Differences are found in patient characteristics between registries and randomized control studies [4], and the former are essential to determine the effectiveness and safety of new therapies in a broad, generalizable population. In the past decades, national and regional registries were established to evaluate anti-TNF agents in the treatment of RA [10]. However, most evaluated the earliest agents, such as IFX and etanercept, and only a few published registries included data on the newer anti-TNFs such as adalimumab [11, 12], certolizumab-pegol [13, 14] and GLM [15]. BioTRAC was one of the longest running RA registries and included data on both old (IFX) and new (GLM) anti-TNF agents.
When anti-TNFs were first approved for the treatment of RA, they were initially used in more refractory patients with longer established disease and higher disease activity. As time passed, they were used earlier, in more moderate activity patients. This can be seen if one compares the baseline characteristics of patients in the registration studies for IFX and GLM [16, 17]. Such a pattern, in which baseline disease activity decreased over time, had been reported in the interim analysis of the IFX-treated patients in BioTRAC [6]. Despite this, it was interesting to notice that baseline disease characteristics of the GLM-treated patients from 2010 to 2012 suggest that the first patients to be treated with GLM may have had more active disease than IFX-treated patients. This could be the result of an unconscious channeling bias towards using newer therapies in more severe patients, as the MDGA scores were identical between the two cohorts. Another possibility is that this was driven by the limited availability of the GLM auto-injector during that period, forcing the use of pre-filled syringes by most patients, along with uncertainties in market dynamics caused by the corporate takeover of Schering-Plough by Merck and the subsequent transition of the immunology portfolio to Janssen. Studies to evaluate the impact of disease duration, baseline disease activity and the adherence to treat-to-target guidelines on long-term function and outcomes are ongoing.
Despite difference in baseline disease activity, all three anti-TNFs showed efficacy with decreased disease activity and improved function. The route of administration does not appear to bring any specific efficacy benefit, as the data curves for GLM and GLM-IV patients are basically superimposable. Differences in the proportion of patients achieving target-specific outcomes such as LDA and remission were noted between IFX- and GLM−/GLM-IV-treated patients. Because we are reporting observed data, these differences could be driven by differences in baseline disease activity, the implementation of treat-to-target guidelines or the use of more stringent targets, such as remission rather than LDA, in later years when GLM and GLM-IV were more likely to be chosen as treatment. Also, the greater availability of additional treatment options could lead to a higher probability of switching therapies if such targets were not achieved. Therefore, caution should be exercised when interpreting the relative effectiveness of the three agents.
The incidence of AEs and SAEs was found to be similar between agents, although there were some notable differences. Patients treated with IFX had a greater incidence of chest discomfort, chest pain, fatigue, headaches, pain, pyrexia, pain in extremities and pruritus compared to GLM and GLM-IV patients, all of which could be due to acute and delayed infusion reactions [18]. Conversely, GLM and GLM-IV patients had a greater incidence of “lack of response” or “loss of response” AEs compared to IFX-treated patients, although this was likely driven by changes in the “End Of Participation” questionnaire and the addition of lack/loss of response as an AE of special interest in a protocol amendment after 2014 (see below).
The incidence of serious infections was 1.2–2.7 events/100 pt.yrs., slightly lower than the incidence of 4–4.4 events/100 pt.yrs. reported in other registries [10, 11, 19]. However, since anti-TNF therapy in RA patients was associated with an increased risk of serious infections, especially in the first 6 months of treatment [20, 21], registries with very long duration of follow-up would have a tendency to report a lower incidence rate. The low incidence of serious infection could also be explained by the low level of disease activity achieved and maintained over time. Indeed, the CORRONA registry assessed the relationship between DAS28 and infection in RA patients and found that high disease activity was associated with an increased risk of infection [22]. Analyses from the BSRBR and Italian LORHEN registries showed similar results [20, 23]. However, other European registries suggested that higher disease activity as measured by DAS28 was not directly associated with an increased incidence of serious infections [24]. Post Hoc analyses could be done in order to determine if serious infections are linked to control of disease activity, age, the use of concomitant MTX, glucocorticoids or survival bias from dropout of patients who developed an infection and subsequently stopped their anti-TNF.
The limitations of this registry are the absence of a non-biologic DMARD control group, the inclusion of predominantly bio-naïve patients and the inherent biases that are common within non-interventional, observational studies. Other limitations are related to non-inclusion of specific data sets that were not “standard of care” among community clinics in the mid-2000’s as this would have led to many missing data points. Examples of these includes radiographic imaging, the complete 66/68 joint count and baseline co-morbidities (although smoking habits were recorded since 2009). Also, the long duration of the registry could have had an impact on data quality over time due to protocol amendments, changes in standard operating procedures from the three sponsors and improvements in adverse event reporting from refining processes and increasing site experience. An example of the above was site training implemented in 2014 following the first interim analysis of the IFX cohort [6] to limit the inappropriate use of the “Other reason; provide details” box within the “End of participation” form when patients were losing response. This led to an increase in the incidence of lack/loss of response AE reporting in later years which had a larger proportion of GLM- and GLM-IV-patients.
Also, despite its respectable size, BioTRAC had limited ability to detect rare AEs unlike large national registries, such as the UK’s BSRBR, Sweden’s ARTIS, Germany’s RABBIT, Denmark’s DANBIO, Spain’s BIOBADASER and the US’s CORRONA [10]. Indeed, most Canadian multi-center registries, such as BioTRAC, CATCH [25], OBRI [26] and RHUMADATA [27], are smaller in scope but still provide significant insights on the treatment of RA at a regional level. CATCH, OBRI and RHUMADATA have the advantage over BioTRAC of being disease registries enrolling RA patients taking any therapy (biologic and non-biologic DMARDs). CATCH is an early RA disease registry enrolling newly diagnosed RA patients while OBRI and RHUMADATA enrolls RA patients from academic and community centers but are restricted to the provinces of Ontario and Quebec, respectively [26, 27]. Despite those differences in design, it has been possible to increase power and answer specific scientific questions by combining patient data from multiple registries [28].
One key strength of BioTRAC is that it included an extensive evaluation of clinical disease parameters, most of which were not collected elsewhere, especially in the early years [10]. Due to its long-term duration, BioTRAC offered a unique opportunity to evaluate the real-world effectiveness and safety of three anti-TNF agents in a community Canadian setting, while assessing regional variations due to differences in patient profiles, practice patterns and local reimbursement policies impacting access to care over 16 years. Although there has been extensive real-world evidence generated on the early anti-TNF agents such as IFX or etanercept, very little efficacy data has been published with other anti-TNF agents such as GLM, and most of those only presented persistence data [15, 29,30,31]. One exception, however, is the GO NICE prospective non-interventional trial in Germany for inflammatory arthritis patients treated with GLM [15, 32]. This 2-year trial also found significant clinical effectiveness among RA patients [15], as well as improvements in patient-reported health status, physical function, and fatigue levels [32].