Early response to therapy predicts 6-month and 1-year disease activity outcomes in psoriatic arthritis patients

Early response to therapy predicts 6-month and 1-year disease activity outcomes in psoriatic... Abstract Objectives In PsA management, remission and low disease activity represent preferential treatment targets. We aimed at evaluating the predictive value and clinical use of initial therapeutic response for subsequent achievement of these targets. Methods Based on data of 216 patients enrolled in a randomized controlled trial of golimumab (GO-REVEAL), we performed diagnostic testing analyses using 3- and 6-month disease activity as tests for treatment outcomes to understand the implications of early response. In regression analyses, we estimated the probabilities for achieving at least LDA. Disease activity was measured by the disease activity index for PsA (DAPSA). Results Three-month DAPSA levels were excellent tests for disease activity at 6 months (and at 1 year), with areas under the receiver operating characteristic curves of 0.92 (and 0.88, respectively). The estimated probability for 6-month LDA could be quantified as <22% if patients did not reach at least moderate disease activity after 3 months on golimumab. Similar data were seen for early DAPSA response: patients achieving a DAPSA 85% at 3 months had an 84% probability for 6-month LDA or REM. All results were validated in an independent trial cohort of patients treated with infliximab (IMPACT 2). Conclusion Three months after implementation of therapy in PsA, it is already possible to evaluate the potential for accomplishing therapeutic goals. This substantiates the choice of the 3-month assessment as essential for treatment adaptations. psoriatic arthritis, disease activity index, outcomes research Rheumatology key messages In PsA, success of a new therapy is already predictable after 3 months. Treatment adaptations in PsA should already be considered at early assessments to optimize long-term outcomes. Using the disease activity index for PsA score enables a treat-to-target approach in PsA patients. Introduction Current guidelines for the management of PsA propose remission (REM) or at least low disease activity (LDA) as a treatment target [1, 2]. Indeed, consistent suppression of inflammation improves clinical outcomes in randomized comparisons [3] and, in particular, the effect of cumulative disease activity appears to influence not only joint damage [4] but also the cardiovascular outcome of PsA patients [5, 6]. Consequently, it seems important to assess the potential of a newly administered therapy early after its introduction to allow timely decision-making on its efficacy. For this purpose, the disease activity index for PsA (DAPSA) serves as a disease-specific compound measure that sums five characteristic disease activity variables [7–10]. Both the whole DAPSA score and its purely clinical version (cDAPSA), which omits CRP, have been validated for the use in PsA [11], and for both indices response criteria and cut-off values to define disease activity states have been developed [12]. Importantly, the DAPSA-defined REM and LDA states have recently been shown to be valid against long-term functional and radiographic outcomes [13]. The availability of parameters to predict response in PsA therapy is very limited; there is evidence from observational data that CRP [14] and ESR [14, 15] as well as male sex [15] predict achievement of minimal disease activity in patients treated with TNF-α inhibitors (TNFi). However, there is a lack of information specifically about the right point in time for the evaluation of core set variables and compound scores to make important treatment decisions and pursue long-term therapeutic targets for PsA patients. Following a treat-to-target approach, the crucial question is what a specific state or response at 3 months means with regard to the chance of achieving the recommended treatment target after 6 months. The implications for decision-making would be to change therapy at 3 months in patients who did not show sufficient improvement (or a sufficiently low state), whereas in those patients who are showing sufficient response, therapy might be continued. The aim of the present study was to analyse the predictive value of disease activity levels and, in particular, changes of disease activity (measured by the continuous DAPSA and cDAPSA) early after the start of therapy with respect to longer-term treatment success to inform clinical decision-making. Methods Patients For our main analyses, we were provided with a random 90% dataset of patient-level data from the GO-REVEAL trial [16]. This large randomized controlled clinical trial evaluated the effectiveness of the TNFi golimumab (GOL) in PsA patients with prior non-response to conventional synthetic DMARDs (csDMARDs) or NSAIDs. Patients with three or more swollen and tender joints were randomly assigned to either 50 or 100 mg GOL, or placebo. In our analyses, we used data from the active treatment groups, merging the two GOL arms, and included all patients in whom DAPSA at baseline (BL), 3, 6 and 12 months was available. Likewise, we received a random 80% dataset of the IMPACT 2 [17] trial, an randomized controlled clinical trial that investigated the effectiveness of another TNFi, infliximab, in PsA patients to validate the findings from GO-REVEAL patients externally. In IMPACT 2, patients were required to have five or more swollen and tender joints and either CRP values of ⩾15 mg/l, ESR ⩾28 mm/1 h and/or morning stiffness of ⩾45 min to receive either infliximab or placebo [17]. Both clinical trials were approved by respective local ethics committees; no additional ethical approval was required for the present study. Disease activity assessment For all analyses, we used the DAPSA, a score that is based on the summation of five variables: number of tender 68 and swollen 66 joints, patient global assessment and patient pain assessment on a 10 cm visual analog scale, as well as CRP (in milligrams per decilitre). The DAPSA score was originally developed for ReA [18] and has recently been validated for the use in PsA [11]. Analyses DAPSA values as tests for subsequent DASPA states We used a diagnostic testing approach to assess the areas under the receiver operating characteristic (ROC) curves of DAPSA values at baseline, 3 and 6 months and of early DAPSA changes regarding the achievement of a desired outcome, that is, REM or LDA after 6 months and after 1 year. Unless otherwise indicated, from now on, LDA also includes REM. An area under the ROC curve of one represents a perfect test, and an area of 0.5 a test no better than random prediction, with one misclassification for each accurate classification. Estimating probabilities of outcomes For guidance in clinical practice and to derive a more comprehensible everyday tool, we perfomed logistic regression analyses to model the probability of achieving at least LDA (i.e. a DAPSA ⩽14) at 6 months depending on DAPSA levels at baseline, and after 3 months, and depending on change from baseline to 3 months in DAPSA scores. We also estimated the probability for LDA at the later 1-year assessment using baseline, 3- and 6-month DAPSA levels as well as the 3- and 6-month percentage response. Our rationale for these analyses was that prediction using 6-month data is specifically relevant in the context of the treat-to-target approach, and prediction of 1-year disease activity is important for a long-term therapeutic perspective. The interpretation of CIs regarding statistical significance needs to be drawn with caution, as there is no formal adjustment for multiple testing. For formal statistical testing without repeated testing in one model, we provide the more complex longitudinal analysis (see below Longitudinal model and in the supplementary data, available at Rheumatology online). Tracking back of disease activity in remitters and non-remitters We then identified disease activity states of DAPSA at 6 months and 1 year (representing the long-term outcome of therapy). To this end, we applied the established DAPSA cut points of ⩽4 for REM, >4 and ⩽14 for LDA, >14 and ⩽28 for moderate disease activity (MDA), and >28 for high disease activity (HDA) [12]. We stratified patients according to their 6-month disease activity outcome (REM, LDA, MDA or HDA) and descriptively tracked their disease activity levels back to the respective baseline values. We repeated this exercise according to the 1-year states. Longitudinal model To supplement our main analyses, we developed a longitudinal statistical model to predict DAPSA states after treatment in PsA. We used a generalized estimation equation with independent covariance structure. For the sake of conciseness and clarity, we present the methodology and results of the longitudinal model, which were fully confirmative of our main analyses, in the Methods section of the supplementary data, available at Rheumatology online. cDAPSA In addition to the analyses focusing on the full DAPSA score, we evaluated the early predictive value of the purely clinical cDAPSA. In this case, we used disease activity cut points of ⩽4 for REM, >4 and ⩽13 for LDA, >13 and ⩽27 for MDA and >27 for HDA, as previously described [12]. All cDAPSA results are presented in the Results section of the supplementary data, available at Rheumatology online. We used SAS statistical software (Version 9.4) for all our analyses. Results Patient characteristics and treatment outcomes Including all patients who had DAPSA available at BL and 3, 6 and 12 months, we analysed a total of 216 patients from the GOL arm of the GO-REVEAL trial. Baseline characteristics are presented in Table 1; briefly, 40.7% were female, with a mean (s.d.) disease duration of 7.7 (7.5) years. Mean DAPSA at baseline amounted to 48.0 (26.7) and, according to the DAPSA classification [12], 0, 2.3, 20.8 and 76.9% of patients were in REM, LDA, MDA and HDA, respectively (supplementary Table S1, available at Rheumatology online). For external validation, we included 64 patients from IMPACT 2 and found very similar results (Table 1; Supplementary Table S1, available at Rheumatology online). Table 1 Patient baseline characteristics by trial and in total study population Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Columns specify patients included in the active treatment arms of GO-REVEAL and IMPACT 2 trials and pooled data of treatment groups in both trials. Numbers specify the mean (s.d.). NA: not applicable; VAS: visual analog scale. Table 1 Patient baseline characteristics by trial and in total study population Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Columns specify patients included in the active treatment arms of GO-REVEAL and IMPACT 2 trials and pooled data of treatment groups in both trials. Numbers specify the mean (s.d.). NA: not applicable; VAS: visual analog scale. DAPSA levels early after treatment initiation are excellent tests for the achievement of good outcomes at later assessments Disease activity levels at different time points during follow-up showed excellent properties as a test for the subsequent achievement of good outcomes. Three-month disease activity levels were very strong predictors of LDA at 6 months, with an area under the ROC curve (AUC; and 95% CIs) of 0.92 (0.89, 0.96; Fig. 1A). Also, early relative change between BL and 3 months was a good predictor for an LDA status at 6 months (AUC 0.84; 95% CI: 0.79, 0.89; Fig. 1C). For prediction of disease activity at 1 year, AUC (95% CI) ranged between 0.73 (0.66, 0.79) for baseline DAPSA and 0.88 (0.84, 0.93) for 3-month values and 0.93 (0.89, 0.96) for 6-month values (Fig. 1B). Taking the percentage change of DAPSA from baseline to 3 and 6 months as a test for LDA after 1 year resulted in AUCs of 0.82 (0.76, 0.88) and 0.88 (0.83, 0.92), respectively (Fig. 1D). Validation analyses using IMPACT 2 data confirmed these results (supplementary Fig. S1, available at Rheumatology online). Fig. 1 View largeDownload slide Diagnostic testing: 6-month and 1-year achievement of low disease activity, based on earlier disease activity index for PsA levels or disease activity index for PsA percentage response Receiver operating characteristic (ROC) curves for achieving at least LDA after 6 months (A and C) or 1 year (B and D) based on DAPSA levels at baseline, 3 and 6 months (A and B) or DAPSA relative change from baseline to 3 or 6 months (C and D) as diagnostic tests. Areas under the ROC curves (AUC) including 95% CIs are presented in panel insets. Data: GO-REVEAL (n = 216). AUC: area under the curve; DAPSA: disease activity index for PsA; LDA: low disease activity; ROC: receiver operating characteristic. Fig. 1 View largeDownload slide Diagnostic testing: 6-month and 1-year achievement of low disease activity, based on earlier disease activity index for PsA levels or disease activity index for PsA percentage response Receiver operating characteristic (ROC) curves for achieving at least LDA after 6 months (A and C) or 1 year (B and D) based on DAPSA levels at baseline, 3 and 6 months (A and B) or DAPSA relative change from baseline to 3 or 6 months (C and D) as diagnostic tests. Areas under the ROC curves (AUC) including 95% CIs are presented in panel insets. Data: GO-REVEAL (n = 216). AUC: area under the curve; DAPSA: disease activity index for PsA; LDA: low disease activity; ROC: receiver operating characteristic. Early treatment response is indicative of achieving good outcomes at later time points Using logistic regression analysis, we estimated the chances for a beneficial outcome according to early relative DAPSA changes and disease activity levels at 3 months. Figure 2 illustrates that, to predict the achievement of the 6-month target of LDA reliably, substantial changes of DAPSA need to be seen at 3 months (Fig. 2A and C), and, likewise, considerable changes at the 3-month assessment are needed in order to achieve a reasonably high probability of a good outcome at 1 year (Fig. 2B and D). To achieve a sufficiently high probability of reaching a good outcome at 1 year, about the same changes in DAPSA are required at 3 and at 6 months (Fig. 2D). In other words, the predictive capacity at 3 months is not worse than at 6 months. Fig. 2 View largeDownload slide Probabilities of achieving low disease activity at 6 months or 1 year Predicted probabilities are shown for achieving at least LDA at 6 months (A and C) or 1 year (B and D) depending on DAPSA values at baseline, 3 and 6 months (A and B) or depending on DAPSA relative change from baseline to 3 or to 6 months (C and D). Data of patients included in the GO-REVEAL treatment arm (n = 216). DAPSA: disease activity index for PsA; LDA: low disease activity. Fig. 2 View largeDownload slide Probabilities of achieving low disease activity at 6 months or 1 year Predicted probabilities are shown for achieving at least LDA at 6 months (A and C) or 1 year (B and D) depending on DAPSA values at baseline, 3 and 6 months (A and B) or depending on DAPSA relative change from baseline to 3 or to 6 months (C and D). Data of patients included in the GO-REVEAL treatment arm (n = 216). DAPSA: disease activity index for PsA; LDA: low disease activity. Looking at absolute DAPSA levels among patients who achieved moderate disease activity (DAPSA >14 and ⩽28) after 3 months of therapy, the estimated probability of achieving at least LDA at 6 months was up to 69.8%, and patients with LDA at 3 months had an estimated probability for staying in LDA or reaching REM at 6 months of 70–90% (Fig. 2A). Achievement of a 50 or 75% response in DAPSA (DAPSA50, DAPSA75) at 3 months implied at least 50 or 73% probability of LDA or REM, respectively, and DAPSA85 even a probability of >80% for a good outcome at 6 months (Fig. 2C). Achieving DAPSA ⩽14 (LDA) after 3 months led to an estimated probability of ⩾78% for continuing to be in LDA (or REM) also after 1 year (Fig. 2B), and a DAPSA75 response from baseline to 3 months predicted >80% probability for at least LDA after 1 year (Fig. 2D). Validation analyses using IMPACT 2 data confirmed these results (supplementary Fig. S2, available at Rheumatology online). Tracking back patients who achieved different states at 6 months and 1 year: early differences in DAPSA levels Mean disease activity improved on TNFi very well over time, leading to a total of 39.9 and 47.5% of patients in at least LDA at 3 and 6 months, respectively (supplementary Table S1, available at Rheumatology online). At the end of follow-up after 1 year, 30.1% of patients were in REM, 28.7% in LDA, 22.2% in MDA and 19.0% in HDA (Table S1, available at Rheumatology online). Figure 3 shows that patients who achieve different DAPSA states at 6 months clearly differ at the group level already at the 3-month assessment (P ⩽ 0.0001). In addition, there were differences in baseline values at the group level, which emphasize that the starting level of disease activity is also an important determinant of the ability to reach a good outcome. Again, validation in the IMPACT 2 dataset was confirmatory (supplementary Fig. S3, available at Rheumatology online). Fig. 3 View largeDownload slide Tracking back patients with 6-month or 1-year remission, low, moderate and high disease activity Lines represent disease activity levels (means and s.e.m.) at baseline, 3 and 6 months and at 1 year in patients who achieved remission, low, moderate or high disease activity at the 6-month (A and B) or the 1-year (C and D) assessment. Disease activity states according to DAPSA (A and C) and cDAPSA (B and D). Data: GO-REVEAL (n = 216). BL: baseline; cDAPSA: clinical DAPSA; DAPSA: disease activity index for PsA; HDA: high disease activity; LDA: low disease activity; M: months; MDA: moderate disease activity; REM: remission; Y: year. Fig. 3 View largeDownload slide Tracking back patients with 6-month or 1-year remission, low, moderate and high disease activity Lines represent disease activity levels (means and s.e.m.) at baseline, 3 and 6 months and at 1 year in patients who achieved remission, low, moderate or high disease activity at the 6-month (A and B) or the 1-year (C and D) assessment. Disease activity states according to DAPSA (A and C) and cDAPSA (B and D). Data: GO-REVEAL (n = 216). BL: baseline; cDAPSA: clinical DAPSA; DAPSA: disease activity index for PsA; HDA: high disease activity; LDA: low disease activity; M: months; MDA: moderate disease activity; REM: remission; Y: year. Longitudinal analysis of DAPSA in relationship to treatment targets All data on longitudinal analyses using different models and statistical approaches are shown in supplementary Tables S2 and S3 and Figures S4–S7, available at Rheumatology online. Importantly, all these analyses using more complex statistical approaches fully confirmed the predictive capacity of DAPSA at 3 months for subsequent outcomes. Validation analyses The analyses presented above used data of the GO-REVEAL trial. External validation using the IMPACT 2 dataset led to consistent results (supplementary Figs S1–S3 and supplementary S7, available at Rheumatology online). In addition, internal validation of the longitudinal model using two different resampling methods in the GO-REVEAL dataset confirmed the presented model results (see supplementary Table S2, available at Rheumatology online, for estimates and CIs, and supplementary Figs S5 and S6, available at Rheumatology online). Finally, we supplemented our ROC results with one analysis using the outcome ACR70 response after 6 months and 1 year as the dependent variable, in contrast to the DAPSA-based analyses in the rest of the manuscript. Here, DAPSA levels at 3 and 6 months, but not baseline DAPSA, could predict 6-month and 1-year ACR70 response (see supplementary Fig. S8, available at Rheumatology online). Analyses of cDAPSA We performed all described analyses also for the cDAPSA, the clinical version of the full DAPSA, which does not include CRP. We found remarkably similar results to the data presented for DAPSA. Detailed results of the cDAPSA analyses are provided in supplementary Table S1 and supplementary Figs S9 and S10, available at Rheumatology online. Discussion In PsA management, prompt evaluation of newly implemented medication is important to avoid losing time on therapies that may not have the potential to reduce disease activity sufficiently in the individual patient. Given the fact that an increasing number of highly efficacious treatment options are available, any state other than LDA or REM should induce a change in therapy, because delay will have consequential effects of the cumulative inflammatory burden [3, 5, 6]. In this respect, a therapeutic evaluation at an early point in time should be attempted. Although response rates are increasing throughout the course of the whole first year of therapy, as evidenced by clinical trial data, the question arises about the minimal observation time that is required to estimate reliably the longer-term disease activity outcomes of the individual patient. According to the treat-to-target algorithm [1], this time frame has been proposed to be 3 months for improvement and 6 months for attaining the target. Indeed, in the present study the time between the 3- and 6-month assessments brought only limited additional information for the 1-year disease activity state. Moreover, when we simply tracked back disease activity levels of patients who had reached REM, LDA, MDA or HDA, it was apparent that disease activity at 3 months was already highly predictive of a good or poor outcome at 6 months and at 1 year; similar to what has been observed in RA [19]. In observational studies, factors that predict minimal disease activity in PsA have been explored, and there was evidence for male sex as well as CRP and ESR [14, 15] predicting favourable outcomes after TNFi treatment. However, minimal disease activity is a state, whereas the novelty of our analyses lies in the use of a continuous scale for PsA that can be used for predicting achievement of a disease state before a particular state is reached and in the identification of specific time points that serve in the process of clinical decision-making early after initiation of treatment. This enables the clinician to ascertain the likelihood of long-term therapeutic success. We show that the potential of a new therapy to suppress disease activity sufficiently at 6 months or 1 year can be predicted well at the 3-month time point, depending on whether or not patients prove to be responding at this early assessment. Consequently, treatment adaptions should be considered on this occasion. For our study, we used the DAPSA to define disease activity and categorize disease activity states. This score has previously been validated as a measure of PsA disease activity and is associated with functional and structural outcomes [8, 11, 20]. As such, it is a continuous outcome measure not only for the purpose of clinical management of PsA patients, but also serves outcomes research in the field of PsA, similar to the way that continuous scores have facilitated RA research over the past decade. Similar to previous findings in RA [19, 21], our results indicate that treatment success in PsA is highly predictable early after start of therapy by evaluating disease activity and, in particular, the change in disease activity after the start of therapy. Importantly, and in accordance with the treat-to-target recommendations, the presence of major DAPSA responses (DAPSA75, DAPSA85) is informative about whether a patient requires treatment modifications, or whether a therapy should be continued for at least an additional 3 months. Specifically, our prediction models indicate that relatively high response rates need to be observed at the 3-month assessment, in order to justify confidence in reaching the target state upon treatment continuation for another 3 months. These findings substantiate the call for an early appraisal of any new therapy, as has been formulated in therapeutic guidelines [1], to allow timely reaction with therapeutic adaptations in those who do not meet the desired level of improvement. Limitations to our analyses include the lack of follow-up data overseeing an even longer time span than 1 year. This was inherent to the use of data from clinical trials, which in our case had only 1 year of treatment with a single compound. Nevertheless, 3- and 6-month levels predicted 1-year outcomes in a similar manner, suggesting that a longer observation period would not necessarily provide additional clinically relevant information. The prediction model might differ slightly according to the mode of action of the drug used, which in our case were two TNFi. The DAPSA lacks enthesitis and dactylitis assessment, as well as skin involvement, for which separate validated instruments exist. The controversy regarding how to construct scores for the assessment of PsA has been discussed broadly at the recent international meeting that validated the treat-to-target approach for spondyloathritis, including PsA; indeed, DAPSA was one of the endorsed tools for follow-up of patients with PsA [1]. In line, we recently showed that patients who achieved DAPSA REM on a TNFi also had minimal enthesitis and dactylitis as well as Psoriasis Area and Severity Index scores, which did not differ from the very LDA state; likewise, DAPSA LDA conveyed similar non-articular outcomes to MDA [22]. Given the available core set variables in trial data, we show that achieving at least LDA in PsA can be predicted based on the parameters included in the DAPSA. The strength of the present study lies in the availability of patient-level data from two large randomized controlled clinical trials that investigated two of the most frequently used biologics in PsA therapy. Outcome measures in clinical trials are consistently evaluated by trained assessors, which is highly valuable for studies like the present one. We found very similar results in the two different trials, inherently validating the data presented. Finally, we consider the consistency of different methods addressing the same question as highly confirmatory. These methods included a diagnostic testing approach, which integrates sensitivity and specificity into the investigated association, and more advanced regression modelling, which is useful for clinicians estimating risks and benefits for their patients. In summary, our study demonstrates that the initial response to PsA therapy holds a highly significant predictive value and predicts long-term outcomes as early as 3 months after initiation of treatment. Thus, our results further support treatment algorithms in their demand for timely adaptations and provide guidance for rheumatologists who apply such strategies, as to whether and when adaptations are justified and, even more importantly, highly required. Our study provides some numerical guidance for the treating rheumatologists to quantify the probability of reaching the therapeutic target in their patients after specific disease activity states and amounts of change from baseline. Funding: No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this manuscript. Disclosure statement: J.S.S. has received grants for his institution from Abbvie, Astra-Zeneca, Janssen, Lilly, Merck Sharp & Dohme (MSD), Pfizer, Roche and has provided expert advice to and/or had speaking engagements for Abbvie, Amgen, Astra-Zeneca, Astro, Bristol-Myers Squibb (BMS), Celgene, Celltrion, Chugai, Gilead, Glaxo, ILTOO, Janssen, Lilly, Medimmune, MSD, Novartis-Sandoz, Pfizer, Roche, Samsung, Sanofi, Union Chimique Belge (UCB). D.B. is an employee of Janssen Research and Development. D.A. has received grants from MSD and BMS, speaker honoraria from AbbVie, Merck, UCB, Janssen, BMS, Pfizer, Medac and Roche and has consulted for Abbvie, Eli Lilly & Co., MSD, Centocor, Janssen. The other authors have declared no conflicts of interest. Supplementary data Supplementary data are available at Rheumatology online. References 1 Smolen JS , Schöls M , Braun J et al. Treating axial spondyloarthritis and peripheral spondyloarthritis, especially psoriatic arthritis, to target: 2017 update of recommendations by an international task force . Ann Rheum Dis 2018 ; 77 : 3 – 17 . Google Scholar CrossRef Search ADS PubMed 2 Gossec L , Smolen JS , Ramiro S et al. European League Against Rheumatism (EULAR) recommendations for the management of psoriatic arthritis with pharmacological therapies: 2015 update . Ann Rheum Dis 2016 ; 75 : 499 – 510 . Google Scholar CrossRef Search ADS PubMed 3 Coates LC , Moverley AR , McParland L et al. Effect of tight control of inflammation in early psoriatic arthritis (TICOPA): a UK multicentre, open-label, randomised controlled trial . Lancet 2015 ; 386 : 2489 – 98 . Google Scholar CrossRef Search ADS PubMed 4 Cresswell L , Chandran V , Farewell VT et al. Inflammation in an individual joint predicts damage to that joint in psoriatic arthritis . Ann Rheum Dis 2011 ; 70 : 305 – 8 . Google Scholar CrossRef Search ADS PubMed 5 Eder L , Thavaneswaran A , Chandran V , Cook R , Gladman DD. Increased burden of inflammation over time is associated with the extent of atherosclerotic plaques in patients with psoriatic arthritis . Ann Rheum Dis 2015 ; 74 : 1830 – 5 . Google Scholar CrossRef Search ADS PubMed 6 Shen J , Shang Q , Li EK et al. Cumulative inflammatory burden is independently associated with increased arterial stiffness in patients with psoriatic arthritis: a prospective study . Arthritis Res Ther 2015 ; 17 : 75 . Google Scholar CrossRef Search ADS PubMed 7 Gladman DD , Mease PJ , Healy P et al. Outcome measures in psoriatic arthritis . J Rheumatol 2007 ; 34 : 1159 – 66 . Google Scholar PubMed 8 Gladman DD , Mease PJ , Strand V et al. Consensus on a core set of domains for psoriatic arthritis . J Rheumatol 2007 ; 34 : 1167 – 70 . Google Scholar PubMed 9 Nell-Duxneuner VP , Stamm TA , Machold KP et al. Evaluation of the appropriateness of composite disease activity measures for assessment of psoriatic arthritis . Ann Rheum Dis 2010 ; 69 : 546 – 9 . Google Scholar CrossRef Search ADS PubMed 10 Coates LC , Fitzgerald O , Mease PJ et al. Development of a disease activity and responder index for psoriatic arthritis–report of the Psoriatic Arthritis Module at OMERACT 11 . J Rheumatol 2014 ; 41 : 782 – 91 . Google Scholar CrossRef Search ADS PubMed 11 Schoels M , Aletaha D , Funovits J et al. Application of the DAREA/DAPSA score for assessment of disease activity in psoriatic arthritis . Ann Rheum Dis 2010 ; 69 : 1441 – 7 . Google Scholar CrossRef Search ADS PubMed 12 Schoels MM , Aletaha D , Alasti F , Smolen JS. Disease activity in psoriatic arthritis (PsA): defining remission and treatment success using the DAPSA score . Ann Rheum Dis 2016 ; 75 : 811 – 8 . Google Scholar CrossRef Search ADS PubMed 13 Aletaha D , Alasti F , Smolen JS. Disease activity states of the DAPSA, a psoriatic arthritis specific instrument, are valid against functional status and structural progression . Ann Rheum Dis 2017 ; 76 : 418 – 21 . Google Scholar CrossRef Search ADS PubMed 14 Perrotta FM , Marchesoni A , Lubrano E. Minimal disease activity and remission in psoriatic arthritis patients treated with Anti-TNF-α drugs . J Rheumatol 2016 ; 43 : 350 – 5 . Google Scholar CrossRef Search ADS PubMed 15 Haddad A , Thavaneswaran A , Ruiz-Arruza I et al. Minimal disease activity and anti-tumor necrosis factor therapy in psoriatic arthritis . Arthritis Care Res 2015 ; 67 : 842 – 7 . Google Scholar CrossRef Search ADS 16 Kavanaugh A , McInnes I , Mease P et al. Golimumab, a new human tumor necrosis factor α antibody, administered every four weeks as a subcutaneous injection in psoriatic arthritis: twenty-four-week efficacy and safety results of a randomized, placebo-controlled study . Arthritis Rheum 2009 ; 60 : 976 – 86 . Google Scholar CrossRef Search ADS PubMed 17 Antoni C , Krueger GG , de Vlam K et al. Infliximab improves signs and symptoms of psoriatic arthritis: results of the IMPACT 2 trial . Ann Rheum Dis 2005 ; 64 : 1150 – 7 . Google Scholar CrossRef Search ADS PubMed 18 Eberl G , Studnicka-Benke A , Hitzelhammer H , Gschnait F , Smolen JS. Development of a disease activity index for the assessment of reactive arthritis (DAREA) . Rheumatology 2000 ; 39 : 148 – 55 . Google Scholar CrossRef Search ADS PubMed 19 Aletaha D , Funovits J , Keystone EC , Smolen JS. Disease activity early in the course of treatment predicts response to therapy after one year in rheumatoid arthritis patients . Arthritis Rheum 2007 ; 56 : 3226 – 35 . Google Scholar CrossRef Search ADS PubMed 20 Husic R , Gretler J , Felber A et al. Disparity between ultrasound and clinical findings in psoriatic arthritis . Ann Rheum Dis 2014 ; 73 : 1529 – 36 . Google Scholar CrossRef Search ADS PubMed 21 Aletaha D , Alasti F , Smolen JS. Optimisation of a treat-to-target approach in rheumatoid arthritis: strategies for the 3-month time point . Ann Rheum Dis 2016 ; 75 : 1479 – 85 . Google Scholar CrossRef Search ADS PubMed 22 Smolen JS , Aletaha D , Gladman DD et al. FRI0498 (2017) Outcomes associated with achievement of various treatment targets in patients with psoriatic arthritis receiving adalimumab . Ann Rheum Dis 2017 ; 76 (Suppl 2) : 677 . © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Rheumatology Oxford University Press

Early response to therapy predicts 6-month and 1-year disease activity outcomes in psoriatic arthritis patients

Loading next page...
 
/lp/ou_press/early-response-to-therapy-predicts-6-month-and-1-year-disease-activity-G0WIST3uXv
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
1462-0324
eISSN
1462-0332
D.O.I.
10.1093/rheumatology/key004
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives In PsA management, remission and low disease activity represent preferential treatment targets. We aimed at evaluating the predictive value and clinical use of initial therapeutic response for subsequent achievement of these targets. Methods Based on data of 216 patients enrolled in a randomized controlled trial of golimumab (GO-REVEAL), we performed diagnostic testing analyses using 3- and 6-month disease activity as tests for treatment outcomes to understand the implications of early response. In regression analyses, we estimated the probabilities for achieving at least LDA. Disease activity was measured by the disease activity index for PsA (DAPSA). Results Three-month DAPSA levels were excellent tests for disease activity at 6 months (and at 1 year), with areas under the receiver operating characteristic curves of 0.92 (and 0.88, respectively). The estimated probability for 6-month LDA could be quantified as <22% if patients did not reach at least moderate disease activity after 3 months on golimumab. Similar data were seen for early DAPSA response: patients achieving a DAPSA 85% at 3 months had an 84% probability for 6-month LDA or REM. All results were validated in an independent trial cohort of patients treated with infliximab (IMPACT 2). Conclusion Three months after implementation of therapy in PsA, it is already possible to evaluate the potential for accomplishing therapeutic goals. This substantiates the choice of the 3-month assessment as essential for treatment adaptations. psoriatic arthritis, disease activity index, outcomes research Rheumatology key messages In PsA, success of a new therapy is already predictable after 3 months. Treatment adaptations in PsA should already be considered at early assessments to optimize long-term outcomes. Using the disease activity index for PsA score enables a treat-to-target approach in PsA patients. Introduction Current guidelines for the management of PsA propose remission (REM) or at least low disease activity (LDA) as a treatment target [1, 2]. Indeed, consistent suppression of inflammation improves clinical outcomes in randomized comparisons [3] and, in particular, the effect of cumulative disease activity appears to influence not only joint damage [4] but also the cardiovascular outcome of PsA patients [5, 6]. Consequently, it seems important to assess the potential of a newly administered therapy early after its introduction to allow timely decision-making on its efficacy. For this purpose, the disease activity index for PsA (DAPSA) serves as a disease-specific compound measure that sums five characteristic disease activity variables [7–10]. Both the whole DAPSA score and its purely clinical version (cDAPSA), which omits CRP, have been validated for the use in PsA [11], and for both indices response criteria and cut-off values to define disease activity states have been developed [12]. Importantly, the DAPSA-defined REM and LDA states have recently been shown to be valid against long-term functional and radiographic outcomes [13]. The availability of parameters to predict response in PsA therapy is very limited; there is evidence from observational data that CRP [14] and ESR [14, 15] as well as male sex [15] predict achievement of minimal disease activity in patients treated with TNF-α inhibitors (TNFi). However, there is a lack of information specifically about the right point in time for the evaluation of core set variables and compound scores to make important treatment decisions and pursue long-term therapeutic targets for PsA patients. Following a treat-to-target approach, the crucial question is what a specific state or response at 3 months means with regard to the chance of achieving the recommended treatment target after 6 months. The implications for decision-making would be to change therapy at 3 months in patients who did not show sufficient improvement (or a sufficiently low state), whereas in those patients who are showing sufficient response, therapy might be continued. The aim of the present study was to analyse the predictive value of disease activity levels and, in particular, changes of disease activity (measured by the continuous DAPSA and cDAPSA) early after the start of therapy with respect to longer-term treatment success to inform clinical decision-making. Methods Patients For our main analyses, we were provided with a random 90% dataset of patient-level data from the GO-REVEAL trial [16]. This large randomized controlled clinical trial evaluated the effectiveness of the TNFi golimumab (GOL) in PsA patients with prior non-response to conventional synthetic DMARDs (csDMARDs) or NSAIDs. Patients with three or more swollen and tender joints were randomly assigned to either 50 or 100 mg GOL, or placebo. In our analyses, we used data from the active treatment groups, merging the two GOL arms, and included all patients in whom DAPSA at baseline (BL), 3, 6 and 12 months was available. Likewise, we received a random 80% dataset of the IMPACT 2 [17] trial, an randomized controlled clinical trial that investigated the effectiveness of another TNFi, infliximab, in PsA patients to validate the findings from GO-REVEAL patients externally. In IMPACT 2, patients were required to have five or more swollen and tender joints and either CRP values of ⩾15 mg/l, ESR ⩾28 mm/1 h and/or morning stiffness of ⩾45 min to receive either infliximab or placebo [17]. Both clinical trials were approved by respective local ethics committees; no additional ethical approval was required for the present study. Disease activity assessment For all analyses, we used the DAPSA, a score that is based on the summation of five variables: number of tender 68 and swollen 66 joints, patient global assessment and patient pain assessment on a 10 cm visual analog scale, as well as CRP (in milligrams per decilitre). The DAPSA score was originally developed for ReA [18] and has recently been validated for the use in PsA [11]. Analyses DAPSA values as tests for subsequent DASPA states We used a diagnostic testing approach to assess the areas under the receiver operating characteristic (ROC) curves of DAPSA values at baseline, 3 and 6 months and of early DAPSA changes regarding the achievement of a desired outcome, that is, REM or LDA after 6 months and after 1 year. Unless otherwise indicated, from now on, LDA also includes REM. An area under the ROC curve of one represents a perfect test, and an area of 0.5 a test no better than random prediction, with one misclassification for each accurate classification. Estimating probabilities of outcomes For guidance in clinical practice and to derive a more comprehensible everyday tool, we perfomed logistic regression analyses to model the probability of achieving at least LDA (i.e. a DAPSA ⩽14) at 6 months depending on DAPSA levels at baseline, and after 3 months, and depending on change from baseline to 3 months in DAPSA scores. We also estimated the probability for LDA at the later 1-year assessment using baseline, 3- and 6-month DAPSA levels as well as the 3- and 6-month percentage response. Our rationale for these analyses was that prediction using 6-month data is specifically relevant in the context of the treat-to-target approach, and prediction of 1-year disease activity is important for a long-term therapeutic perspective. The interpretation of CIs regarding statistical significance needs to be drawn with caution, as there is no formal adjustment for multiple testing. For formal statistical testing without repeated testing in one model, we provide the more complex longitudinal analysis (see below Longitudinal model and in the supplementary data, available at Rheumatology online). Tracking back of disease activity in remitters and non-remitters We then identified disease activity states of DAPSA at 6 months and 1 year (representing the long-term outcome of therapy). To this end, we applied the established DAPSA cut points of ⩽4 for REM, >4 and ⩽14 for LDA, >14 and ⩽28 for moderate disease activity (MDA), and >28 for high disease activity (HDA) [12]. We stratified patients according to their 6-month disease activity outcome (REM, LDA, MDA or HDA) and descriptively tracked their disease activity levels back to the respective baseline values. We repeated this exercise according to the 1-year states. Longitudinal model To supplement our main analyses, we developed a longitudinal statistical model to predict DAPSA states after treatment in PsA. We used a generalized estimation equation with independent covariance structure. For the sake of conciseness and clarity, we present the methodology and results of the longitudinal model, which were fully confirmative of our main analyses, in the Methods section of the supplementary data, available at Rheumatology online. cDAPSA In addition to the analyses focusing on the full DAPSA score, we evaluated the early predictive value of the purely clinical cDAPSA. In this case, we used disease activity cut points of ⩽4 for REM, >4 and ⩽13 for LDA, >13 and ⩽27 for MDA and >27 for HDA, as previously described [12]. All cDAPSA results are presented in the Results section of the supplementary data, available at Rheumatology online. We used SAS statistical software (Version 9.4) for all our analyses. Results Patient characteristics and treatment outcomes Including all patients who had DAPSA available at BL and 3, 6 and 12 months, we analysed a total of 216 patients from the GOL arm of the GO-REVEAL trial. Baseline characteristics are presented in Table 1; briefly, 40.7% were female, with a mean (s.d.) disease duration of 7.7 (7.5) years. Mean DAPSA at baseline amounted to 48.0 (26.7) and, according to the DAPSA classification [12], 0, 2.3, 20.8 and 76.9% of patients were in REM, LDA, MDA and HDA, respectively (supplementary Table S1, available at Rheumatology online). For external validation, we included 64 patients from IMPACT 2 and found very similar results (Table 1; Supplementary Table S1, available at Rheumatology online). Table 1 Patient baseline characteristics by trial and in total study population Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Columns specify patients included in the active treatment arms of GO-REVEAL and IMPACT 2 trials and pooled data of treatment groups in both trials. Numbers specify the mean (s.d.). NA: not applicable; VAS: visual analog scale. Table 1 Patient baseline characteristics by trial and in total study population Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Patient baseline characteristics GO-REVEAL (n = 216) IMPACT 2 (n = 64) P-value All (pooled treatment arms; n = 280) Age, years 46.4 (10.7) 48.8 (12.7) 0.1614 47.0 (11.2) Female, % 40.7 31.3 <0.0001 38.6 Disease duration, years 7.7 (7.5) 8.9(6.8) 0.0854 7.9 (7.3) HAQ 1.0 (0.6) NA NA 1.0 (0.6) Patient global assessment, VAS 52.9 (23.3) 52.3 (20.4) 0.9153 52.8 (22.6) Evaluator global assessment, VAS 53.6 (18.3) 53.3 (16.8) 0.8029 53.5 (18.0) Patient pain, VAS 55.1 (24.1) 55.6 (20.0) 0.9453 55.2 (23.2) CRP, mg/dl 1.4 (1.6) 1.5 (1.5) 0.0059 1.4 (1.6) Swollen joint count (SJC66) 13.1 (10.3) 10.7 (5.8) 0.2086 12.6 (9.5) Tender joint count (TJC68) 22.7 (16.9) 20.4 (13.6) 0.2558 22.2 (16.2) Simplified disease activity index 29.4 (13.9) 26.8 (9.75) 0.4701 28.8 (13.1) Clinical disease activity index 28.0 (13.6) 25.3 (9.59) 0.3785 27.4 (12.8) DAS28-CRP 4.9 (1.10) 4.8 (0.9) 0.6301 4.8 (1.1) PsA disease activity score 48.0 (26.7) 43.4 (17.1) 0.7812 46.9 (24.9) Clinical DAPSA 46.6 (26.4) 41.9 (17.1) 0.7212 45.5 (24.7) Columns specify patients included in the active treatment arms of GO-REVEAL and IMPACT 2 trials and pooled data of treatment groups in both trials. Numbers specify the mean (s.d.). NA: not applicable; VAS: visual analog scale. DAPSA levels early after treatment initiation are excellent tests for the achievement of good outcomes at later assessments Disease activity levels at different time points during follow-up showed excellent properties as a test for the subsequent achievement of good outcomes. Three-month disease activity levels were very strong predictors of LDA at 6 months, with an area under the ROC curve (AUC; and 95% CIs) of 0.92 (0.89, 0.96; Fig. 1A). Also, early relative change between BL and 3 months was a good predictor for an LDA status at 6 months (AUC 0.84; 95% CI: 0.79, 0.89; Fig. 1C). For prediction of disease activity at 1 year, AUC (95% CI) ranged between 0.73 (0.66, 0.79) for baseline DAPSA and 0.88 (0.84, 0.93) for 3-month values and 0.93 (0.89, 0.96) for 6-month values (Fig. 1B). Taking the percentage change of DAPSA from baseline to 3 and 6 months as a test for LDA after 1 year resulted in AUCs of 0.82 (0.76, 0.88) and 0.88 (0.83, 0.92), respectively (Fig. 1D). Validation analyses using IMPACT 2 data confirmed these results (supplementary Fig. S1, available at Rheumatology online). Fig. 1 View largeDownload slide Diagnostic testing: 6-month and 1-year achievement of low disease activity, based on earlier disease activity index for PsA levels or disease activity index for PsA percentage response Receiver operating characteristic (ROC) curves for achieving at least LDA after 6 months (A and C) or 1 year (B and D) based on DAPSA levels at baseline, 3 and 6 months (A and B) or DAPSA relative change from baseline to 3 or 6 months (C and D) as diagnostic tests. Areas under the ROC curves (AUC) including 95% CIs are presented in panel insets. Data: GO-REVEAL (n = 216). AUC: area under the curve; DAPSA: disease activity index for PsA; LDA: low disease activity; ROC: receiver operating characteristic. Fig. 1 View largeDownload slide Diagnostic testing: 6-month and 1-year achievement of low disease activity, based on earlier disease activity index for PsA levels or disease activity index for PsA percentage response Receiver operating characteristic (ROC) curves for achieving at least LDA after 6 months (A and C) or 1 year (B and D) based on DAPSA levels at baseline, 3 and 6 months (A and B) or DAPSA relative change from baseline to 3 or 6 months (C and D) as diagnostic tests. Areas under the ROC curves (AUC) including 95% CIs are presented in panel insets. Data: GO-REVEAL (n = 216). AUC: area under the curve; DAPSA: disease activity index for PsA; LDA: low disease activity; ROC: receiver operating characteristic. Early treatment response is indicative of achieving good outcomes at later time points Using logistic regression analysis, we estimated the chances for a beneficial outcome according to early relative DAPSA changes and disease activity levels at 3 months. Figure 2 illustrates that, to predict the achievement of the 6-month target of LDA reliably, substantial changes of DAPSA need to be seen at 3 months (Fig. 2A and C), and, likewise, considerable changes at the 3-month assessment are needed in order to achieve a reasonably high probability of a good outcome at 1 year (Fig. 2B and D). To achieve a sufficiently high probability of reaching a good outcome at 1 year, about the same changes in DAPSA are required at 3 and at 6 months (Fig. 2D). In other words, the predictive capacity at 3 months is not worse than at 6 months. Fig. 2 View largeDownload slide Probabilities of achieving low disease activity at 6 months or 1 year Predicted probabilities are shown for achieving at least LDA at 6 months (A and C) or 1 year (B and D) depending on DAPSA values at baseline, 3 and 6 months (A and B) or depending on DAPSA relative change from baseline to 3 or to 6 months (C and D). Data of patients included in the GO-REVEAL treatment arm (n = 216). DAPSA: disease activity index for PsA; LDA: low disease activity. Fig. 2 View largeDownload slide Probabilities of achieving low disease activity at 6 months or 1 year Predicted probabilities are shown for achieving at least LDA at 6 months (A and C) or 1 year (B and D) depending on DAPSA values at baseline, 3 and 6 months (A and B) or depending on DAPSA relative change from baseline to 3 or to 6 months (C and D). Data of patients included in the GO-REVEAL treatment arm (n = 216). DAPSA: disease activity index for PsA; LDA: low disease activity. Looking at absolute DAPSA levels among patients who achieved moderate disease activity (DAPSA >14 and ⩽28) after 3 months of therapy, the estimated probability of achieving at least LDA at 6 months was up to 69.8%, and patients with LDA at 3 months had an estimated probability for staying in LDA or reaching REM at 6 months of 70–90% (Fig. 2A). Achievement of a 50 or 75% response in DAPSA (DAPSA50, DAPSA75) at 3 months implied at least 50 or 73% probability of LDA or REM, respectively, and DAPSA85 even a probability of >80% for a good outcome at 6 months (Fig. 2C). Achieving DAPSA ⩽14 (LDA) after 3 months led to an estimated probability of ⩾78% for continuing to be in LDA (or REM) also after 1 year (Fig. 2B), and a DAPSA75 response from baseline to 3 months predicted >80% probability for at least LDA after 1 year (Fig. 2D). Validation analyses using IMPACT 2 data confirmed these results (supplementary Fig. S2, available at Rheumatology online). Tracking back patients who achieved different states at 6 months and 1 year: early differences in DAPSA levels Mean disease activity improved on TNFi very well over time, leading to a total of 39.9 and 47.5% of patients in at least LDA at 3 and 6 months, respectively (supplementary Table S1, available at Rheumatology online). At the end of follow-up after 1 year, 30.1% of patients were in REM, 28.7% in LDA, 22.2% in MDA and 19.0% in HDA (Table S1, available at Rheumatology online). Figure 3 shows that patients who achieve different DAPSA states at 6 months clearly differ at the group level already at the 3-month assessment (P ⩽ 0.0001). In addition, there were differences in baseline values at the group level, which emphasize that the starting level of disease activity is also an important determinant of the ability to reach a good outcome. Again, validation in the IMPACT 2 dataset was confirmatory (supplementary Fig. S3, available at Rheumatology online). Fig. 3 View largeDownload slide Tracking back patients with 6-month or 1-year remission, low, moderate and high disease activity Lines represent disease activity levels (means and s.e.m.) at baseline, 3 and 6 months and at 1 year in patients who achieved remission, low, moderate or high disease activity at the 6-month (A and B) or the 1-year (C and D) assessment. Disease activity states according to DAPSA (A and C) and cDAPSA (B and D). Data: GO-REVEAL (n = 216). BL: baseline; cDAPSA: clinical DAPSA; DAPSA: disease activity index for PsA; HDA: high disease activity; LDA: low disease activity; M: months; MDA: moderate disease activity; REM: remission; Y: year. Fig. 3 View largeDownload slide Tracking back patients with 6-month or 1-year remission, low, moderate and high disease activity Lines represent disease activity levels (means and s.e.m.) at baseline, 3 and 6 months and at 1 year in patients who achieved remission, low, moderate or high disease activity at the 6-month (A and B) or the 1-year (C and D) assessment. Disease activity states according to DAPSA (A and C) and cDAPSA (B and D). Data: GO-REVEAL (n = 216). BL: baseline; cDAPSA: clinical DAPSA; DAPSA: disease activity index for PsA; HDA: high disease activity; LDA: low disease activity; M: months; MDA: moderate disease activity; REM: remission; Y: year. Longitudinal analysis of DAPSA in relationship to treatment targets All data on longitudinal analyses using different models and statistical approaches are shown in supplementary Tables S2 and S3 and Figures S4–S7, available at Rheumatology online. Importantly, all these analyses using more complex statistical approaches fully confirmed the predictive capacity of DAPSA at 3 months for subsequent outcomes. Validation analyses The analyses presented above used data of the GO-REVEAL trial. External validation using the IMPACT 2 dataset led to consistent results (supplementary Figs S1–S3 and supplementary S7, available at Rheumatology online). In addition, internal validation of the longitudinal model using two different resampling methods in the GO-REVEAL dataset confirmed the presented model results (see supplementary Table S2, available at Rheumatology online, for estimates and CIs, and supplementary Figs S5 and S6, available at Rheumatology online). Finally, we supplemented our ROC results with one analysis using the outcome ACR70 response after 6 months and 1 year as the dependent variable, in contrast to the DAPSA-based analyses in the rest of the manuscript. Here, DAPSA levels at 3 and 6 months, but not baseline DAPSA, could predict 6-month and 1-year ACR70 response (see supplementary Fig. S8, available at Rheumatology online). Analyses of cDAPSA We performed all described analyses also for the cDAPSA, the clinical version of the full DAPSA, which does not include CRP. We found remarkably similar results to the data presented for DAPSA. Detailed results of the cDAPSA analyses are provided in supplementary Table S1 and supplementary Figs S9 and S10, available at Rheumatology online. Discussion In PsA management, prompt evaluation of newly implemented medication is important to avoid losing time on therapies that may not have the potential to reduce disease activity sufficiently in the individual patient. Given the fact that an increasing number of highly efficacious treatment options are available, any state other than LDA or REM should induce a change in therapy, because delay will have consequential effects of the cumulative inflammatory burden [3, 5, 6]. In this respect, a therapeutic evaluation at an early point in time should be attempted. Although response rates are increasing throughout the course of the whole first year of therapy, as evidenced by clinical trial data, the question arises about the minimal observation time that is required to estimate reliably the longer-term disease activity outcomes of the individual patient. According to the treat-to-target algorithm [1], this time frame has been proposed to be 3 months for improvement and 6 months for attaining the target. Indeed, in the present study the time between the 3- and 6-month assessments brought only limited additional information for the 1-year disease activity state. Moreover, when we simply tracked back disease activity levels of patients who had reached REM, LDA, MDA or HDA, it was apparent that disease activity at 3 months was already highly predictive of a good or poor outcome at 6 months and at 1 year; similar to what has been observed in RA [19]. In observational studies, factors that predict minimal disease activity in PsA have been explored, and there was evidence for male sex as well as CRP and ESR [14, 15] predicting favourable outcomes after TNFi treatment. However, minimal disease activity is a state, whereas the novelty of our analyses lies in the use of a continuous scale for PsA that can be used for predicting achievement of a disease state before a particular state is reached and in the identification of specific time points that serve in the process of clinical decision-making early after initiation of treatment. This enables the clinician to ascertain the likelihood of long-term therapeutic success. We show that the potential of a new therapy to suppress disease activity sufficiently at 6 months or 1 year can be predicted well at the 3-month time point, depending on whether or not patients prove to be responding at this early assessment. Consequently, treatment adaptions should be considered on this occasion. For our study, we used the DAPSA to define disease activity and categorize disease activity states. This score has previously been validated as a measure of PsA disease activity and is associated with functional and structural outcomes [8, 11, 20]. As such, it is a continuous outcome measure not only for the purpose of clinical management of PsA patients, but also serves outcomes research in the field of PsA, similar to the way that continuous scores have facilitated RA research over the past decade. Similar to previous findings in RA [19, 21], our results indicate that treatment success in PsA is highly predictable early after start of therapy by evaluating disease activity and, in particular, the change in disease activity after the start of therapy. Importantly, and in accordance with the treat-to-target recommendations, the presence of major DAPSA responses (DAPSA75, DAPSA85) is informative about whether a patient requires treatment modifications, or whether a therapy should be continued for at least an additional 3 months. Specifically, our prediction models indicate that relatively high response rates need to be observed at the 3-month assessment, in order to justify confidence in reaching the target state upon treatment continuation for another 3 months. These findings substantiate the call for an early appraisal of any new therapy, as has been formulated in therapeutic guidelines [1], to allow timely reaction with therapeutic adaptations in those who do not meet the desired level of improvement. Limitations to our analyses include the lack of follow-up data overseeing an even longer time span than 1 year. This was inherent to the use of data from clinical trials, which in our case had only 1 year of treatment with a single compound. Nevertheless, 3- and 6-month levels predicted 1-year outcomes in a similar manner, suggesting that a longer observation period would not necessarily provide additional clinically relevant information. The prediction model might differ slightly according to the mode of action of the drug used, which in our case were two TNFi. The DAPSA lacks enthesitis and dactylitis assessment, as well as skin involvement, for which separate validated instruments exist. The controversy regarding how to construct scores for the assessment of PsA has been discussed broadly at the recent international meeting that validated the treat-to-target approach for spondyloathritis, including PsA; indeed, DAPSA was one of the endorsed tools for follow-up of patients with PsA [1]. In line, we recently showed that patients who achieved DAPSA REM on a TNFi also had minimal enthesitis and dactylitis as well as Psoriasis Area and Severity Index scores, which did not differ from the very LDA state; likewise, DAPSA LDA conveyed similar non-articular outcomes to MDA [22]. Given the available core set variables in trial data, we show that achieving at least LDA in PsA can be predicted based on the parameters included in the DAPSA. The strength of the present study lies in the availability of patient-level data from two large randomized controlled clinical trials that investigated two of the most frequently used biologics in PsA therapy. Outcome measures in clinical trials are consistently evaluated by trained assessors, which is highly valuable for studies like the present one. We found very similar results in the two different trials, inherently validating the data presented. Finally, we consider the consistency of different methods addressing the same question as highly confirmatory. These methods included a diagnostic testing approach, which integrates sensitivity and specificity into the investigated association, and more advanced regression modelling, which is useful for clinicians estimating risks and benefits for their patients. In summary, our study demonstrates that the initial response to PsA therapy holds a highly significant predictive value and predicts long-term outcomes as early as 3 months after initiation of treatment. Thus, our results further support treatment algorithms in their demand for timely adaptations and provide guidance for rheumatologists who apply such strategies, as to whether and when adaptations are justified and, even more importantly, highly required. Our study provides some numerical guidance for the treating rheumatologists to quantify the probability of reaching the therapeutic target in their patients after specific disease activity states and amounts of change from baseline. Funding: No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this manuscript. Disclosure statement: J.S.S. has received grants for his institution from Abbvie, Astra-Zeneca, Janssen, Lilly, Merck Sharp & Dohme (MSD), Pfizer, Roche and has provided expert advice to and/or had speaking engagements for Abbvie, Amgen, Astra-Zeneca, Astro, Bristol-Myers Squibb (BMS), Celgene, Celltrion, Chugai, Gilead, Glaxo, ILTOO, Janssen, Lilly, Medimmune, MSD, Novartis-Sandoz, Pfizer, Roche, Samsung, Sanofi, Union Chimique Belge (UCB). D.B. is an employee of Janssen Research and Development. D.A. has received grants from MSD and BMS, speaker honoraria from AbbVie, Merck, UCB, Janssen, BMS, Pfizer, Medac and Roche and has consulted for Abbvie, Eli Lilly & Co., MSD, Centocor, Janssen. The other authors have declared no conflicts of interest. Supplementary data Supplementary data are available at Rheumatology online. References 1 Smolen JS , Schöls M , Braun J et al. Treating axial spondyloarthritis and peripheral spondyloarthritis, especially psoriatic arthritis, to target: 2017 update of recommendations by an international task force . Ann Rheum Dis 2018 ; 77 : 3 – 17 . Google Scholar CrossRef Search ADS PubMed 2 Gossec L , Smolen JS , Ramiro S et al. European League Against Rheumatism (EULAR) recommendations for the management of psoriatic arthritis with pharmacological therapies: 2015 update . Ann Rheum Dis 2016 ; 75 : 499 – 510 . Google Scholar CrossRef Search ADS PubMed 3 Coates LC , Moverley AR , McParland L et al. Effect of tight control of inflammation in early psoriatic arthritis (TICOPA): a UK multicentre, open-label, randomised controlled trial . Lancet 2015 ; 386 : 2489 – 98 . Google Scholar CrossRef Search ADS PubMed 4 Cresswell L , Chandran V , Farewell VT et al. Inflammation in an individual joint predicts damage to that joint in psoriatic arthritis . Ann Rheum Dis 2011 ; 70 : 305 – 8 . Google Scholar CrossRef Search ADS PubMed 5 Eder L , Thavaneswaran A , Chandran V , Cook R , Gladman DD. Increased burden of inflammation over time is associated with the extent of atherosclerotic plaques in patients with psoriatic arthritis . Ann Rheum Dis 2015 ; 74 : 1830 – 5 . Google Scholar CrossRef Search ADS PubMed 6 Shen J , Shang Q , Li EK et al. Cumulative inflammatory burden is independently associated with increased arterial stiffness in patients with psoriatic arthritis: a prospective study . Arthritis Res Ther 2015 ; 17 : 75 . Google Scholar CrossRef Search ADS PubMed 7 Gladman DD , Mease PJ , Healy P et al. Outcome measures in psoriatic arthritis . J Rheumatol 2007 ; 34 : 1159 – 66 . Google Scholar PubMed 8 Gladman DD , Mease PJ , Strand V et al. Consensus on a core set of domains for psoriatic arthritis . J Rheumatol 2007 ; 34 : 1167 – 70 . Google Scholar PubMed 9 Nell-Duxneuner VP , Stamm TA , Machold KP et al. Evaluation of the appropriateness of composite disease activity measures for assessment of psoriatic arthritis . Ann Rheum Dis 2010 ; 69 : 546 – 9 . Google Scholar CrossRef Search ADS PubMed 10 Coates LC , Fitzgerald O , Mease PJ et al. Development of a disease activity and responder index for psoriatic arthritis–report of the Psoriatic Arthritis Module at OMERACT 11 . J Rheumatol 2014 ; 41 : 782 – 91 . Google Scholar CrossRef Search ADS PubMed 11 Schoels M , Aletaha D , Funovits J et al. Application of the DAREA/DAPSA score for assessment of disease activity in psoriatic arthritis . Ann Rheum Dis 2010 ; 69 : 1441 – 7 . Google Scholar CrossRef Search ADS PubMed 12 Schoels MM , Aletaha D , Alasti F , Smolen JS. Disease activity in psoriatic arthritis (PsA): defining remission and treatment success using the DAPSA score . Ann Rheum Dis 2016 ; 75 : 811 – 8 . Google Scholar CrossRef Search ADS PubMed 13 Aletaha D , Alasti F , Smolen JS. Disease activity states of the DAPSA, a psoriatic arthritis specific instrument, are valid against functional status and structural progression . Ann Rheum Dis 2017 ; 76 : 418 – 21 . Google Scholar CrossRef Search ADS PubMed 14 Perrotta FM , Marchesoni A , Lubrano E. Minimal disease activity and remission in psoriatic arthritis patients treated with Anti-TNF-α drugs . J Rheumatol 2016 ; 43 : 350 – 5 . Google Scholar CrossRef Search ADS PubMed 15 Haddad A , Thavaneswaran A , Ruiz-Arruza I et al. Minimal disease activity and anti-tumor necrosis factor therapy in psoriatic arthritis . Arthritis Care Res 2015 ; 67 : 842 – 7 . Google Scholar CrossRef Search ADS 16 Kavanaugh A , McInnes I , Mease P et al. Golimumab, a new human tumor necrosis factor α antibody, administered every four weeks as a subcutaneous injection in psoriatic arthritis: twenty-four-week efficacy and safety results of a randomized, placebo-controlled study . Arthritis Rheum 2009 ; 60 : 976 – 86 . Google Scholar CrossRef Search ADS PubMed 17 Antoni C , Krueger GG , de Vlam K et al. Infliximab improves signs and symptoms of psoriatic arthritis: results of the IMPACT 2 trial . Ann Rheum Dis 2005 ; 64 : 1150 – 7 . Google Scholar CrossRef Search ADS PubMed 18 Eberl G , Studnicka-Benke A , Hitzelhammer H , Gschnait F , Smolen JS. Development of a disease activity index for the assessment of reactive arthritis (DAREA) . Rheumatology 2000 ; 39 : 148 – 55 . Google Scholar CrossRef Search ADS PubMed 19 Aletaha D , Funovits J , Keystone EC , Smolen JS. Disease activity early in the course of treatment predicts response to therapy after one year in rheumatoid arthritis patients . Arthritis Rheum 2007 ; 56 : 3226 – 35 . Google Scholar CrossRef Search ADS PubMed 20 Husic R , Gretler J , Felber A et al. Disparity between ultrasound and clinical findings in psoriatic arthritis . Ann Rheum Dis 2014 ; 73 : 1529 – 36 . Google Scholar CrossRef Search ADS PubMed 21 Aletaha D , Alasti F , Smolen JS. Optimisation of a treat-to-target approach in rheumatoid arthritis: strategies for the 3-month time point . Ann Rheum Dis 2016 ; 75 : 1479 – 85 . Google Scholar CrossRef Search ADS PubMed 22 Smolen JS , Aletaha D , Gladman DD et al. FRI0498 (2017) Outcomes associated with achievement of various treatment targets in patients with psoriatic arthritis receiving adalimumab . Ann Rheum Dis 2017 ; 76 (Suppl 2) : 677 . © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

RheumatologyOxford University Press

Published: Feb 22, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off