Chronic gout is characterised by the presence of tophi, chronic granulomatous synovitis and bone erosion. Tophi are collections of monosodium urate (MSU) crystals surrounded by chronic inflammatory cells including giant cells and connective tissue and may be located in subcutaneous and synovial tissues.1 Bone erosion is a strong predictor of musculoskeletal disability in many forms of arthritis, including gout.2^–4 The underlying mechanisms of bone erosion in gout remain speculative. Some histological case reports have demonstrated tophus adjacent to bone erosion.5 However, recent ultrasound studies have implicated hypervascularity as a factor associated with gouty bone erosions, suggesting that the pathogenesis of bone erosion in gout may have similarities with erosive synovial arthropathies such as rheumatoid arthritis.6 7

Computed tomography (CT) allows excellent visualisation of tophi in patients with gout.7^–10 We have recently demonstrated that CT is able to assess subcutaneous tophus volume in a reliable and reproducible manner.11 Furthermore, CT has superior capability over both plain radiography and magnetic resonance imaging to detect bone erosion in erosive arthropathies such as rheumatoid arthritis.12 13 Through systematic assessment of both tophi and bone erosion, CT has the potential to provide further understanding of the relationship between tophus and bone erosion in chronic gout. The aim of the current analysis was to understand further the mechanisms of bone erosion in gout using CT as a non-invasive imaging technique. In particular, we wished to determine the relationship between bone erosion and tophus, whether all joints with bone erosion also have intraosseous tophus and whether intraosseous tophi have distinct characteristics.

METHODS

Patients

This was a predefined study of paired plain radiographs and CT scans of 798 individual hand and wrist joints in 20 patients with chronic gout attending rheumatology outpatient clinics. Two joints were not available for assessment because of previous amputation. As previously described,4 11 the inclusion criteria were the diagnosis of gout as defined by the Wallace classification for acute gout,14 the ability to provide written informed consent and the ability to flex and abduct the shoulders fully (required for positioning in the CT scanner). Women of childbearing age and patients with acute gout flares were excluded from the study. The patients were assessed to obtain clinical characteristics related to gout, including disease duration, the total number of subcutaneous tophi, serum urate, C-reactive protein and dominant hand grip strength using a Jamar hand dynamometer. The Northern Regional Ethics Committee approved this study. All patients provided written informed consent before inclusion in the study.

Plain radiographs

Plain radiographs of both hands were obtained on the same day as the clinical assessments and CT scans. The plain radiographs were scored by a rheumatologist (ND), using a radiographic scoring system that we have recently validated using a separate set of radiographs, with excellent interobserver and intraobserver reliability.15 This method includes scoring each joint for erosion (0–5) using the Sharp/van der Heijde scoring method for rheumatoid arthritis, with additional scoring of the distal interphalangeal joints (DIPJ).

CT scans

The CT scans of the hands were performed on a Philips Brilliance 16-slice scanner (Philips Medical Systems, Best, The Netherlands) as previously described.11 CT scans were analysed for erosion and tophus in each joint by two independent observers; a musculoskeletal radiologist (BC) and a rheumatologist (ND) with experience in the assessment of CT analysis in gout. The CT scans were analysed more than 6 months after the radiographic erosion scoring and both observers were blinded to the results of the radiographic erosion scores. For the purposes of this analysis, serial axial images were examined for erosion and tophus.

Each joint was first scored for the presence or absence of CT bone erosion. Bone erosion was defined as evidence of complete cortical disruption in the axial plane. When readers differed, the presence or absence of erosion was decided by consensus. Pre-consensus, there was 89.1% agreement for the presence of CT bone erosion.

Each joint was then measured for CT erosion diameter, by recording the longest distance of cortical defect in the axial plane. If more than one erosion within a single joint was identified, the largest erosion was measured. The mean erosion diameter was recorded for each joint. The intraclass correlation coefficient for erosion diameter was 0.835 (0.812–0.855).

Individual joints were also scored for the presence of tophus adjacent to or within the joint of interest on CT scanning. CT tophus was defined as a well-defined opacity, with density greater than that of surrounding soft tissue but less than that of bone. For each tophus, Hounsfield units and the presence of calcification was also recorded, as previously described.11 When readers differed, the presence of tophus adjacent to or within the joint of interest was decided by consensus. Pre-consensus, there was 87.7% agreement for the presence of tophus.

Each joint was then assessed for the presence of intraosseous tophus on CT scanning, as defined by the presence of tophus within bone. When readers differed, the presence of intraosseous tophus was decided by consensus. Pre-consensus, there was 88.2% agreement for the presence of intraosseous tophus.

The longest diameter of the tophus in the axial plane was recorded. Previous analysis demonstrated excellent correlation between tophus axial diameter and three-dimensional volume assessments (r  =  0.894, p<0.001, Dalbeth et al, unpublished data). The mean tophus size was recorded for each joint. The intraclass correlation coefficient for tophus diameter was 0.804 (0.778–0.827).

Statistical analysis

Medians with ranges and percentages were used to describe the clinical characteristics, radiographic and CT scores. Radiographic and CT erosion data were analysed by χ² tests and Spearman correlations. As lesions were nested within individuals and were likely to be correlated, two additional approaches were used; first, a Spearman correlation analysis of the average score for each patient and, second, a general estimating equation (GEE) approach. This approach was used to provide adjustment for repeated measures within subjects. The Genmod procedure of SAS was used. Tests for linear and higher order trend were conducted using contrast coefficients. Comparisons between intraosseous and non-intraosseous tophi were made using χ² analysis and Student’s t tests. All tests were two tailed and p<0.05 was considered significant.

RESULTS

Patient and individual joint characteristics

The clinical characteristics of the patients have been reported previously.4 11 In brief, there were 19 (95%) men with a median age of 58 years (range 38–75). The median disease duration was 17 years (range 1–50) and grip strength was 31 kg (range 4–71). Median serum urate was 0.48 mmol/l (range 0.25–0.68) and C-reactive protein was 2.0 mg/l (range 0.5–20.0). The majority of patients (16/20, 80%) had subcutaneous tophi and the median number of subcutaneous tophi was 10 (range 0–244).

CT erosions were present in 237/798 (29.7%) individual joints. The median erosion diameter was 4.8 mm (range 0.8–24.3). Tophus adjacent to or within a joint was present in 265/798 (33.2%) individual joints on CT scanning. The median tophus diameter was 7.4 mm (range 1.0–44.6). Of these tophi, 192/265 (72.5%) had an intraosseous component.

The location of plain radiographic and CT abnormalities is shown in table 1. Overall, the radius/ulna and proximal interphalangeal joints were most frequently affected. However, when expressed as the rate of affected joints per patient at each site, these differences did not reach statistical significance (analysis of variance, all p>0.26).

Radiographic erosions were present in 270/798 (33.8%) individual joints. There were 84 (10.5%) joints with a radiographic erosion score of 1, 59 (7.4%) with a score of 2, 29 (3.6%) with a score of 3, 40 (5.0%) with a score of 4 and 58 (7.3%) with a score of 5.

Relationship between radiographic and CT erosion

Of the 798 individual joints assessed, there was agreement regarding the presence of erosion between plain radiographs and CT in 697 joints (87.3%). In 494 joints, no erosions were identified by plain radiography or CT, in 67 joints erosions were identified by radiography but not CT, in 34 joints erosions were identified by CT but not radiography and in 203 joints erosions were identified by both modalities. Mismatch between radiography and CT erosion was greatest at the DIPJ; of the 67 joints with erosion identified on plain radiography but not by CT, 22 (32.8%) were DIPJ. In 104 DIPJ, no erosions were identified by plain radiography or CT; in 22 DIPJ, erosions were identified by radiography but not CT; in four DIPJ, erosions were identified by CT but not radiography; and in 29 DIPJ, erosions were identified by both modalities.

With increasing radiographic erosion score, the percentage of joints with CT erosion increased (fig 1A, χ² analysis, p<0.001 for trend). For those joints with a radiographic erosion score of 4 or 5, 97/98 (99%) also had CT erosion. In addition, there was a strong relationship between the radiographic erosion scores and CT erosion diameter (fig 1B, p<0.001 for linear trend). The individual joint radiographic erosion scores were strongly correlated with the CT erosion diameters (r  =  0.78, p<0.001). Average radiographic erosion scores for each individual were also highly correlated with average CT erosion diameters for each individual (r  =  0.95, p<0.001). Using GEE analysis to account for repeated measures within each patient, the relationship between individual joint radiographic erosion scores and CT scores was still observed (p<0.001 for linear trend).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

The relationship between radiographic erosion score and computed tomography (CT) erosion. (A) The percentage of joints with erosion present on CT scanning. (B). The median (interquartile range) CT erosion diameter.

Relationship between radiographic erosion and intraosseous tophus on CT

There were 179/270 (66.3%) joints with radiographic erosion that had visible intraosseous tophus on CT scanning. Examples of joints with intraosseous tophi are shown in fig 2. With increasing radiographic erosion scores, the percentage of joints with intraosseous tophus increased (fig 3A, χ² analysis, p<0.001 for trend). For those joints with a radiographic erosion score of 4 or 5, 96/98 (98%) had evidence of intraosseous tophus. In addition, there was a strong relationship between the radiographic erosion scores and intraosseous tophus size as measured by the axial diameter (fig 3B, p<0.001 for linear trend). The individual joint radiographic erosion scores were strongly correlated with the intraosseous tophus diameters (r  =  0.72, p<0.001). Average radiographic erosion scores for each individual were also highly correlated with average intraosseous tophus diameters for each individual (r  =  0.92, p<0.001). Using GEE analysis to account for repeated measures within each patient, the relationship between individual joint radiographic erosion scores and intraosseous tophus diameters was still observed (p<0.001 for linear trend).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Representative computed tomography (CT) scans from three separate individuals showing intraosseous tophi associated with radiographic and CT bone erosion. Arrows indicate the areas of bone erosion with associated intraosseous tophi. For all joints indicated by the arrows, radiographic erosion scores were 5. These examples demonstrate the spectrum of intraosseous tophus size ranging from small tophi in (A) to large tophi in (B) and (C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The relationship between radiographic erosion scores and intraosseous tophus in individual joints. (A) The percentage of joints with intraosseous tophus on computed tomography scanning. (B) The median (interquartile range) diameter of intraosseous tophus.

Relationship between CT erosion and intraosseous tophus

For those joints with erosion on CT, 194/237 (81.8%) had visible intraosseous tophus. Of the joints with CT erosions greater than 5 mm, 106/112 (94.6%) had visible intraosseous tophus, and all (56/56) erosions greater than 7.5 mm had intraosseous tophus. There was a strong correlation between CT erosion diameter and intraosseous tophus diameter in individual joints (r  =  0.92, p<0.001, fig 4). Average CT erosion diameters for each individual were also highly correlated with average intraosseous tophus diameters for each individual (r  =  0.93, p<0.001). Using GEE analysis to account for repeated measures within each patient, the relationship between individual joint radiographic erosion scores and intraosseous tophus diameters persisted (p<0.001 for linear trend).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

The relationship between computed tomography (CT) erosion size and intraosseous tophus diameter.

Comparison between intraosseous and non-intraosseous tophi on CT

There were 194 intraosseous tophi identified and 73 tophi that were adjacent to or within a joint but not extending into bone (non-intraosseous tophi). There was no difference in Hounsfield units between intraosseous and non-intraosseous tophi (mean 191.9 (SE 3.8) and 189.9 (SE 6.2), respectively, p = 0.78). Similarly, there was no difference in the presence of calcification; 59/194 (30.4%) of intraosseous tophi and 16/73 (22%) of non-intraosseous tophi had evidence of calcification (p = 0.17). However, intraosseous tophi were larger than non-intraosseous tophi; the mean axial diameter of intraosseous tophi was 9.869 mm (SE 0.541) compared with 7.137 mm (SE 0.496) for non-intraosseous tophi (p = 0.004).

Relationships between clinical characteristics and CT findings

There was no relationship between disease duration and the number of joints with either CT erosion or intraosseous tophus in the 20 patients (table 2). However, there was a strong correlation between the number of subcutaneous tophi and the number of joints affected by both erosion and intraosseous tophus. A weaker relationship was evident with serum urate concentrations taken at the time of the study visit. C-reactive protein was strongly associated with the number of eroded joints and to a lesser extent with the number of joints with intraosseous tophi. Dominant hand grip strength was strongly associated with both the number of eroded joints and the number of joints with intraosseous tophi. There was no relationship between CT findings and other clinical characteristics such as age, body mass index or serum creatinine (data not shown).

DISCUSSION

This study has demonstrated that there is a strong relationship between bone erosion and the presence of intraosseous tophus. The use of CT scanning has enabled non-invasive visualisation of tophi within areas of bone erosion. Although additional processes such as associated synovitis cannot be excluded, these results indicate that tophus infiltration into bone is the dominant mechanism for the development of bone erosion and joint damage in gout.

This study has provided an opportunity to compare erosions on plain radiography and CT scanning. Overall, there was moderately good agreement between plain radiographic and CT erosion. These results provide further validation for the use of a radiographic damage score that includes an erosion score in gout. However, some mismatch was evident and overall more erosions were detected by plain radiographs than by CT. This is in contrast to reported studies in rheumatoid arthritis, in which CT has been shown to detect more erosions than plain radiography.13 It is possible that true erosions were not adequately detected by CT due to insufficient image quality or lack of training by the observers. These explanations seem unlikely, as CT has much better image resolution than plain radiography and both readers were experienced in the assessment of both radiographic erosions in gout and CT tophus.11 15 An alternative explanation is that some changes scored as plain radiographic erosions were not in fact true erosions. The finding that mismatch occurred most often in DIPJ (which are not routinely analysed in rheumatoid arthritis scoring systems) supports this explanation. It is likely that the assessment of these joints for erosion by plain radiography is less reliable due to their small size and the presence of concurrent degenerative joint disease.16 17 Whether the DIPJ should be included in a radiographic damage index is the subject of ongoing analysis by our group.

In this study, intraosseous tophi were not consistently identified in all small erosions. However, virtually all joints with large erosions on plain radiography or CT did have visible intraosseous tophus. These observations suggest that pathology other than tophus infiltration may be causing some small erosions. However, it seems more likely that intraosseous tophus in small erosions is not easily visualised due to the limits of resolution by CT scanning.

The comparison between clinical findings and CT abnormalities has important implications for clinical practice. Our data indicate that gout disease duration alone is not a key determinant of joint damage in gout. The close relationship between the presence of subcutaneous tophi and CT abnormalities highlights the importance of subcutaneous tophi as a marker of underlying joint and bone damage in gout. The strong correlation between grip strength and CT abnormalities indicates that the functional consequences of structural damage in gout are considerable.

Although there was no difference in the density or presence of calcification between the intraosseous tophi and non-intraosseous tophi, intraosseous tophi were larger in size. This finding is perhaps surprising; given the clinical appearance of large subcutaneous tophi that are frequently present in patients with severe gout. Importantly, the CT analysis only included tophi directly adjacent to or within joints (rather than subcutaneous tophi distant from joints); this may explain the relatively small size of many of the tophi that were analysed in this study. Our data suggest that the burden of MSU crystals within the joint, rather than particular biological characteristics of the tophus, is an important determinant of the development of associated bone erosion in gout.

The underlying pathological mechanisms of bone erosion in chronic gout remain uncertain. It seems unlikely that local pressure by the intra-articular tophus is sufficient to cause such lesions. Local production of enzymes within the tophus that degrade bone and cartilage matrix may contribute to the development of erosion in gout.18^–20 Chronic foreign body granulomatous synovitis is a frequent feature of chronic gout,1 21 and some ultrasound studies have suggested that synovitis is associated with bone erosion in gout. The current study does not exclude the potential role of granulomatous synovitis induced by intra-articular MSU crystals in the pathogenesis of bone erosion in gout. It is possible that intraosseous tophi are extensions of chronic granulomatous synovitis within the joint, or that the burden of MSU crystals may influence the degree of associated synovitis. Comparisons between CT scanning, ultrasonography and pathological analysis may clarify this issue in future studies. Our observation that C-reactive protein is strongly associated with CT erosion suggests that an inflammatory response (either related to the tophus or within synovial membrane) may contribute to bone erosion in gout. We have recently identified osteoclast-like cells within subcutaneous tophi and also at the soft tissue–bone interface in patients with gout, indicating that osteoclast activation within the tophus is a potential mechanism for bone erosion in gout.22

Our findings raise further questions about the impact of intensive urate-lowering therapy on bone erosion in gout. It is well documented that effective urate-lowering therapy leads to the regression of subcutaneous and intra-articular tophi,23^–25 and some reports have also indicated that radiographic erosions can repair in patients with gout.26 Further prospective studies are needed to determine the time course for tophus formation and dissolution, whether gouty erosions reliably repair with effective urate-lowering therapy, what rate of erosion repair is achievable and what target serum urate is needed to achieve such repair. These data also provide further support for early urate-lowering therapy, in order to prevent the development of intraosseous tophi and bone erosion.

In summary, using quantitative analysis of non-invasive imaging techniques, we have demonstrated a close relationship between bone erosion and the presence and size of intraosseous tophus. Direct visualisation of these tophi within sites of erosion strongly implicates these lesions as causative in the pathogenesis of bone erosion in gout.