Crohn's disease (CD) is a chronic immune-mediated disease involving uncontrolled inflammation of the intestinal tract, which can lead to poor quality of life, disability, and surgeries. Conventional treatment strategies have focused on induction of a clinical remission using a step-wise approach to medical therapy with 5-aminosalicylates (5-ASA), corticosteroids, and immune modulators (e.g., thiopurines), but in recent years clinical trials of earlier use of immune-modifying or biologic therapies (or combinations of them) have been associated with more rapid remission and improved short- and longer-term outcomes.1-4

Tumor necrosis factor-α (TNF) plays an important role in mediating the inflammatory cascade in CD.5-7 The development of anti-TNF therapy for CD was an important advance in the management of this disease, and these therapies have become a cornerstone in the management of moderate to severe disease refractory to conventional therapy.8-11 Currently, there are three anti-TNF agents (infliximab, adalimumab, and certolizumab pegol) approved for use in patients with moderate to severe CD for whom conventional therapy has failed, and there is a substantial body of evidence demonstrating the efficacy of these agents in inducing and maintaining response and remission.12-19 Up to 40% of patients with active CD receive anti-TNF agents in combination with thiopurines or methotrexate.20 As experience and success with anti-TNF therapy have grown, some have suggested that an early introduction (top-down) of these agents may improve outcomes versus the traditional step-up approach.21 However, it has been postulated that physician and patient concerns about safety and cost of the earlier use of anti-TNF therapies have limited acceptance of this approach. The frequency with which such top-down strategies are employed and the outcomes in a nonclinical trial setting of community practitioners is not known. Therefore, the aim of this study was to determine, using claims data of patients with CD, if a top-down approach with anti-TNF therapy is associated with improved clinical outcomes.

MATERIALS AND METHODS Patients

The patient database was from PharMetrics, an administrative claims database that includes a systematic sample of more than 94 commercial health plans throughout the United States and contains more than 58 million commercially insured patients. To be eligible for the study, patients had to be continuously enrolled in the same health plan for at least 6 months prior to the initial diagnostic claim for CD and remain enrolled through 12 months after an initial anti-TNF agent claim for CD.

To verify their diagnosis, patients also had to have two or more claims for CD (International Classification of Diseases, Ninth Revision [ICD-9] 555, 555.0, 555.1, 555.2, 555.9) within 6 months and one or more claims for anti-TNF therapy after their initial diagnosis for CD. Eligible patients were followed from the date of their CD diagnosis through the initiation of anti-TNF therapy and thereafter up to 24 months. Study data contained claims from January 2000 through January 2009.

Three patient groups were identified based on treatment pathways through biologic initiation: a group which used 5-ASA and/or corticosteroids and/or immunosuppressants prior to anti-TNF therapy (called “Step-Up”); an immunosuppression (IS)-to-TNF inhibitor group, which used immunosuppressive therapy (not 5-ASA) prior to anti-TNF agent initiation (called “IS-to-TNF”); and a group that initiated anti-TNF therapy within 30 days of the first prescription for CD (called “Early-TNF”). A change between therapies was defined as ≥30 days of one class of drugs prior to the start of a second drug.

Outcomes

Response to anti-TNF therapy was measured at 6-month increments up to 24 months following TNF inhibitor initiation. The following response measures were evaluated to determine response to therapy: concomitant corticosteroid use (corticosteroid use was assumed to be indicative of corticosteroid-dependence, ongoing disease activity, and/or flares); CD surgery (at least one CD surgery during the anti-TNF treatment period); anti-TNF dose escalation as an indicator of loss of response (defined as a claim for the biologic therapy that is at least double the standard maintenance dose); and TNF inhibitor discontinuation/switch (combined). Anti-TNF dose escalation was defined as a two-fold increase in the estimated number of pens/syringes for a weekly dose of adalimumab and either a two-fold increase in the estimated number of vials every 4 or every 8 weeks or 4-weekly infusions for infliximab. Discontinuation was defined as the cumulative percentage of patients (not including surgery patients) who discontinued the anti-TNF agent without starting another anti-TNF agent. Rate of surgery was an independent metric. A switch was defined as the cumulative percentage of patients that initiated a second anti-TNF agent after discontinuing the first one. Response metrics were analyzed overall and separately among the three cohorts.

Statistical Analysis

Demographic data were analyzed to control for key differences among cohorts. One-way analysis of variance was used to test for differences in the mean age of the three cohorts. Other demographics (sex, geographic region, and site of intestinal involvement) were compared using the χ2 test. Relative risk (RR) values and their confidence limits were estimated by log-binomial regression. The statistical differences were tested using the χ2 test. All data analyses, including the statistical testing, were performed using SAS v. 9.2 (SAS Institute, Cary, NC).

RESULTS Patient Characteristics and Disposition

Demographic data across the three patient groups were similar (Table 1). The majority of patients were female (60%), age 20–60 years (75%), and from the Eastern, Southern, and Midwest regions of the U.S. (90%). Most patients had CD involvement in both the colon and small intestine (61%). The most common comorbid conditions were hypertension (16%), anal fistula (11%), hypercholesterolemia (10%), and asthma (5%).

Table 1. Patient Demographics and Clinical Characteristics Early-TNF (n = 1321) IS-to-TNF (n = 1031) Step-up (n = 1398) Probability Female, n (%) 743 (56) 578 (56) 789 (56) 0.9829 Male, n (%) 578 (44) 453 (44) 609 (44) Age, y  Mean 41 39 40 0.098  Median 41 39 40  25th percentile 28 26 28  75th percentile 53 49 52 Geographic region, n (%)  East 325 (25) 278 (27) 343 (25) 0.2192  Midwest 486 (37) 405 (39) 564 (40)  South 374 (28) 255 (25) 362 (26)  West 136 (10) 93 (9) 129 (9) Location of CD, n (%)  Ileum 117(9) 84 (8) 122 (9) 0.7899  Colon 199 (15) 166 (16) 211 (15)  Ileum/colon 795 (60) 631 (61) 870 (62)  Not specified 210 (16) 150 (15) 195 (14) Key Claims Data Analysis

The treatment path prior to biologic use among patients newly diagnosed with CD is shown in Figure 1. There were 1398 (37%) patients in the Step-Up group, divided into 973 patients who switched from 5-ASA to immunosuppressant and then to an anti-TNF agent and 425 patients who switched directly from 5-ASA to a TNF inhibitor. There were 1031 and 1321 patients in the IS-to-TNF and Early-TNF groups, respectively. The duration between a CD diagnosis to the first biologic claim among the groups was the longest in the Step-Up group (median = 463 days) and the shortest in the Early-TNF group (median = 87 days). Patients in the Early-TNF group did indeed receive anti-TNF earlier in their disease compared with IS-to–TNF and Step-Up groups (P < 0.001, Kolmogorov–Smirnov test).

Treatment path of patients newly diagnosed with CD prior to biologic use. The change between therapies was defined as ≥30 days of one class of therapy prior to the next therapy. At each stage of the treatment path patients could have received combined therapies. Figure shows the median time (in days) between each step for each type of treatment path.

Response Measures

In general, the Early-TNF group had a lower RR of concomitant corticosteroid use, dose escalation, discontinuation/switch, and CD-related surgery compared with the Step-Up and IS-to-TNF groups (Table 2).

Table 2. Patient Response to Anti-TNF Therapy Based on Treatment Path Response Measure, RR (95% CI for RR) Time After Start of Anti-TNF Therapy 6 Months 12 Months 18 Months 24 Months Step-up vs. Early-TNF IS-to-TNF vs. Early-TNF Step-up vs. Early-TNF IS-to-TNF vs. Early-TNF Step-up vs. Early-TNF IS-to-TNF vs. Early-TNF Step-up vs. Early-TNF IS-to-TNF vs. Early-TNF Corticosteroid use 1.86* (1.64–2.10) 1.60* (1.40–1.83) 1.91* (1.66–2.20) 1.45* (1.23–1.70) 1.99* (1.67–2.37) 1.47* (1.21–1.79) 1.79* (1.47–2.18) 1.42* (1.14–1.76) Dose escalationa 1.05 (0.86–1.28) 1.01 (0.82–1.25) .93 (0.76–1.14) .84 (0.67–1.05) 1.28* (1.03–1.60) .87 (0.68–1.11) 1.36* (1.05–1.76) .85 (0.64–1.15) Discontinue/ switch 1.13* (0.98–1.30) 1.08 (0.93–1.27) 1.20* (1.09–1.33) 1.14* (1.02–1.28) 1.34* (1.23–1.47) 1.22* (1.11–1.35) 1.47* (1.36–1.59) 1.36* (1.24–1.48) CD-related surgery 1.33 (0.93–1.90) 1.57* (1.09–2.27) 1.44* (1.08–1.92) 1.62* (1.20–2.18) 1.50* (1.16–1.93) 1.61* (1.24–2.10) 1.43* (1.12–1.81) 1.68* (1.32–2.15) * P < 0.05. a Defined as a two-fold increase in the estimated number of pens/syringes for a weekly dose of adalimumab and either a two-fold increase in the estimated number of vials every 4 or every 8 weeks or 4-weekly infusions for infliximab. CI, confidence interval.

The Early-TNF group had a lower risk of concomitant corticosteroid use compared with the Step-Up and IS-to-TNF groups within 6 months following biologic initiation (P < 0.05 for both groups vs. Early-TNF group) (Fig. 2A). Although the concomitant corticosteroid use declined in the Step-Up and IS-to-TNF groups at months 6 through 24 of biologic initiation, the Early-TNF group had significantly lower rates of concomitant corticosteroid use than both the Step-Up and IS-to-TNF groups (P < 0.05 for both groups vs. Early-TNF group). The percentage of patients with continuous corticosteroid use decreased more than 50% at months 6 through 24 of biologic initiation in all three patient groups (Fig. 2B). The percentage of patients with continuous corticosteroid use in the IS-to-TNF group (16%–40%) was higher than in the Early-TNF group at months 6 through 18 of biologic initiation (10%–25%) but not at month 24. Compared with the Early-TNF group, the cumulative percentage of patients with corticosteroid exposure was significantly higher in the IS-to-TNF and Step-Up groups at months 6 through 24 of biologic initiation (P < 0.05 for both groups vs. Early-TNF group). Forty-four percent of patients in the Early-TNF group had corticosteroid exposure at month 24 following biologic initiation versus 64% and 71% in the IS-to-TNF and Step-Up groups, respectively (P < 0.05).

Concomitant (A) and continuous (B) corticosteroid use during anti-TNF therapy. *P < 0.05 Early-TNF groups versus other groups. †P < 0.05IS-to-TNF group versus other groups.

During each 6-month evaluation period, more than 15% of patients in each of the three groups had one or more claims at an increased dose (more than double the recommended maintenance level) of the anti-TNF agent (Fig. 3). During months 13–18 and 19–24 (following biologic initiation), patients in the Step-Up group had significantly higher rates of dose escalation compared with the Early-TNF and IS-to-TNF groups (25% vs. 18%–19% and 16%–17%, respectively) (P < 0.05 for both groups vs. Step-up group). By month 24 the cumulative percentage of patients with dose escalation exposure was significantly higher in the Step-Up group (38%) compared with the IS-to-TNF group (32%) (P < 0.05).

Dose escalation during anti-TNF therapy. Dose escalation of adalimumab was defined as a two-fold increase in the estimated number of pens/syringes for a weekly dose. Dose escalation for infliximab was defined as either a two-fold increase in the estimated number of vials every 4 or every 8 weeks or 4-weekly infusions for infliximab.

Compared with the Early-TNF group, discontinuation/switch was significantly higher for the Step-Up group by month 6 (RR 1.13, P < 0.05) and the IS-to-TNF group by month 12 (RR 1.14, P < .05). At 1 year, discontinuation/switch in the Early-TNF group was 38%, while the IS-to-TNF and Step-Up groups were 43% and 45%, respectively. At 2 years, discontinuation/switch in the Early-TNF group was 53%, while the IS-to-TNF and Step-Up groups were 71% and 77%, respectively (P < 0.05 for both groups vs. Early-TNF group).

As shown in Figure 4 and Table 2, the cumulative percentage of patients with CD-related surgery and the RR of CD-related surgery was significantly lower in the Early-TNF group compared with the Step-Up and IS-to-TNF groups by the first year (RR 1.44 and 1.62, respectively, P < 0.05 for both groups vs. Early-TNF group). The higher rate of CD-related surgery continued through month 24, during which 14% of patients in the IS-to-TNF had surgery (RR 1.68 vs. Early-TNF group, P < 0.05) and 12% of patients in the Step-Up group had surgery (RR 1.43 vs. Early-TNF group, P < 0.05).

CD-related surgery during anti-TNF therapy.

DISCUSSION

This study is the first assessment of real-world management of patients with CD in the era of evolving evidence favoring earlier use of immune suppressive and biologic therapies and demonstrates that a significant number of patients with CD are receiving such therapies with favorable outcomes. The results demonstrate that, in general, earlier use of anti-TNF therapy is associated with a lower risk of concomitant corticosteroid use, anti-TNF dose escalation, discontinuation/switch of anti-TNF therapy, and CD-related surgery compared with conventional models of step-up or immunosuppression to anti-TNF strategies.

This is the largest assessment of the outcome of different strategies of anti-TNF use reported and suggests that earlier use is associated with enhanced outcomes using the measures that could be ascertained from the source data. Conversely, it demonstrates that early use of 5-ASA is associated with worse outcomes. The large number of patients in this study who were treated with 5-ASA earlier in their disease management demonstrates the ongoing prevalence of this therapeutic modality in CD in the U.S. This study demonstrates in a “real-world” assessment that these patients were more likely to have undesirable outcomes at 2 years, and lends to the evidence and ongoing vigorous discussions that the safety of 5-ASA therapies does not warrant their inappropriate use in many patients with CD. Although we could not break the patient groups into their disease behavior types, it is acknowledged that penetrating CD and CD in the perineum are certainly not disease types in which 5-ASA would play any role. Given that this entire study was focused on patients who ultimately received anti-TNF therapies, it is worthwhile to underscore these outcomes, and continue to educate gastroenterologists about the significantly limited role of 5-ASA for many CD patients. This study shows that the time lost while treating with an ineffective therapy is not at no cost to patients or their outcome.

Although the use of a large administrative dataset provides the ability to assess outcomes in the largest number of patients assessed in this manner to date, the use of administrative databases do have limitations. This is a study of all patients with CD who received TNF-inhibitor therapy in this dataset. It must be noted that patients with CD who were managed with other therapies and never received anti-TNF therapy were by design not included. This was intentionally performed as a way to analyze what was assumed to be similar patient groups. However, differences in disease severity or behavior by conventional clinical trial measures (such as the Crohn's Disease Activity Index or even the Montreal Classification system) could not be assessed. Therefore, it is possible that patients in one cohort or another may have had more or less disease severity or behavioral type than the others. In addition, while it is possible that some patients with treatment discontinuations may be due to stable disease remission, current standard of care and standard of practice in the U.S. does not include drug discontinuation. Therefore, and especially in the shorter duration of follow-up in this study, we believe this was a less likely explanation for discontinuation.

Claims data analysis requires continuous enrollment in the same health plan. Because of the high rates of switching health plans in the U.S. (the source of the PharMetrics data), the overall patient sample in this study is biased toward patients who were more recently diagnosed with CD and who started anti-TNF therapy earlier than the general CD patient population. As with any claims analysis, it is certainly possible that patients with known CD were followed without a code or prescription therapy for the 6-month lead-in, but this seems very unlikely, especially with the group that was ultimately identified for this study. Such a lead-time bias might in fact not favor an Early-TNF approach, since these patients would most likely have milder disease or smoldering fibrotic disease. Despite these standard limitations of claims analysis, the results of this analysis do demonstrate that a “top-down” or early approach to anti-TNF therapy, compared with more traditional “step-up” approaches, is associated with improved clinical outcomes. Patients who start anti-TNF therapy earlier after a CD diagnosis may require less overall and less continuous use of corticosteroids and may experience lower rates of anti-TNF dose escalation and improved persistency (i.e., lower rates of discontinuation and switching).

As with any claims database, the accuracy of the claims is an important potential limitation. We believe that the enrollment criteria, multiple claims, and receipt of the anti-TNF therapy are all evidence that this was indeed a group of patients with CD.

Chronic use of corticosteroids is limited by serious adverse effects, which correlate with the dose and the duration of treatment.22 In the present study, patients in the Early-TNF group had significantly lower rates of concomitant corticosteroid use than both Step-Up and IS-to-TNF groups, and fewer than one half of the patients in the Early-TNF group had corticosteroid exposure at month 24 following biologic initiation. In contrast, 64% and 71% of patients had corticosteroid exposure at month 24 in the IS-to-TNF and Step-Up groups, respectively.

Loss of response to TNF inhibitors, defined as loss of response and/or development of acute or delayed hypersensitivity reactions or injection-site reactions, is an important problem from both the clinical and economic perspectives.12, 13, 18 Patients in the Early-TNF group had a lower risk of dose escalation compared with Step-Up group patients 18 months following biologic initiation and a lower risk of discontinuing or switching treatment compared with both the Step-Up and IS-to-TNF groups 12 months after biologic initiation. Thus, a top-down approach to anti-TNF therapy may avert secondary loss of response in some patients.

Despite increased acceptance and use of immunosuppressive therapy in CD, the need for CD-related surgery has remained significant over the past few decades.23 In the present study, patients in the Early-TNF group experienced a lower percentage of CD-related surgery compared with Step-Up and IS-to-TNF patients at 1 year following biologic initiation. These results are consistent with those of D'Haens et al,3 who compared early use of infliximab and azathioprine with conventional management in patients with active CD who had not previously received corticosteroids, antimetabolites, or infliximab. In a 2-year, open-label trial, 133 patients were randomly assigned to either early combined immunosuppression or conventional treatment. At week 26, 60% of patients in the combined infliximab/azathioprine group were in remission without corticosteroids and without surgical resection, compared with 36% of controls (P = 0.0062). Corresponding rates at 1 year were 62% and 42% (P = 0.0278). Thus, reduced rates of CD-related surgery may be related to earlier use of anti-TNF therapy in some patients.1

In summary, the results of this study suggest that patients in this dataset with active CD who received anti-TNF therapy earlier in their treatment course had lower rates of corticosteroid use, fewer CD-related surgeries, and a greater rate of stable dosing and persistency of the anti-TNF therapy. As we work toward improved outcomes and disease-modifying therapeutic strategies, future effort should clarify the efficacy, safety, and cost-effectiveness of this approach.

Acknowledgements

This study was conceived and designed by the investigator (D.R.) and scientists at C1 Consulting (R.S. and O.U.). All of the authors were involved in the analysis and interpretation of data, writing the article, including development of the first draft, and writing or approving all subsequent revisions. The corresponding author (D.R.) had full access to all of the data and takes full responsibility for the veracity of the data and analysis. All authors were involved in the decision to submit for publication and approved the final version. The statistical analysis of the entire datasets was confirmed by a C1 Consulting-employed statistician according to stringent and validated standard operating procedures. Editorial assistance was provided by a contract medical writer (Ann P. Tighe, PhD, PPSI, a PAREXEL company, US) supported by UCB. Conflicts of interest: David T. Rubin, MD, Consultant: UCB, Abbott Immunology and Janssen (formerly Centocor). Financial support for research: Abbott Immunology (safety registry). No payment was received for work on this study. Robert Sederman, Consultant, UCB. Ozgecan Uluscu, PhD, Consultant, UCB.