Crohn s disease (CD) is characterized by recurrent

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2006;4: Serious Infections and Mortality in Association With Therapies for Crohn s Disease: TREAT Registry GARY R. LICHTENSTEIN,* BRIAN G. FEAGAN, RUSSELL D. COHEN, BRUCE A. SALZBERG, ROBERT H. DIAMOND, DONNY M. CHEN, # MICHELLE L. PRITCHARD, # and WILLIAM J. SANDBORN** *Department of Medicine, Division of Gastroenterology, Hospital of the University of Pennsylvania, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania; Robarts Clinical Trials, Robarts Research Institute, London, Ontario, Canada; Department of Medicine, Section of Gastroenterology, University of Chicago Medical Center, Chicago, Illinois; Atlanta Gastroenterology Associates, Atlanta, Georgia; Centocor, Inc., Horsham, Pennsylvania; # Ovation Research Group, Highland Park, Illinois; and the **Division of Gastroenterology and Hepatology, Mayo Clinic and Foundation, Rochester, Minnesota Background & Aims: Long-term safety data for infliximab and other therapies in Crohn s disease (CD) are needed. Methods: We prospectively evaluated patients for prespecified safety-related outcomes. Results: As of August 2004, 6290 patients were enrolled; 3179 received infliximab (5519 patient-years), 87% of whom received at least 2 infusions, and 3111 received other therapies (6123 patient-years). The mean length of follow-up evaluation was 1.9 years. More infliximab-treated patients had moderate-to-severe (30.8% vs 10.3%) or severe-fulminant (2.5% vs.6%) CD, and had surgical (17.5% vs 13.8%) or medical (14.4% vs 9.1%) hospitalizations in the previous year. More patients were taking prednisone (27.4% vs 16.1%), immunomodulators (49.4% vs 32.2%), or narcotic analgesics (9.8% vs 5.4%) when compared with those receiving other therapies (P <.001, all comparisons). The mortality rates were similar for infliximaband non infliximab-treated patients (.53 per 100 patient-years vs.43; relative risk, 1.24; 95% confidence interval [CI], ). In multivariate logistic regression analysis, only prednisone was associated with an increased mortality risk (odds ratio [OR], 2.10; 95% CI, ; P.016). Although the unadjusted analysis showed an increased risk for infection with infliximab use, multivariate logistic regression analysis suggested that infliximab was not an independent predictor of serious infections (OR,.99; 95% CI, ). Factors independently associated with serious infections included prednisone use (OR, 2.21; 95% CI, ; P <.001), narcotic analgesic use (OR, 2.38; 95% CI, ; P <.001), and moderateto-severe disease activity (OR, 2.11; 95% CI, ; P.024). Conclusions: Mortality rates were similar between infliximab- and non infliximabtreated patients. The increased risk for serious infection observed with infliximab likely was owing to disease severity and prednisone use. Crohn s disease (CD) is characterized by recurrent inflammation at any location in the gastrointestinal tract. The pharmacologic management of CD is based on the location, extent, and disease severity. 1 3 Therapeutic options include aminosalicylates, antibiotics, and immunosuppressive agents such as corticosteroids, antimetabolite immunomodulators (eg, 6-mercaptopurine, azathioprine, methotrexate), and infliximab. Infliximab has been proven to be efficacious in patients with either luminal 4 7 or fistulizing CD. 8,9 The use of corticosteroids is associated with a high incidence of both short- and long-term side effects including Cushing s syndrome, osteoporosis, diabetes, and infection. 10,11 The use of immunomodulators is complicated by a delayed onset of action and side effects including bone marrow suppression, hepatotoxicity, and pancreatitis. 12 Because both corticosteroids and immunomodulators have broad effects on the immune system, patients treated with these drugs may be at increased risk for serious or fatal infections. 13,14 Infliximab, which specifically targets tumor necrosis factor-, also is associated with adverse events including infections. In large controlled clinical trials of infliximab in patients with CD, the percentage of patients with serious infections ranged from 4.0% to 4.6% and the mortality rate ranged from.7% to 1.3%. 6,9 The occurrence of infliximab-related adverse events, including serious opportunistic infections such as tuberculosis, listeriosis, histoplasmosis, and death, also has been documented in reports of open-label trials or retrospective trials Abbreviations used in this paper: CD, Crohn s disease; CI, confidence interval; OR, odds ratio; RR, relative risk; SAE, serious adverse event 2006 by the American Gastroenterological Association Institute /06/$32.00 doi: /j.cgh

2 622 LICHTENSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 5 Gastroenterologists remain concerned about the safety of CD treatments. Recognizing that registries are useful in evaluating safety by prospectively following-up large numbers of patients in real-world clinical settings for an extended time and representing cohorts of patients with the same disease who are treated with different drugs, we established a registry, The Crohn s Therapy, Resource, Evaluation, and Assessment Tool (TREAT). The TREAT Registry is a large-scale, ongoing, observational registry that was designed to examine the safety of CD therapies, including infliximab. Methods Study Design The TREAT Registry is a prospective, observational, multicenter, long-term registry of North American patients with CD. The registry was initiated in 1999 to evaluate the clinical safety outcomes of various treatment regimens, including infliximab, in the management of CD. Approximately 350 gastroenterologists from both community-based and academic practice settings were each to enroll up to 150 patients for a target enrollment of at least 5000 patients. Most physicians were identified from the membership list of the Crohn s and Colitis Foundation Association. Patients were treated at the discretion of their physicians (ie, there was no defined treatment protocol). Physician participation in the registry could be withdrawn if patient enrollment requirements were not met, complete data were not submitted, a physician was leaving the practice, or if a physician elected to discontinue participation. The design of the TREAT Registry was approved by institutional review boards. All patients provided written informed consent. Physicians or their designees were paid a small honorarium on a per-patient basis as compensation for participating in the registry. Centocor, Inc. (Malvern, PA), the manufacturer of infliximab, sponsored the TREAT Registry, but all data were collected, collated, managed, and analyzed by Ovation Research Group (Highland Park, IL), an independent research organization, under the supervision and guidance of an Advisory Committee comprising several authors of this publication (G.R.L., B.G.F., R.D.C., B.A.S., and W.J.S.), and the TREAT medical monitor (R.H.D.). Registry Participants Patients must have had a diagnosis of CD and could not be participating in any clinical trial. Initially, patients younger than 18 years of age participated but the protocol subsequently was amended to limit enrollment to those age 18 years or older. Study Evaluations In this ongoing registry, data are collected at enrollment and then on a semiannual basis (each January and July). Data collection will continue for a minimum of 5 years. Patient demographic information and physicians assessments of overall patient health were collected at enrollment. Physicians assessed disease severity (remission, mild-moderate, moderate-severe, or severe-fulminant) according to the American College of Gastroenterology Guidelines. 19 The follow-up data included assessment of disease severity, medication use, adverse events, dates of infliximab infusions, and outcomes of each infusion. The start and stop dates for other medications were not collected. Serious adverse events (SAEs) and changes in pregnancy status were collected as they occurred. Physicians were required to report SAEs within 24 hours of occurrence. A detailed description of the SAE including dates of onset and resolution; designation that the SAE was expected or unexpected; the seriousness and intensity of the SAE; the relationship of the SAE to the underlying CD; concomitant CD medications or surgical procedures for CD; and the treatment and outcome of the SAE, was collected. Statistical Methods Baseline demographic and disease characteristics are summarized for the cohort of all patients combined and by treatment cohort. The cohort of patients who received infliximab within 12 weeks before enrollment, who were scheduled to receive infliximab within 30 days of enrollment, or who received infliximab at some other point in the registry was termed infliximab-treated patients. Because it is not possible to document infliximab administration before participation in the registry, only patients who met the criteria outlined previously are considered infliximab-treated. Patients who did not receive infliximab during the registry are termed other-treatments-only patients. Means and SDs were calculated for continuous variables (age, body mass index, and years between CD diagnosis and enrollment). Frequencies and percentages were generated for categoric outcomes (sex, race, involved intestinal segment[s], disease severity, use of immunomodulators, use of prednisone, use of narcotic analgesics, and health resource use). Corticosteroid use represented prednisone use exclusively; immunomodulator use represented the use of azathioprine, 6-mercaptopurine, and/or methotrexate. The Student t test was used to test for equality of means across treatment groups. The 2 test was used to evaluate the association between treatment group and categoric variables. The primary analyses evaluated the rates of death and rates of serious infection among medication categories. If a patient had multiple serious infections, then the first serious infection was considered for this analysis. For each event, the rate of the events per 100 person-years was calculated for each medication category (infliximab vs other treatments) as the quotient of the total number of on-therapy events and patient-years of exposure to medication multiplied by 100. Two different definitions of exposure time were used: (1) total exposure to infliximab was defined as the time between the first infliximab infusion through the date of the last follow-up evaluation, and (2) exposure to infliximab was defined as having received infliximab within 3 months. Patients could accumulate exposure time in both infliximab and othertreatments-only categories, depending on infliximab start and

3 May 2006 SAFETY WITH INFLIXIMAB: TREAT REGISTRY 623 Table 1. Baseline Patient Demographics Characteristic All patients Infliximab treated a Other treatments only P value b Total patients, N 6290 (100.0%) 3179 (50.5%) 3111 (49.5%) Age at enrollment, y Mean SD Sex, N Female 3619 (57.5%) 1842 (57.9%) 1777 (57.1%).24 Male 2559 (40.7%) 1289 (40.5%) 1270 (40.8%) Unknown 112 (1.8%) 48 (1.5%) 64 (2.1%) Race, N Caucasian 5601 (89.0%) 2823 (88.8%) 2778 (89.3%).25 African American 421 (6.7%) 228 (7.2%) 193 (6.2%) Asian 24 (0.4%) 15 (.5%) 9 (.3%) Hispanic 81 (1.3%) 39 (1.2%) 42 (1.4%) Other 42 (.7%) 22 (.7%) 20 (.6%) Unknown/missing 121 (1.9%) 52 (1.6%) 69 (2.2%) BMI at enrollment, kg/m 2 Mean SD BMI, body mass index. a Infliximab-treated patients are those patients who received infliximab within 12 weeks before enrollment, who were scheduled to receive infliximab within 30 days of enrollment, or who received infliximab at some other point in the registry. b P value from t test (continuous variables) or 2 test (categoric variables). stop dates. A Poisson regression analysis, accounting for within-patient correlations and as implemented by the GENMOD procedure in SAS (SAS Institute Inc., Cary, NC), tested for the differences in event rates between medication categories. In these analyses, a 2-sided P value of.05 was considered statistically significant. Multivariable regression modeling was used to explore the effects of baseline demographic characteristics (age, sex, race, diseased segment[s], disease severity, and years between diagnosis and enrollment in the TREAT Registry) and medication use as predictors of death and serious infection. For mortality analyses, CD medications, designated as ever used were obtained from all available 6-month periods ( January June and July December) that occurred between enrollment and the time of death. Medication use during the period in which the death occurred was included in this analysis. For analysis of serious infection, CD medication ever used was obtained from all available 6-month periods that occurred from enrollment through the period before the onset of the serious infection. Medication use during the period in which the infection occurred was not included in these analyses because medication start and stop dates were not collected for any medications other than infliximab. Therefore, it was unknown whether the medication was given in response to the event, or whether the medication caused the event. After examining the unadjusted models, demographic characteristics and medication ever used between enrollment and the event were included in a multivariate logistic model. 20 The Cox proportional hazards model with time-varying covariates also was used to assess the risk for an event in a given period based on drug use in the prior period. Because the core results identifying risk factors were consistent between the 2 approaches, results from the logistic regression are presented. The data collected through August 23, 2004, were analyzed using SAS software, version 8.02 or higher of the SAS System for Windows. No adjustments were made for multiple comparisons. Results Through August 2004, 6290 CD patients from 212 centers were enrolled from both community (82%) and academic (18%) centers. Most patients were female (58%) and Caucasian (89%), with an average age of years. The population was distributed almost equally between those who had been treated with infliximab (N 3179, 50.5%) and those who had not. More than 85% of patients treated with infliximab had received at least 2 infusions. Among patients who received infliximab, a median of 5 (range, 1 32) infusions were received. Most patients (86%) received infliximab 5 mg/ kg. Among the patients who had received at least 2 infusions, the average time between infusions was days. Approximately 16% of patients took budesonide and.5% of patients took cyclosporine at some point while enrolled in the registry. Infliximabtreated patients were younger than patients in the othertreatments-only group (P.001), but did not differ with regard to sex or race (Table 1). Seventy-four patients who were younger than 18 years of age were enrolled, but their enrollment was discontinued when the protocol was amended to include only patients age 18 or older. Patients in the infliximab-treated group differed significantly from patients in the other-treatments-only group with regard to intestinal segment(s) affected by

4 624 LICHTENSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 5 Table 2. Baseline Disease Characteristics Characteristic All patients Infliximab treated a Other treatments only P value b Total patients, N 6290 (100.0%) 3179 (50.5%) 3111 (49.5%) Years between diagnosis and enrollment Mean SD Diseased segment(s), N Ileum only 1806 (28.7%) 799 (25.1%) 1007 (32.4%).001 Colon only 1767 (28.1%) 895 (28.2%) 872 (28.0%) Ileum and colon 2475 (39.3%) 1374 (43.2%) 1101 (35.4%) Unknown/missing 242 (3.8%) 111 (3.5%) 131 (4.2%) Disease severity, N Remission c 1535 (24.4%) 387 (12.2%) 1148 (36.9%).001 Mild/moderate d 3081 (49.0%) 1592 (50.1%) 1489 (47.9%) Moderate/severe e 1299 (20.7%) 979 (30.8%) 320 (10.3%) Severe/fulminant f 99 (1.6%) 81 (2.5%) 18 (.6%) Unknown/missing 276 (4.4%) 140 (4.4%) 136 (4.4%) Health resource use in year before enrollment Surgical admissions, N 985 (15.7%) 557 (17.5%) 428 (13.8%).001 Medical admissions, N 741 (11.8%) 459 (14.4%) 282 (9.1%).001 Medication use at enrollment Prednisone, N 1372 (21.8%) 870 (27.4%) 502 (16.1%).001 Immunomodulators, N (%) 2575 (40.9%) 1572 (49.4%) 1003 (32.2%).001 Narcotic analgesics, N (%) 480 (7.6%) 313 (9.8%) 167 (5.4%).001 a Infliximab-treated patients are those patients who received infliximab within 12 weeks before enrollment, who were scheduled to receive infliximab within 30 days of enrollment, or who received infliximab at some other time during the registry. b P value from t test (continuous variables) or 2 test (categoric variables). c Remission refers to patients who are asymptomatic or without inflammatory sequelae and refers to patients who have responded to acute medical intervention or who have undergone surgical resection without evidence of residual disease. Patients requiring corticosteroids to maintain well-being are considered to be steroid-dependent and are not in remission. d Mild to moderate disease applies to ambulatory patients who are able to tolerate oral alimentation without manifestations of dehydration, toxicity (high fevers, rigors, prostration), abdominal tenderness, painful mass, obstruction, or 10% weight loss. e Moderat-to-severe disease applies to patients who have failed to respond to treatment for mild to moderate disease or those with more prominent symptoms of fevers, significant weight loss, abdominal pain or tenderness, intermittent nausea or vomiting (without obstructive findings), or significant anemia. f Severe-fulminant disease applies to patients with persistent symptoms despite the introduction of corticosteroids, or individuals presenting with high fever, persistent vomiting, evidence of intestinal obstruction, rebound tenderness, cachexia, or evidence of an abscess. disease, disease severity, health resource use in the year before enrollment, and medication use at enrollment (P.0001 for all) (Table 2). Patients who received infliximab, compared with patients who received other treatments only, were more likely to have both their ileum and colon involved (43.2% vs 35.4%, respectively), and to have moderate-to-severe or severe-to-fulminant disease at enrollment (33.3% vs 10.9%, respectively). Hospitalization in the year before enrollment for either surgical (17.5% vs 13.8%) or medical (14.4% vs 9.1%) management also was more frequent among infliximab-treated patients compared with patients who received other treatments only. Patients in the infliximab-treated group were more likely to be receiving prednisone (27.4% vs 16.1%), immunomodulators (49.4% vs 32.2%), and narcotic analgesics (9.8% vs 5.4%) at enrollment. Patients were followed-up for a mean of years (median, 1.77 y; range, y), and as of August 2004, approximately 79% of patients were participating actively in the registry. Patient discontinuations are outlined in Figure 1. Mortality Fifty-five (.87%) of the 6290 patients died. Of these 55 patients, 29 had received infliximab. Mortality rates were similar for patients who had received infliximab and those who had not:.53 per 100 patient-years vs.43 per 100 patient-years (relative risk [RR] [95% confidence interval (CI)] 1.24 [ ]; P.43). Similar rates were observed when patients were assessed by exposure to infliximab within the past 3 months. Sixteen patients died within 3 months of receiving an infliximab infusion (mortality rate per 100 patient-years,.41) compared with 39 who died who were not in this time frame (mortality rate,.50) (RR [95% CI].83 [ ]; P.54). Without adjusting for other factors, patients who died were older (odds ratio [OR], 1.07 for each 1-year increase in age; P.001) and had a longer duration of disease (OR, 1.06 for each 1-year increase in CD duration; P.001). When medications were analyzed without adjusting for other factors, the use of prednisone (OR, 2.13; P.007) and narcotic analgesics (OR,

5 May 2006 SAFETY WITH INFLIXIMAB: TREAT REGISTRY 625 Figure 1. Flow of participation in the TREAT Registry. Infiximabtreated patients are those patients who received infiximab within 12 weeks prior to enrollment, who were scheduled to receive infiximab within 30 days of enrollment, or who received infiximab at some other point in the registry. 1.84; P.044) were associated with increased mortality. In the adjusted model, only age (OR, 1.07; P.001), duration of CD (OR, 1.03; P.006), and use of prednisone (OR, 2.10; P.016) remained independent predictors of death. The use of infliximab was not a significant predictor of mortality in either the unadjusted or adjusted models (Table 3). The reported causes of death among patients are shown in a Table included in online supplementary materials (see Supplemental Table 1 at Serious Infection Among the 6253 patients who had adequate follow-up data for this analysis, 106 patients (1.70%) had a serious infection. Without adjusting for other factors, the rate of serious infections for the 69 patients who had received infliximab was significantly higher than that of 37 patients who were not treated with infliximab (1.37 per 100 patient-years vs.65; RR [95% CI], 2.15 [ ]; P.001). Without adjustment for other factors, there was a significantly higher rate of serious infection within 3 months of an infliximab infusion, that is, 1.41 per 100 patient-years among patients who had received infliximab in the prior 3 months compared with.77 per 100 patientyears among patients who had not received infliximab in the prior 3 months (RR [95% CI], 1.86 [1.269, 2.719]; P.001). Without adjusting for other factors, patients with a serious infection tended to be older (OR, 1.01; P.038), have a longer CD duration (OR, 1.03; P.001), and have moderate-severe CD (OR, 3.16 vs remission; P.001). Fewer serious infections were observed in patients with disease of either the ileum or colon, but not both. When medications were examined without adjusting for other factors, the use of prednisone (OR, 2.65; P.001) or narcotic analgesics (OR, 3.27; P.001) also was associated with an increase in serious infections. Although in the unadjusted analysis infliximab was associated with an increased risk for serious infections (OR, 1.49; P.045), on adjusting for other factors, the effect of infliximab on the risk for serious infection was not significant (OR,.99; P.97). Race (OR,.54 for Caucasian compared with non-caucasian, P.030), CD duration (OR, 1.02; P.011), moderateto-severe CD (OR, 2.11 vs remission; P.02), use of prednisone (OR, 2.21; P.001), and use of narcotic analgesics (OR, 2.38; P.001) were independent predictors of serious infection in the model (Table 5). Discussion This is the first report from the TREAT Registry, a large, prospective, observational research program designed to address the long-term safety of medications, including infliximab, for the treatment of CD. After adjustment for confounding factors including disease severity and the use of other medications, the risk for serious infection or death with infliximab use was similar to that observed with the use of conventional immunomodulators, and was not higher than the overall incidence of serious infections among all CD patients. The use of prednisone was a strong independent risk factor for both serious infection and death. Likewise, the use of narcotic analgesics also was associated with a significantly increased risk for serious infection. Colombel et al 15 reported the annual incidence of serious infections and mortality to be 2.1% and 1.3%, respectively, in a cohort of 500 CD patients treated with infliximab. A second series from Sweden, described by Ljung et al, 21 reported on the experience with 217 patients treated with infliximab. Eighteen serious infections (8.29%) were observed (11 occurred in CD patients) and 6 deaths (2.8% [2 deaths resulting from serious infection]) were reported during 28 months, resulting in an annual mortality rate of 1.2%. Although the mortality rate observed in these cohorts did not differ from published mortality rates in CD, these findings have raised some concerns. An important limitation of these published case series is the lack of a comparator group. Many of the serious infections occurred in patients with severe CD who also were receiving other immunosuppressive agents. The relatively small size of these studies precluded conducting multivariate analyses

6 626 LICHTENSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 5 Table 3. Predictors of Mortality Predictor variable All patients (N 6290) Deaths (N 53 a ) Unadjusted results Adjusted results OR (95% CI) c P value b OR (95% CI) c P value b Age Mean SD ( ) c ( ).001 Sex Male/unknown % Female %.825 ( ) ( ).28 Race Non-Caucasian/unknown % Caucasian % ( ) ( ).52 Diseased segment(s) Ileum and colon % Ileum only %.766 ( ) ( ).74 Colon only %.726 ( ) ( ).70 Unknown %.407 ( ) ( ).74 Years between diagnosis and enrollment Mean SD ( ) ( ).006 Severity at baseline Remission % Mild % ( ) ( ).57 Moderate/severe % ( ) ( ).083 Unknown % ( ) ( ).11 Infliximab No % Yes % ( ) ( ).96 Prednisone No % Yes % ( ) ( ).016 Immunomodulators No % Yes %.695 ( ) ( ).31 Narcotic analgesics No % Yes % ( ) ( ).074 a Data representing patient status between registration and June 30, 2004, were used in the logistic regression analysis because this is the timeframe for which there is complete information or medication usage. Two patients died after June 30, b P value from Wald 2 test. c OR reported as OR per year increase in age. to assess specific risk factors contributing to infection while controlling for confounding variables. Two controlled, 1-year, clinical trials of infliximab in 879 patients with moderately to severely active CD showed that the annual rate of serious infection ranged from 4.0% to 4.5%. 6,9 It is unclear if this rate is higher than expected in CD patients treated with other immunosuppressive agents because most of the patients in the control arms of the trials ultimately were exposed to infliximab. Present et al 26 evaluated data from 396 patients over 1800 patient-years of follow-up evaluation who were treated with 6-mercaptopurine and reported rates of significant infection to be 7.4%. The data from the TREAT Registry are consistent with the studies described previously. Results of univariate analysis suggest the occurrence of significantly more serious infections in patients who had received infliximab in the 3 months preceding the event. Unlike previous studies, the large number of patients in the TREAT Registry permitted multivariate analyses, the results of which indicated that the increased rate of infection likely is attributed to disease severity and the concomitant use of corticosteroids. This does not necessarily mean that the use of infliximab and/or conventional immunomodulators is not associated with any increased risk for serious infection. These are all immunosuppressive drugs and, thus, carry an increased risk for infection. This analysis allows the relative comparison of the risk associated with one immunosuppressive drug with that of another; however, this analysis does not allow for the assessment of risk associated with various combinations of therapies. Likewise, this analysis does not take into consideration contributing risk factors such as duration of use and cumulative exposure of a particular immunosuppressive

7 May 2006 SAFETY WITH INFLIXIMAB: TREAT REGISTRY 627 Table 5. Predictors of Serious Infection All patients Serious infections (N 6253) a (N 106) b Unadjusted results Adjusted results OR (95% CI) P value c OR (95% CI) P value c Age at enrollment Mean SD ( ) ( ).12 Sex Male/unknown % Female % ( ) ( ).35 Race Non-Caucasian/unknown % Caucasian %.674 ( ) ( ).030 Diseased segment(s) Ileum and colon % Ileum only %.606 ( ) ( ).15 Colon only %.594 ( ) ( ).17 Unknown %.188 ( ) ( ).35 Years between diagnosis and enrollment Mean SD ( ) ( ).011 Severity at baseline Remission % Mild % ( ) ( ).36 Moderate/severe % ( ) ( ).024 Unknown %.771 ( ) ( ).96 Infliximab d No % Yes % ( ) ( ).97 Prednisone d No % Yes % ( ) ( ).001 Immunomodulators d No % Yes %.963 ( ) ( ).24 Narcotic analgesics d No % Yes % ( ) ( ).001 a Of the 6290 patients enrolled in TREAT, 6253 provided adequate follow-up data to participate in this analysis. b Only the first occurrence of an event was considered for this analysis. c P value from Wald 2 test. d For those with a serious infection, medication ever used was obtained from all available 6-month periods (January June, July December) that occurred between enrollment and the period before the onset of the serious infection. Medication use during the period in which the infection occurred was not included in this analysis. agent or combination regimen. After adjusting for other factors, the odds ratio for serious infections with recent infliximab use was not shown to be increased in the TREAT Registry. Likewise, after adjustment for other factors, an increase in serious infections was not observed in patients who recently had received immunomodulators. The odds of a serious infection were increased significantly in patients with more severe CD, recent use of corticosteroids, and recent use of narcotics. A similar pattern was observed with regard to mortality. Although neither univariate nor multivariate analysis showed an increased risk for death in infliximab-treated patients in the TREAT Registry, a significantly increased risk of mortality was associated with recent use of corticosteroids. The finding that the use of infliximab is not associated with an increased risk for death is consistent with the Mayo Clinic and Swedish experiences, in which the reported annual mortality rates were similar to the mortality rates in previously published series of CD patients. 15,21 25 TREAT allows the risks associated with different CD medications to be interpreted in context. It appears that the risk for serious infections and mortality associated with infliximab therapy is similar to that of conventional immunomodulators. These data should be reassuring to physicians who use these drugs for the treatment of CD. Also, the risk for serious infection and death are increased significantly in patients recently taking corticosteroids. Although it is well known that corticosteroids are asso-

8 628 LICHTENSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 5 ciated with several complications (eg, osteoporosis, diabetes, and cardiovascular complications), the short-term risk of these drugs with regard to serious infections and mortality has not been well quantified. The data from this registry reinforce that risks should be considered when prescribing corticosteroids in the treatment of CD, especially with long-term use. Patients who require frequent corticosteroids to control their CD should be considered for steroid-sparing therapy with immunomodulators and/or infliximab. This is consistent with findings reported by Aberra et al 27 that indicated a higher risk for postoperative infections with the use of corticosteroids but not with the use of immunomodulators in patients with inflammatory bowel disease. Finally, the TREAT data indicate that patients receiving narcotic analgesics are also at greater risk for developing serious infections, perhaps because some of the signs and symptoms of infection are masked by these agents. It is not uncommon for narcotic analgesics to be used for CDassociated pain because nonsteroidal anti-inflammatory drugs are avoided in this population. 28,29 Physicians should instruct CD patients receiving narcotic analgesics to be particularly alert to signs and symptoms suggestive of infection because some of these symptoms may be less obvious in patients receiving these medications. The results from the TREAT Registry need to be interpreted with an understanding of both the strengths and limitations of patient registries. Although registries allow a large number of patients to be followed-up prospectively in a real-world setting, they are not monitored as rigorously as randomized controlled studies. It is possible that some SAEs occurred that were not captured in the registry; therefore, the rate of these events may be underestimated in the TREAT Registry. However, most analyses presented here represent comparisons of different CD treatment regimens, and there is no reason to suspect that reporting of SAEs would be different in one treatment group compared with another. In fact, because maintenance infusions associated with regular office visits are required with infliximab, it may be likely that more adverse events associated with infliximab would be reported. A second limitation of registries in this regard is that the population included may not be as sick as the general CD population and accordingly there may be underrepresentation of adverse events. To address this point, the rate of hospitalizations and surgeries in TREAT was compared with a general CD population (from the MedStat MarketScan insurance claims database; Medstat, Ann Arbor, MI). In this analysis, a higher percentage of TREAT vs MarketScan patients had at least 1 CD-related hospitalization and at least 1 surgery (age-adjusted rates for patients ages in 2003, 5.36% vs 5.15% and 3.95% vs 1.90%, respectively). These data indicate that substantial underreporting of adverse events in TREAT is unlikely. A second limitation of patient registries is the potential lack of follow-up evaluation. Although relatively few patients were lost to follow-up evaluation in this registry, a number of patients discontinued because they transferred their care to other physicians who were not participating in the registry. In many cases, patients had been required to change their treating physician as mandated by insurance plans. Although the majority of patients who discontinued from the registry provided a reason for withdrawal, a smaller subset did not. Unfortunately, additional follow-up evaluation on this issue could not be conducted once the patient withdrew consent to participate. For a similar reason, it is not known whether events occurred in this cohort after discontinuation from the registry. Overall, the attrition rate was similar to that observed in other clinical trials in CD, and the reasons for patient discontinuation were similar irrespective of treatment group. 6,9 It is notable that the proportions of patients discontinuing and lost to follow-up evaluation in TREAT were similar in both the infliximab and other-treatments-only cohorts. Taken together with the large overall sample size of the registry, it is unlikely that a disproportionate number of SAEs was missed in one group compared with the other. A potential limitation of these data is that only medication use during the registry or during the period immediately before enrollment was considered. It is possible that some patients categorized as other-treatmentsonly had in fact received infliximab in the more distant past. However, prior use of infliximab cannot be documented and analyzed in this registry owing to patientrecall bias. Furthermore, the timing and dose of infliximab administration, the concomitant medications at the time of infliximab administration, and any adverse events that may have occurred subsequently but before registry enrollment cannot be known. Incomplete data collection is unavoidable in patient registries. In any event, this limitation is not relevant to the analysis of serious infections presented here because these analyses assessed whether or not serious infections occurred within 3 months of an infliximab infusion. Because TREAT is an ongoing registry, further analyses will be conducted in the future to obtain a better understanding of the issues raised here. Analysis to examine risk factors for serious infection and mortality in greater detail (eg, dosage, length of time on medications, medication combinations, and so

9 May 2006 SAFETY WITH INFLIXIMAB: TREAT REGISTRY 629 forth) are needed, but to date the number of events in the registry is too small for multivariate analysis of these details. The TREAT Registry represents one of the largest databases of prospectively followed-up CD patients. In this regard, it is likely to be a valuable resource that can address a number of important questions about CD in the years to come. Conclusions Because CD is a chronic, potentially debilitating disease requiring ongoing treatment, physicians must consider the benefit/risk profiles of all therapies. Data from the TREAT Registry indicate that the risks for serious infection and death associated with infliximab are similar to those of conventional immunomodulators over 2 years of follow-up evaluation. Possible factors implicated with a significantly increased risk for serious infections include the use of prednisone, the use of narcotic analgesics, older age, and moderate or severe CD. Prednisone treatment was associated with a significantly increased risk for death. Although the TREAT data, when adjusted for confounding factors, have not indicated a relative increased risk for serious infection or death associated with infliximab, one should not completely disregard the possibility of increased risks with infliximab, alone or in combination with corticosteroids and/or immunomodulators, because the data does not yet allow a subanalysis of the role of duration of exposure and cumulative doses of different immunosuppressive drugs. Longer-term follow-up evaluation from this registry will provide additional insight in understanding the safety profile of infliximab and other therapies for the treatment of CD. Supplementary Data Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at References 1. Rutgeerts P, van Assche G, Vermeire S. Optimizing anti-tnf treatment in inflammatory bowel disease. Gastroenterology 2004; 126: Hanauer SB. Drug therapy: inflammatory bowel disease. N Engl J Med 1996;334: Podolsky DK. Inflammatory bowel disease. N Engl J Med 2002; 347: Targan SR, Hanauer SB, van Deventer SJH, et al, for the Crohn s Disease ca2 Study Group. A short-term study of chimeric monoclonal antibody ca2 to tumor necrosis factor for Crohn s disease. N Engl J Med 1997;337: Rutgeerts P, Feagan BF, Lichtenstein GR, et al. Comparison of scheduled and episodic treatment strategies of infliximab in Crohn s disease. Gastroenterology 2004;126: Hanauer SB, Feagan BG, Lichtenstein GR, et al. Maintenance infliximab for Crohn s disease: the ACCENT I randomized trial. Lancet 2002;359: Rutgeerts P, D Haens G, Targan S, et al. Efficacy and safety of retreatment with anti-tumor necrosis factor antibody (infliximab) to maintain remission in Crohn s disease. Gastroenterology 1999;117: Present DH, Rugeerts P, Targan S, et al. Infliximab for the treatment of fistulas in patients with Crohn s disease. N Engl J Med 1999;340: Sands BE, Anderson FH, Bernstein CN, et al. Infliximab maintenance therapy for fistulizing Crohn s disease. N Engl J Med 2004; 350: Stuck AE, Minder C, Frey FJ. Risk of infectious complication in patients taking glucocorticosteroids. Rev Infect Dis 1989;11: Yang YX, Lichtenstein GR. Corticosteroids in Crohn s disease. Am J Gastroenterol 2002;97: Sandborn WJ. A review of immune modifier therapy for inflammatory bowel disease: azathioprine, 6-mercaptopurine, cyclosporine, and methotrexate. Am J Gastroenterol 1996;91: Aucott JN. Glucocorticoids and infection. Endocrinol Metab Clin North Am 1994;23: Klein NC, Go CH, Cunha BA. Infections associated with steroid use. Infect Dis Clin North Am 2001;15: Colombel JF, Loftus EV, Tremaine WJ, et al. The safety profile of infliximab in patients with Crohn s disease: the Mayo Clinic experience in 500 patients. Gastroenterology 2004;126: Ricart E, Panaccione R, Loftus EV, et al. Infliximab for Crohn s disease in clinical practice at the Mayo Clinic: the first 100 patients. Am J Gastroenterol 2001;96: Cohen RD, Tsang JF, Hanauer SB. Infliximab in Crohn s disease: first anniversary clinical experience. Am J Gastroenterol 2000; 95: Farrell RJ, Shah SA, Lodhavia PJ, et al. Clinical experience with infliximab therapy in 100 patients with Crohn s disease. Am J Gastroenterol 2000;95: Hanauer SB, Sandborn W, and the Practice Parameters Committee of the American College of Gastroenterology. Management of Crohn s disease in adults. Am J Gastroenterol 2001;96: Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons, Inc., Ljung T, Karlén P, Schmidt D, et al. Infliximab in inflammatory bowel disease: clinical outcome in a population based cohort from Stockholm County. Gut 2004;53: Jess T, Winther KV, Munkhol P, et al. Mortality and causes of death in Crohn s disease: follow-up of a population based cohort in Copenhagen, Denmark. Gastroenterology 2002;122: Loftus EV, Silverstein MD, Sandborn WJ, et al. Crohn s disease in Olmstead County, Minnesota, : incidence, prevalence, and survival. Gastroenterology 1998;114: Farmer RG, Whelan G, Fazio VW. Long-term follow-up of patients with Crohn s disease. Relationship between clinical pattern and prognosis. Gastroenterology 1985;88: Ekbom A, Helmick CG, Zack M, et al. Survival and causes of death in patients with inflammatory-bowel disease: a populationbased study. Gastroenterology 1992;103: Present DH, Meltzer SJ, Krumholz MP, et al. 6-mercaptopurine in the management of inflammatory bowel disease: short- and longterm toxicity. Ann Intern Med 1989;111: Aberra FN, Lewis JD, Hass D, et al. Corticosteroids and immunomodulators: postoperative infectious complication risk in inflam-

10 630 LICHTENSTEIN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 5 matory bowel disease patients. Gastroenterology 2003;125: Edwards JT, Radford-Smith GL, Florin TH. Chronic narcotic use in inflammatory bowel disease patients: prevalence and clinical characteristics. J Gastroenterol Hepatol 2001;16: Kaplan MA, Korelitz BI. Narcotic dependence in inflammatory bowel disease. J Clin Gastroenterol 1988;10: Address requests for reprints to: Gary R. Lichtenstein, Gastroenterology Division, Hospital of the University of Pennsylvania, University of Pennsylvania School of Medicine, 3rd Floor Ravidin Building, 3400 Spruce Street, Philadelphia, Pennsylvania grl@uphs.upenn.edu; fax: (215) Gary R. Lichtenstein has received honorarium and research funding from Centocor, the manufacturers of infliximab; honorarium from Prometheus; honorarium from Salix, the manufacturers of azathioprine; and honorarium from Astra Zeneca, the manufacturers of Budesonide. Michelle L. Pritchard is employed by Ovatich Research Group, a company that receives funding from Centocor Inc. William J. Sandler has received research support and is a Consultant for Centocor, and has participated in CME events indirectly sponsored by Centocor. Bruce A. Salzberg is on the Speakers Bureau for Centocor. Robert H. Diamond is employed by Centocor. Donny M. Chen is a Consultant for Centocor. Russell D. Cohen has received research grants and is a consultant for Centocor, and is a stockholder in Johnson & Johnson.