Does digital X-ray radiogrammetry have a role in identifying patients at increased risk for joint destruction in early rheumatoid arthritis?

Rheumatoid arthritis (RA) is a progressive autoimmune disease characterized by inflammatory activity in the joints, synovial sheets of tendons and bursae as well as extra-articular manifestations such as vasculitis and serositis [1‐3]. The increased amount of pro-inflammatory cytokines mediates osteoclast activation, which in turn provokes joint destruction [4].

Both the disease and its drug treatment can cause systemic bone loss and also a periarticular disease-related osteoporosis [5]. The periarticular bone loss in hands is an early feature of RA and may precede erosions [6, 7]. As periarticular demineralization and joint damage are both related to imbalance in osteoclast and osteoblast activity, measures of hand bone loss may be a marker for active deterioration of bone and a predictor of subsequent erosive joint damage.

Radiogrammetry was introduced in the sixties for assessment of skeletal status, using various measures of the cortical bone on conventional hand radiographs. The diaphysis of the second metacarpal of the hand was often selected for radiogrammetry. Measurements of the total and medullar width of the bone can be used to calculate different indices and to consecutively quantify changes of cortical bone [8].

The availability of digital images provides the opportunity for quantitative measurements of radio-geometric features and offers a refinement for radiogrammetry [9, 10]. Digital X-ray radiogrammetry (DXR) is a technique that uses automated image analyses of standard hand radiographs, either in digital or conventional analog format, to estimate bone mineral density (DXR-BMD) [8, 11, 12]. DXR was first introduced into clinical practice in 1999 for osteoporosis assessment. Several studies have shown that this technique has a potential to predict progressive joint damage in RA [13‐15].

The heterogeneity of the RA disease and the need for rapid adjustment of disease management puts high requirements on markers for disease progress. The currently available predictors for poor outcome are not perfect and there is a need for improved decision support. The aim of this study was to investigate the potential of hand BMD loss analyzed with DXR to predict progression of joint damage in patients with early RA followed for up to eight years.

In this study on 379 patients with early RA who were followed for eight years we have studied the potential of HBL during the first year measured by DXR to predict radiographic joint damage progression after two, five and eight years.

In accordance with previous studies, HBL above SDC was significantly associated with radiographic joint damage. This was the case also when HBL was defined by tertiles as great, moderate or no/small.

In the early stage of the disease the patients had bone loss more often than they had radiographic progression. Thus, after one year HBL was present in 68% of the patients while progressive joint damage was seen in only 25%. With time radiographic progression increased to more than 60%, which is in agreement with the report by Güler-Yüksel et al. [26], who showed bone loss after one year in 68%, progressive joint damage in 18% and after four years progressive joint damage in 30%. The explanation for this difference is that localized loss of BMD predates erosions and joint destruction [27, 28].

Change in DXR-BMD between baseline and one year correlated significantly and inversely with change in SHS at all time-points. This is in agreement with previous studies [13, 14, 29]. In an observational study, Stewart et al. showed that measurement of HBL over one year, using DXR, correlated with erosive changes in patients with early RA at four years follow-up [29] and Hoff et al. also showed that patients with hand BMD loss at one year had more radiographic damage at five and ten years in comparison with patients without HBL [30]. However, in these studies, the correlations between hand BMD and joint damage were found to be small suggesting that the predictive ability of HBL for joint damage may be limited. This notion is further supported by the modest PPVs of HBL for joint damage found in the present study.

The change in SHS from baseline to one year and the presence of anti-CCP, but not change in DXR-BMD, were the only independent predictors of radiographic progression at two years. This is in line with the BeSt study reporting that hand BMD loss as a risk factor for further radiographic damage is largely overridden by change in SHS measured over one year [26]. This implies that the first year change in DXR-BMD does not add to the therapeutic decision if radiographic scores are available. However, in clinical practice, scoring radiographs is not frequently performed, as it is a time-consuming method requiring special training.

Therefore, in the absence of radiographic scores, DXR-BMD, which today may be easily accessible in clinical practice, might be a helpful predictor. Indeed, the present data show that in the absence of radiographic scoring the change in DXR-BMD was an independent predictor of radiographic progression, significant at two years, but not at five and eight years.

In the present study, change in SHS over the first year was the single best predictor of further radiological progression. This predictor attained the highest PPVs and contributed highly to create the best regression model for predicting radiological outcome at two years, in which also the presence or absence of baseline erosions, anti-CCP and number of swollen joints after one year were independent predictors. By this model Nagelkerke R² was 77% and 91% of the patients could be correctly classified. Thus, change in SHS over the first year is indeed a promising predictor for use at this stage of the disease. However, it would be ideal if reliable predictors could be identified at an earlier stage of the disease.

Ongoing studies including measuring changes in DXR-BMD over shorter time periods, for example, three and six months, are awaited with great interest. Recently a study with six months data on hand BMD loss was published [31]. Here DXR-BMD loss was measured from baseline to six months in 80 patients with early (mean disease duration 11 month) undifferentiated arthritis. They found that great BMD loss was associated with RA development after one year. It should be noted that the cut-off value for 'great bone loss' was considerably higher than that used in the present study (≥2.5 versus 1.5 mg/cm²/month).

It has been suggested that treatment with biologics may decelerate the development of erosions more than it reduces HBL [32]. However, since the present study was performed before the introduction of biologics, this issue could not be further addressed.

Loss of bone measured by HBLsdc and HBLtertiles was more frequent in patients with high disease activity and in patients not in remission after two years but not in later phases of the disease. HBL did not come out as an independent predictor of clinical outcome in multiple regression analyses. In 145 patients from the BeSt study, where change in hand BMD by DXR was investigated, an increase in BMD occurred mostly in patients with continuous remission by the EULAR criterion [33]. However, continuously low disease activity did not appear to provide less loss in hand BMD than continuously high disease activity. Therefore, although reduction of BMD is a consequence of inflammation, DXR-BMD is not a suitable tool for predicting disease activity in the future.

A shortcoming of this study was that, at several sites, the technical conditions at acquisition of X-rays were such that BMD could not be measured for some patients which led to the loss of several patients from the study, but the included patients did not differ in baseline characteristics from those not included. Furthermore, the patients with early RA in this study fulfilled the ACR 1987 criteria, whereas patients with very early RA who presented with limited clinical symptoms were not included until they fulfilled the criteria. This means that we have no data, neither of radiographic damage nor of bone mineral density from the very early phase of RA.

To conclude, the present study confirms previous studies reporting an association of change in DXR-BMD over the first year of RA disease with progression of radiographic joint damage over an extended period of time. However, the change in DXR-BMD was not an independent predictor of radiographic joint damage progression. Instead, change in SHS from baseline to one year was a strong predictor of radiographic progression both after two years and after longer follow-up visits at five and eight years. On the other hand, when radiological findings were excluded the association was independent. This suggests a possible predictive role of DXR-BMD in clinical practice when radiographic evaluation is not available. However, since the association was significant only after the follow-up at two years and the PPVs of bone loss for radiological progression were quite small, further studies are required before this can be established. Further studies on DXR may focus on the diagnosis of early RA also considering sex- and age-related DXR data that define a threshold value to discriminate possible patients with RA. In particular, studies with a shorter interval between measurements of DXR-BMD are encouraged to evaluate its predictive power in early RA, ideally before joint damage has become identified radiographically.