 3.

At the individual joint basis, painful joints were significantly more likely to have osteophytes (χ2 = 156.76, 5 df, P < 0.001), joint space narrowing (χ2 =146.517, 1 df, P < 0.001), synovitis (GS) (χ2 = 123.342, 3 df, P < 0.001), and PDS (χ2 = 55.27, 3 df, P < 0.001) than nonpainful joints. Exploratory analyses of the association between whether a joint was painful or not and the osteophyte count (Mann-Whitney U [MWU] Z = −11.05, P < 0.001), semiquantitative scores for synovitis (GS) (MWU Z = −12.03, P < 0.001), and PDS (MWU Z = −7.25, P < 0.001) demonstrated low P values, which may be worth exploring in further studies focused and statistically powered to address these associations.

Tender joints were also significantly more likely to have osteophytes (χ2 = 137.43, 5 df, P < 0.001), joint space narrowing (χ2 = 94.84, 1 df, P < 0.001), synovitis (GS) (χ2 = 108.02, 3 df, P < 0.001), and PDS (χ2 = 75.12, 3 df, P < 0.001). Exploratory analyses of the association between whether a joint was tender or not and the osteophyte count (MWU Z = −11.27, P < 0.001), the semiquantitative scores for synovitis (GS) (MWU Z = −9.839, P < 0.001), and PDS (MWU Z = −8.11, P < 0.001) also demonstrated low P values, which may be worth exploring in further studies focused and statistically powered to address these associations.

In an individual, US pathology was considered either by summating the semiquantitative scores for each US-detected pathology at each joint (here termed summative degree) or dichotomizing the presence or absence of pathology at each joint, therefore considering the number of joints in an individual with pathology (termed extent). When the summative degree was examined, the cumulative score for synovitis (GS), PDS, osteophytes, and joint space narrowing in an individual did not correlate with global pain VAS, AUSCAN pain, stiffness, or function subscales. When the extent of pathology was examined, neither the number of joints in an individual with synovitis (GS), number of joints with PDS, number of joints with osteophytes, nor joint space narrowing in an individual correlated with global VAS, AUSCAN pain, stiffness, or function subscales. The correlation coefficients and P values are presented in Table 4.

Table 4. Spearman's correlations between the number of joints in an individual with pain, tenderness, and ultrasound pathology and an individual's symptom scores*
 No. with synovitis (GS)No. with synovitis (PD)No. with osteophytesNo. with JSN
  • *

    None of the correlations were significant after application of the Holm modification of the Bonferroni correction. GS = gray scale; PD = power Doppler; JSN = joint space narrowing; VAS = visual analog scale; AUSCAN = Australian/Canadian Osteoarthritis Hand Index.

Pain VAS    
 ρ0.001−0.3110.0490.128
 P (2-tailed)0.9950.0780.7880.479
Global VAS    
 ρ−0.009−0.2790.1450.198
 P (2-tailed)0.9610.1100.4140.262
AUSCAN pain    
 ρ−0.045−0.078−0.1750.012
 P (2-tailed)0.7950.6520.3070.945
AUSCAN stiffness    
 ρ0.2130.0980.1810.217
 P (2-tailed)0.2120.5680.2920.204
AUSCAN function    
 ρ0.051−0.0430.0670.113
 P (2-tailed)0.7660.8050.7000.510

Comparison of US findings between the SOA and control groups.

Joints in the SOA group were significantly more likely to have osteophytes (χ2 = 21.19, 1 df, P < 0.001), joint space narrowing (χ2 = 28.76, 1 df, P < 0.001), synovitis (GS) (χ2 = 21.60, 1 df, P < 0.001), and PDS (χ2 = 9.38, 1 df, P = 0.002) than the control group even after correction for multiple comparisons. Individuals in the SOA group displayed a trend toward a greater number of joints in each individual with US-detected osteophytes, joint space narrowing, synovitis (GS), and PDS; however, the findings were not significant after correction for multiple comparisons (Table 1).

DISCUSSION

The prevalence and importance of soft tissue pathology in hand OA is not well documented. In the current study, US-detected pathology was seen in a distribution consistent with epidemiologic studies of radiographic hand OA, predominantly affecting the first CMC and second and third DIP joints bilaterally. Importantly, GS synovitis and PDS occur frequently in symptomatic hand OA and less often in the joints of controls. Recently, Tan et al utilized magnetic resonance imaging (MRI) to examine the structural pathology in a single joint in OA of the hand, but did not examine the distribution of pathology in other joints of the hand, or the relationship with symptoms (25, 26). The distribution and extent of synovial inflammation in OA of the hands has not previously been investigated with US or MRI; however, these findings are in keeping with studies examining the frequency of synovial inflammation in a large US study of knee OA (27).

Community studies have demonstrated that a large percentage of small joints in the hand with severe radiographic changes are asymptomatic (6, 10, 28). This is the first attempt to utilize modern imaging to better understand the relationship between symptoms and pathology in OA of the hands. In the SOA group, at the level of the individual joint, painful joints were more likely to demonstrate US pathology. Similarly, the most painful joint was more likely than other joints to have US-detected pathology; however, the degree of pain in that joint did not correlate with the degree of US changes. As might be expected, symptomatic individuals had more joints involved ultrasonographically than the control group. However, neither an individual's cumulative semiquantitative score nor the number of joints with US pathology correlated with the degree of symptoms. These findings suggest that symptomatic joints are more likely to demonstrate US-detected pathology, but that the degree of pathology does not correlate with the degree of symptoms in individual joints or in individuals.

Synovial pathology has variably been associated with some pain measures in MRI and US studies of painful knee OA (29–32). In the current study, painful joints were more likely to have US-detected synovitis (GS) and PDS. Given this, further US studies are warranted to further establish the relationship between US-detected pathology and peripheral sources of pain in OA.

This study has limitations. A clinically relevant definition of symptomatic hand OA was used; subjects either met ACR criteria or had symptoms and radiographic structural changes. This approach has recently been used in some studies (17) because other diagnostic criteria for hand OA are problematic (15).

The sample size of 36 subjects with SOA may be considered small; however, according to a priori power calculations, 33 subjects would be needed to demonstrate associations between pain and US-detected pathology in the most painful joint.

Another potential limitation was that the control group answered negatively to verbal questioning about the presence of hand pain, but when symptoms were assessed with validated outcome measures, such as the VAS or AUSCAN, low levels of symptoms in the control group became apparent. This may reflect the sensitivity and specificity of the outcome tools, and has important trial implications for studies looking at asymptomatic elderly individuals.

The control cohort had structural pathology. Between 75% and 85% of elderly community-based subjects in the Framingham cohort have structural changes in the small joints of the hand; however, the majority are asymptomatic (33, 34). Given this, it was expected that our control cohort (who were chosen to be a demographically similar but relatively pain-free group compared with the SOA group) would have structural pathology. The study was designed with this in mind, as it was thought that a young control group with a high likelihood of being structurally normal may have resulted in overestimation of the association between structural pathology and symptoms.

We did not examine US-detected cartilage thickness or erosions. US has been demonstrated to be less sensitive to the presence of OA erosions than conventional radiography (35), and noninvasive US techniques need further investigation before being considered valid measures of cartilage loss in hand OA. Given the poor relationship between radiographic changes of hand OA and symptoms (4), we focused on the pathologies that can be better detected with US than conventional radiography.

Another important clinical study limitation was that the ultrasonographer was not blinded to diagnosis or clinical features; however, in practice, blinding a sonographer to joint deformities and joint tenderness is not feasible.

As well as limitations in the measurement of US pathology, there are limitations in using the simple instruments used here to measure the complex phenomena of pain; socioeconomic factors, illness perception, mental state, and other factors have previously been found to be confounders of symptom assessment in hand OA (18). This must necessarily limit the strength of reported structure–pain relationships.

In this study, novel use of US has allowed the phenotype of OA to be examined with respect to structural changes that cannot be studied with conventional radiography. In this way, it has systematically been demonstrated that US-detected abnormalities of GS synovitis, Doppler signal, osteophytosis, and joint space narrowing are common and more likely to occur in SOA subjects than controls. The distribution of changes in SOA is similar to the patterns demonstrated in conventional radiography studies (4, 10); however, it is of interest to recognize the involvement of MCP joints, as they are often not included in population studies of OA.

It is known that radiographic changes do not correlate well with symptoms in hand OA (10). This study is the first attempt to examine the relationship between US-detectable changes and symptoms in hand OA. While symptomatic individuals and symptomatic joints are more likely to demonstrate US-detected pathology, the degree of pain in individual joints does not correlate well with the degree of US-detected changes. In addition, in an individual, the extent or summative degree of US-detected pathology does not correlate with symptoms. The causes of symptoms in hand OA remain complex, with US-detected structural changes and inflammation possibly having a part to play.

AUTHOR CONTRIBUTIONS

Dr. Conaghan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Keen, Grainger, Wakefield, Emery, Conaghan.

Acquisition of data. Keen.

Analysis and interpretation of data. Keen, Hensor, Grainger, Wakefield, Emery, Conaghan.

Manuscript preparation. Keen, Grainger, Wakefield, Emery, Conaghan.

Statistical analysis. Keen, Hensor.

Ancillary