Patient Profiles in Randomized Controlled Trials Versus a Real-World Study in Psoriatic Arthritis: Scoping Review and Metaanalysis

DISCUSSION

This study provides important information on the profiles of patients with PsA included in RCTs and in routine clinical practice. In this metaanalysis of 10 phase III RCTs of bDMARDs in PsA, patients had a highly active musculoskeletal disease, as reflected by a pooled mean SJC of 11.8 and TJC of 21.5, frequent enthesitis (64.7%) and dactylitis (37.7%), and active skin PsO—with a vast majority (62.2%) having a BSA ≥ 3%. In contrast, in the real-world PsABio study of 930 patients over the same time frame, although demographic data were similar, musculoskeletal disease activity was much lower (mean SJC and TJC were 5.7 and 11.9, respectively, and 48.2% and 19.8% had enthesitis and dactylitis, respectively), as was skin disease (46% had very limited PsO). On the other hand, PGA and disease impact, as assessed through PROs, were high and comparable between RCTs and the real-world data study. Conversely, CRP was statistically significantly higher in PsABio, suggesting that CRP may play a role in the decision to start a bDMARD therapy in clinical practice. These findings emphasize the need for caution when applying results from RCTs to usual care in terms of generalizability. They also support the physicians’ clinical decision when prescribing bDMARDs in usual practice, since although joint counts were lower in PsABio, other markers indicated those patients were indeed in need of a bDMARD.

This study has strengths and weaknesses. We included only trials published between 2015 and 2020, to align with the time frame of PsABio. This study did not include a formal risk-of-bias assessment for the literature review. However, such an assessment is not standard in the context of scoping reviews, which aim to map available evidence rather than provide an assessment of management options.⁸ Given that our metaanalysis is not based on individual patient data, the conclusions are applicable only at the group level. We could not, for example, separate and compare bDMARD-naïve and bDMARD-experienced groups in both settings, though such a comparison would have been of interest. Also, we did not include data on axial disease or comorbidities, as they were not uniformly described among the different studies. For real-world data, we relied only on a single data source, the PsABio study, which might restrict the scope of our analysis, as it may not capture the diversity of patient populations in usual practice. However, PsABio is one of the largest clinical practice studies in PsA, providing data in a multinational setting on more than 900 participants starting a bDMARD. Further research in a broader population and/or in other observational studies would be useful to confirm our findings.

Disease activity was high in the RCTs. Although it was much lower in PsABio than in the RCTs, it was still remarkably high for routine clinical practice, with a mean of 6 swollen and 12 tender joints. The higher disease activity identified in this meta-analysis of RCTs aligns with previous research, indicating that evidence-based knowledge of PsA treatments focuses mainly on polyarticular disease, which accounts only for a subgroup of PsA.^4,5 Patients in PsABio more commonly had an oligoarticular phenotype but were apparently as much in need of a bDMARD as those in the RCTs, as shown by the fact that 50% had already received another bDMARD before, whereas this was the case in less than 30% of RCT patients. The lower joint counts observed in PsABio patients is interesting when considering that countries may have rules for minimal disease activity when starting a bDMARD, and in PsABio (an observational study), the drugs were not provided in the context of the study, nor was there a washout period. Despite the overall lower disease activity and similar disease duration, PsABio patients may have had more refractory disease than most of the RCT patients, which is also reflected by the similar disease burden according to PROs and the relatively high PGA score and elevated CRP.

Despite differences in joint counts and presence of enthesitis/dactylitis between RCTs and the observational study, patients with PsA reported a high and similar disease impact in both settings. This may point toward a disconnect between clinical examination and patient-reported symptoms and concerns, which has been previously indicated to affect 30-70% of patients with PsA.^21,22 In the present study, widespread pain was not assessed; it is possible that differences in rates of this comorbidity may have influenced our findings. On the other hand, PGA was statistically significantly higher in RCTs than in PsABio (60 vs 54). This difference is moderate and numerically not nearly as far apart between the studied groups as the joint counts, indicating that physicians regarded the disease activity within PsABio as quite significant. Also, CRP was lower in RCTs than PsABio; thus, patient assessment is aligned in some ways with some of the markers of systemic disease activity. Both these elements in PsABio go in the direction of supporting the need for bDMARDs in real-world patients, thus confirming the physician’s decision.

Skin disease was much more prevalent and severe in RCTs than in the PsABio study. This is likely due to a requirement of minimum skin involvement in 6 of the evaluated studies, whereas this was not the case in PsABio. However, the observation of a low frequency of major skin involvement in this real-world setting lends further support to previous findings that PsO is relatively mild in clinical practice in patients with PsA.²³

Data from clinical practice are increasingly being used to supplement those from RCTs to improve our understanding of response to treatment in PsA. The selection criteria of RCTs of bDMARDs in PsA may not always allow the inclusion of all patients who would benefit from these treatments.⁴ In the PsABio registry, almost half of the cohort could not have been included in RCTs, confirming other studies in which lack of polyarticular disease was also often reported as a limiting factor for inclusion.^3,24 It is currently unclear if treatment effectiveness and drug survival are similar in patients with PsA who are eligible vs those who are ineligible for RCTs.²⁴ Of note, the present analyses did not compare the efficacy of drugs in the observational study vs the RCTs; thus, we cannot make conclusions on the relative usefulness of drug prescriptions in these 2 settings. Although a recent evolution of the patient population considered for RCTs has been observed, with inclusion of patients with earlier and less severe disease,⁵ the present findings indicate that subanalyses of patients who are more similar to the ones in clinical practice (eg, patients with oligoarticular disease) are needed, and these subanalyses should be prespecified in the design of RCTs.

In summary, our study demonstrated that patients with PsA in routine practice had mild/moderate joint disease and often limited skin PsO, leading to lower overt joint disease, whereas CRP and PGA were high. Although these patients differed significantly from patients included in RCTs, these findings are in line with the decision for bDMARD therapy in PsABio. RCT populations are not fully representative of routine care populations; future studies should address the population of patients who are ineligible for RCTs but are nonetheless in need of effective therapy.