Combination Immunomodulator and Antibiotic Treatment in Patients With Inflammatory Bowel Disease and Clostridium difficile Infection

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2009;7:981–987

Combination Immunomodulator and Antibiotic Treatment in Patients With Inflammatory Bowel Disease and Clostridium difficile Infection SHOMRON BEN-HORIN,* MAYA MARGALIT,‡ PETER BOSSUYT,§ JOCHEN MAUL,储 YAMI SHAPIRA,¶ DANIELA BOJIC,# IRIT CHERMESH,** AHMAD AL-RIFAI,‡‡ ALAIN SCHOEPFER,§§ MATTEO BOSANI,储 储 MATTHIEU ALLEZ,¶¶ PETER LASZLO LAKATOS,## FABRIZIO BOSSA,*** ALEXANDER ESER,‡‡‡ TOMMASO STEFANELLI,§§§ FRANCK CARBONNEL,储 储 储 KONSTANTINOS KATSANOS,¶¶¶ DAVIDE CHECCHIN,### INÉS SÁENZ DE MIERA,**** YEHUDA CHOWERS,** and GORDON WILLIAM MORAN,‡‡‡‡ FOR THE EUROPEAN CROHN’S AND COLITIS ORGANIZATION (ECCO) *Sheba Medical Center, Tel-Aviv University, Israel; ‡Hadassah Medical Center, Hebrew University, Jerusalem, Israel; §University of Leuven Hospitals, Leuven, Belgium; 储 Medizinische Klinik I, Campus Benjamin Franklin, Charité-Universitätsmedizin, Berlin, Germany; ¶Tel-Aviv Medical Center, Tel-Aviv, Israel; #Zvezdara University Medical Center, Belgrade, Serbia; **Rambam Medical Center, The Technion, Haifa, Israel; ‡‡Salford NHS University Hospital, Manchester, United Kingdom; §§University Hospital of Bern, Switzerland; 储 储Sacco University Hospital, Milan, Italy; ¶¶Saint-Louis Hospital, Paris, France; ##Semmelweis University, Budapest, Hungary; ***Casa Sollievo della Sofferenza Hospital, San Giovanni Rotondo, Italy; ‡‡‡Medical University of Vienna, Austria; §§§Università degli Studi di Milano, Italy; 储 储 储University Hospital of Besançon, France; ¶¶¶ Univeristy Hospital of Ioannina, Greece; ###University of Padua, Italy; ****Hospital General Yagüe, Burgos, Spain; and ‡‡‡‡University of Birmingham, United Kingdom

BACKGROUND & AIMS: Management of Clostridium difficile infection in patients with flaring inflammatory bowel disease (IBD) has not been optimized. We investigated the effects of combination therapy with antibiotics and immunomodulators in patients with IBD and C difficile infection. METHODS: We analyzed data from 155 patients (59% with ulcerative colitis [UC]) from a retrospective, European Crohn’s and Colitis organization, multi-center study comparing outcome of hospitalized IBD patients with C difficile infection who were treated with antibiotics (n ⫽ 51) or antibiotics and immunomodulators (n ⫽ 104). The primary composite outcome was death or colectomy within 3 months of admission, in-hospital megacolon, bowel perforation, hemodynamic shock, or respiratory failure. RESULTS: The primary outcome occurred in 12% of patients given the combination treatment vs none of the patients given antibiotics alone (P ⫽ .01). UC, abdominal tenderness, or severe bloody diarrhea was more common among patients that received the combined therapy. However, multivariate analysis revealed that only the combination therapy maintained a trend for an independent association with the primary outcome (likelihood ratio ⫽ 11.9; CI, 0.9 –157; P ⫽ .06). Treatment with 2 or 3 immunomodulators was correlated with the primary outcome, independent of disease severity at presentation (odds ratio [OR] ⫽ 17; CI, 3.2–91; P ⬍ .01). Acid-suppressing medications increased the risk of C difficile relapse (OR ⫽ 3.8; CI, 1.1–12.9; P ⫽ .03), whereas recent hospitalization correlated with increased rate of C difficile persistence (OR ⫽ 8; CI, 2.1–29; P ⫽ .002). CONCLUSIONS: Patients with IBD that also have C difficile infection are frequently treated with a combination of antibiotics and immunomodulators. However, this combination tends to associate with a worse outcome than antibiotic therapy alone. Prospective controlled trials are urgently needed to optimize the management of these challenging patients.

C

lostridium difficile infection (CDI) can cause life-threatening colitis in patients with inflammatory bowel disease (IBD), as well as in other populations.1 It has been shown that the incidence of CDI in hospitalized IBD patients has increased

compared with incidence in control populations,2,3 possibly due to an altered immune status caused by immunomodulator (IM) therapy,1 more frequent contact with medical facilities, and predisposing antibiotic therapy. Similar to other enteric pathogens, CDI can present with abdominal symptoms indistinguishable from an exacerbation of IBD. Therefore, routine screening for CDI has been advocated for any flaring IBD patient.4,5 The importance of considering this diagnosis in flaring IBD patients has been underscored by recent evidence that hospitalization with a dual diagnosis of IBD and CDI is associated with a greater mortality compared with patients with IBD or CDI alone.3,6 The mainstay of treatment of CDI is antibiotics, most commonly consisting of metronidazole or vancomycin.7 Nevertheless, when faced with a laboratory test result indicating CDI in a flaring IBD patient, some clinicians elect to administer IMs conjointly with antibiotics, suspecting that at least part of the patient’s symptoms are due to exacerbation of underlying IBD. Conversely, other clinicians tend to treat with antibiotics alone and to withhold all IMs, fearing that further immune suppression will worsen CDI outcome. This unresolved management decision is particularly important given the significant clinical burden and potential grave consequences of CDI in IBD patients. However, there are no studies specifically addressing this clinical dilemma in the setting of a flaring IBD patient found to be inflicted with C difficile. Therefore, the primary aim of the present study was to investigate the clinical outcome of CDI-associated IBD in hospitalized patients treated with antibiotics alone, compared with patients in whom antibiotics and IM were coadministered. As a secondary goal, the study also aimed to identify other potential risk factors for an adverse outcome in these patients. Abbreviations used in this paper: CDI, Clostridium difficile infection; CI, confidence interval; IBD, inflammatory bowel disease; IM, immunomodulator; OR, odds ratio; PPI, proton pump inhibitor; UC, ulcerative colitis. © 2009 by the AGA Institute 1542-3565/09/$36.00 doi:10.1016/j.cgh.2009.05.031

982

BEN-HORIN ET AL

Methods Study Design and Population This was a European Crohn’s & Colitis Organization (ECCO) multicenter retrospective study. All members of Y-ECCO (the Young branch of ECCO), as well as several senior ECCO members, were contacted and were offered to participate. The study population was defined to be all patients hospitalized with a co-diagnosis of IBD and concomitant CDI from January 1, 2000, to March 1, 2008 in the participating centers. CDI-IBD cases were identified by a systematic search of discharge diagnoses of admitted patients and of the bacteriology laboratory database. In addition, in some centers, physicians of the Gastroenterology department were also actively questioned about past patients with CDI-IBD. To be included in the study, patients had to have an established diagnosis of Crohn’s disease, ulcerative colitis (UC), or colonic IBD unclassified according to accepted clinical radiological and histopathological criteria, for at least 3 months prior to admission. Patients also had to have at least 3 months of follow-up to be included in the study. CDI was diagnosed if any of the following were found positive: toxin AB assay (or toxin A assay alone), bacterial culture of C difficile, cell culture toxicity assay, or histological evidence of pseudo membranes compatible with CDI. The study was approved centrally by the ethics committee of the Sheba Medical Center, and was also approved or exempted locally by the respective ethics committee at each of the participating centers.

Study Definitions and Outcomes Patients who were treated with antibiotics and immunomodulators during the hospitalization with CDI-IBD were designated the AB⫹IM group, whereas patients who were treated solely by antibiotics were defined as the AB group. IM treatments were defined as any of the following: corticosteroids at a dose equal to or above 20 mg of prednisone daily, thiopurines at any dose, methotrexate, cyclosporine, tacrolimus, or biologics of any kind. Topical per rectal treatments were not considered as IM. While IMs may have different pharmacoimmunokinetics, we chose the hospitalization period as the most clinically relevant time frame by which to define the administration of IMs or their lack of administration. Thus, patients were allocated to the AB-IM group if their IM medications were carried on from their regular medication, intensified in dose, or if IMs were initiated at admission and given for 3 days or more during hospitalization. Patients were allocated to the AB group if they had not received IMs during hospitalization, regardless of whether IMs were discontinued at CDI diagnosis or not initiated. Additional parameters previously reported to affect prognosis in C difficile colitis were also assessed. These included the concurrent use of proton pump inhibitors (PPIs) or H2 blockers, advanced age, choice of metronidazole over vancomycin, CDI after the year 2003 (possibly corresponding to emergence of a super pathogenic strain), elevated baseline creatinine, leukocytosis, or hypoalbuminemia ⬍2.5 mg/dL.8 –11 The primary outcome was a composite comprised of the following parameters: mortality or colectomy within 3 months of admission with CDI and IBD (in-hospital or after discharge), colon perforation or megacolon during hospitalization, shock (defined by the need for intravenous vasopressive support), or the need for mechanical ventilation.

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 9

Secondary outcomes were: the duration of hospitalization, the need to substitute antibiotics due to CDI persistence and lack of response, CDI relapse defined as rehospitalization within 3 months with an admission diagnosis of CDI, or colectomy and/or mortality at 12 months follow-up. The AB⫹IM group was compared with the AB group with respect to the study primary and secondary outcomes.

Statistical Analysis Continuous variables were analyzed by 2-tailed Student t test or Mann–Whitney test, as appropriate, and categorical variables were analyzed by Fisher exact test. Correlations were tested by Spearman rank correlation. Unless noted otherwise, variables differing between groups with a significance level of P ⬍ .1 were then entered into multivariate analysis models. These consisted of a multiple backward logistic regression model to identify factors independently affecting dichotomous clinical outcomes, and backward linear regression model for continuous clinical outcomes (ie, the duration of hospitalization). Near complete separation of data designates a situation when no events occur in 1 of the compared groups, thereby causing the equation denominator to be 0 and rendering logistic regression not valid. In this case, penalized likelihood ratios (bias-reduced logistic regression) were computed according to Firth’s method.12 A power analysis was done to estimate the sample size needed. Because of the lack of previous data regarding the effect size of the 2 treatment modalities (AB⫹IM versus AB alone), an interim analysis was conducted after 50 patients were identified. Based on this interim analysis, a sample size of 90 patients (45 in each arm) was computed to be required in order to detect a 15% difference in the occurrence of the composite outcome with a power of 80% and with ␣ level of 5%. All statistics were performed using MedCalc software (Mariakerke, Belgium), except for the bias-reduced logistic regression which was performed using the Wessa.net open code software.13 P ⬍ .05 was considered significant.

Results Characteristics of Study Population One hundred fifty-nine patients were identified in 20 centers across Europe and Israel. Four patients were excluded due to insufficient data or uncertain diagnosis. Thus, there remained 155 patients who comprised the study cohort. All but 5 of these patients were admitted due to exacerbation of gastrointestinal symptoms with initial presumed diagnosis of worsening IBD. The 5 remaining patients were hospitalized for other reasons when diarrhea and/or other gastrointestinal symptoms appeared. Several methods were used for CDI diagnosis in the study patients, the most widely used being toxin AB assay (Figure 1). Interestingly, histopathologic examination detected pseudomembranes compatible with CDI in only 4 out of the 77 patients (5%) in whom colonic biopsies were obtained. Altogether, 104 patients (67%) were treated by concurrent AB⫹IM during hospitalization, whereas 51 (33%) received AB alone. Demographic and clinical characteristics of the study population are shown in Table 1. Clinical parameters pertaining to the index hospitalization with CDI and IBD are shown in Table 2.

September 2009

COMBINATION TREATMENT OF IBD AND C DIFFICILE 983

Figure 1. The rate of use for the different modes of Clostridium difficile infection diagnosis in the study population.

Primary Adverse Outcome Twelve out of the 155 patients met an adverse outcome compatible with the primary composite outcome (some had more than 1 adverse outcome). Specifically, 9 patients had colectomy within 3 months (8 during the index hospitalization), and 2 patients died (1 during hospitalization). Five patients had megacolon and all underwent in hospital colectomy. Two of the 5 also suffered from bowel perforation. Three patients experienced hemodynamic shock or respiratory failure. All primary adverse outcomes occurred in patients in the AB⫹IM treatment group (P ⫽ .01 for the comparison to the AB group; Table 2). However, more patients in the AB⫹IM group had UC, and they tended to present more commonly with severe diarrhea, abdominal tenderness, or blood in stool compared with the AB group (Table 2). In addition, more patients in the AB⫹IM group had received maintenance IM, compared

with the AB group. In order to at least partially support or refute a confounding effect by these disparities between the 2 treatment groups, 2 additional analyses were undertaken. First, we performed a univariate comparison of the clinical parameters of disease severity at presentation among patients with or without an adverse primary outcome (Table 3). This analysis showed that despite a more severe presentation among the AB⫹IM treated patients as a group, the group with the adverse outcome did not differ from the rest of the cohort in the various parameters of severity of presentation. Moreover, on multivariate analysis, none of the parameters observed more frequently among the AB⫹IM group was found to affect the primary outcome, whereas treatment with AB⫹IM still maintained a trend for an independent adverse impact on the primary outcome (Table 4). A separate analysis showed that most of this risk was attributable to treatment with 2 or 3 IMs, in a manner that was statistically independent of clinical parameters of disease severity at presentation (Table 5). In contrast to the effect of double/triple IM, the association of cyclosporine treatment per se with the adverse outcome lost its statistical significance on multivariate analysis (odds ratio [OR], 5.4; 95% confidence interval [CI], 0.8 –37; P ⫽ .08). Finally, as an additional measure to exclude a possible confounding effect of the severity of underlying chronic IBD, the primary outcome was reanalyzed only for the subgroup of 102 patients who were not receiving maintenance IM prior to hospitalization. The adverse primary outcome was experienced by 8/56 patients (14%) in the AB⫹IM group versus 0/46 patients in the AB group (P ⫽ .008; Fisher exact test). Multivariate analysis performed selectively for this subgroup of patients who had not received maintenance IM prior to admission, showed that AB⫹IM treatment and the presence of severe diarrhea were the only 2 variables exhibiting

Table 1. Background Demographic and Clinical Characteristics of the Study Population and the Breakdown for the 2 Treatment Groups

Age (y, median ⫾ SD) Females UC CD Colonic IBD unclassified Extent of disease, UCa Pan/extensive colitis Left-sided colitis Proctitis Extent of disease, CDa Colonic Ileocolonic Ileal CD phenotypea Inflammatory Penetrating Fibrostenoting Duration of disease (y) Hospitalization in prior 3 months AB in prior 3 months PPI/H2B in prior month Any co-morbidities

All patients (n ⫽ 155)

AB⫹IM patients (n ⫽ 104)

AB patients (n ⫽ 51)

P value

43 ⫾ 18 90 (58%) 92 (59%) 61 (39%) 2 (1%)

41 ⫾ 17 55 (53%) 68 (65%) 35 (34%) 1 (1%)

45 ⫾ 20 35 (69%) 24 (47%) 26 (51%) 1 (2%)

.3 .08 .04 .04 ⬎.9

47 (52%) 40 (48%) 2 (2%)

38 (57%) 27 (40%) 2 (3%)

9 (41%) 13 (59%) 0

.3 .3 .3

25 (41%) 16 (26%) 20 (33%)

17 (48%) 9 (26%) 9 (26%)

8 (31%) 7 (27%) 11 (42%)

.3 .3 .3

40 (66%) 11 (19%) 8 (13%) 6.6 ⫾ 7 39 (25%) 59 (41%) 33 (23%) 59 (37%)

24 (73%) 4 (12%) 5 (15%) 6.5 ⫾ 8 24 (23%) 37 (37%) 21 (22%) 39 (36%)

16 (62%) 7 (27%) 3 (11%) 6.2 ⫾ 8 15 (29%) 22 (47%) 12 (26%) 20 (38%)

.4 .4 .4 .7 .4 .5 .5 .8

NOTE. The P values are for the comparison of the AB group to the AB⫹IM group. AB, antibiotic; CD, Crohn’s disease; H2B, H2 blockers. aPercentages are out of the number of patients with either UC or CD infection, respectively.

984

BEN-HORIN ET AL

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 9

Table 2. Clinical Parameters Pertaining to the Index Hospitalization of the Study Population With CDI and IBD

Severe diarrhea (⬎6 BM/d) Blood in stool Abdominal pain Abdominal tenderness Fever (⬎37.5°C) Leukocytosis (⬎16 K/␮L) Anemia (Hb ⬍10.5 g/dL) Elevated CRP or ESR ⬎30 mm/h Hypoalbuminemia (⬍2.5 mg/dL) Elevated creatinine Endoscopy done Endoscopic pseudomembranesa Histological pseudomembranesa Initial treatment metronidazole Index admission after 2003 Concurrent PPI/H2B Time to CDI diagnosis (d)b Outcomes Primary adverse outcome Secondary outcomes CDI relapse CDI antibiotic failure Hospitalization duration (d) Death/colectomy within 1 year

All patients (n ⫽ 155)

AB⫹IM patients (n ⫽ 104)

AB patients (n ⫽ 51)

P value

97 (66%) 101 (66%) 97 (63%) 85 (56%) 58 (41%) 17 (7%) 44 (30%) 120 (89%) 10 (8%) 27 (18%) 92 (60%) 10 (11%) 4 (3%) 119 (78%) 105 (69%) 56 (39%) 3 ⫾ 3.9

73 (73%) 74 (71%) 70 (68%) 63 (63%) 38 (41%) 10 (10%) 30 (29%) 83 (90%) 5 (6%) 20 (20%) 67 (64%) 9 (13%) 4 (4%) 79 (76%) 74 (71%) 40 (42%) 2 ⫾ 3.3

24 (52%) 27 (54%) 27 (53%) 22 (43%) 20 (42%) 7 (14%) 14 (39%) 36 (84%) 5 (12%) 7 (15%) 25 (52%) 1 (4%) 0 (0%) 40 (80%) 31 (62%) 16 (33%) 3 ⫾ 4.9

.01 .04 .11 .04 ⬎.9 .5 ⬎.9 .4 .2 .5 .15 .2 .3 .7 .2 .4 .15

12 (8%)

12 (12%)

0 (0%)

.01

17 (11%) 15 (10%) 15 ⫾ 13 2 (1%)

12 (12%) 7 (7%) 15 ⫾ 14 2 (2%)

5 (10%) 8 (16%) 14 ⫾ 11 0 (0%)

.8 .09 .9 .5

AB, antibiotic; BM, bowel movement; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; H2B, H2 blockers; Hb, hemoglobin. aPercentage noted is out of the number of endoscopies done. bDenotes median number of days (⫾SD) since admission until CDI diagnosis.

a trend for independent association with the primary outcome (Table 6).

Secondary Outcomes Seventeen out of 155 patients (11%) had CDI relapse within 3 months of admission. The only parameter associated with an increased risk of relapse was the use of PPI or H2 blockers during hospitalization (OR, 3.8; 95% CI, 1.1–12.9; P ⫽ .03). The median duration of hospitalization in the study cohort was 14.7 ⫾ 13 days. The factors found to be predictive of a more prolonged hospitalization were the presence of anemia (r ⫽ 0.23; P ⫽ .02) or significant hypoalbuminemia ⬍2.5 mg/dL on presentation (r ⫽ 0.22; P ⫽ .01). As evident from the respective r values, the strength of the correlation was weak due to significant overlap between the groups. Most patients (77%) were treated by metronidazole as their initial CDI-directed antibiotic, while 18% were treated by vancomycin or vancomycin plus metronidazole, and 8 patients (5%) did not receive an antibiotic. Persistent CDI led to a change of the antibiotic regimen in 13/119 (11%) of metronidazole-treated and in 2/27 (7%) of vancomycin-treated patients (P ⫽ not significant). Prior hospitalization within 3 months of admission was the only parameter predictive of the need for antibiotic change due to persistent C difficile (OR, 8; 95% CI, 2.1–29; P ⫽ .002). There were too few events for evaluation of the last secondary outcome, namely, death/colectomy beyond 3 months from admission.

Discussion The present study examined the clinical course of hospitalized IBD patients found to be inflicted with CDI. To the

best of our knowledge, this is the first study to specifically address the prevalence of coadministration of antibiotics plus IM to these patients, and to investigate the outcome of this management approach. The first observation of this study is indeed that the majority of patients with CDI-associated IBD flare-up is treated by a combination of IM and antibiotic by their physicians, despite the absence of evidence to support this management approach. The design of the study precluded the ability to inquire about physicians’ considerations in choosing IM plus antibiotic combination over antibiotic alone. Some, possibly relying on prior works, may interpret the positive assay for C difficile as merely indicating a carrier state,14 and may therefore attribute the worsening symptoms to exacerbation of underlying IBD. As previously reported1 and reproduced in our study, most episodes of CDI in IBD presented with bloody diarrhea and in the absence of endoscopically-detectable pseudo membranes. This may augment the tendency to ascribe the symptoms to underlying IBD rather than to superimposed CDI. Alternatively, a perceived benefit of cotreatment with IMs plus antibiotics may be rationally deduced by analogy to other infectious disorders in which coadministration of corticosteroids with antibiotics has proven superior to antibiotics alone, as for instance, in bacterial meningitis, tuberculous meningitis, and other disorders.15–18 The paradoxical superiority of such combination therapy may be related to the amelioration of infection-associated inflammatory processes in the tissue through various mechanisms. Whether corticosteroids may propagate such beneficial mechanisms in CDI has been investigated in 2 animal studies, which yielded conflicting results.19,20

September 2009

COMBINATION TREATMENT OF IBD AND C DIFFICILE

985

Table 3. Univariate Analysis Comparing the Clinical Parameters of Patients With Adverse Primary Outcome to Those Without (Benign Outcome Group)

Age (y, median ⫾ SD) Females UC CD Duration of disease (y) Hospitalization in prior 3 months AB in prior 3 months Concurrent PPI/H2B Any comorbidities Index admission after 2003 Time to CDI diagnosis (d) Severe diarrhea (⬎6 BM/d) Blood in stool Abdominal pain Fever (⬎37.5°C) Abdominal tenderness Leukocytosis (⬎16 K/␮L) Anemia (Hb ⬍10.5 g/dL) Elevated CRP or ESR ⬎30 mm/h Hypoalbuminemia (⬍2.5 mg/dL) Elevated creatinine Endoscopic pseudomembranesb Treated with AB⫹IM Corticosteroidsc Thiopurines/MTXc Cyclosporinec Anti-TNFc Treated with a single IM Treated with double/triple IM Initial treatment metronidazole CDI relapse CDI antibiotic failure Hospitalization duration (d)

Benign outcome patients (n ⫽ 143)

Adverse outcome patients (n ⫽ 12)

P value

42 ⫾ 18 84 (59%) 83 (58%) 58 (40%) 6.8 ⫾ 8 37 (26%) 53 (39%) 40 (42%) 52 (36%) 97 (68%) 3 ⫾ 3.8 89 (66%) 91 (64%) 90 (63%) 54 (41%) 79 (56%) 14 (11%) 39 (28%) 110 (87%) 8 (7%) 24 (18%) 10 (12%) 92 (64%) 79 (86%)a 26 (28%)a 6 (6%)a 7 (8%)a 67 (73%)a 25 (27%)a 110 (77%) 14 (10%) 13 (9%) 13 ⫾ 9

45 ⫾ 18 6 (50%) 9 (75%) 3 (25%) 4.1 ⫾ 5.3 2 (16%) 5 (52%) 16 (33%) 6 (50%) 8 (66%) 1 ⫾ 4.9 8 (75%) 10 (83%) 7 (64%) 4 (44%) 6 (55%) 3 (25%) 5 (41%) 10 (91%) 2 (20%) 3 (25%) 0 12 (100%) 11 (92%) 5 (42%) 4 (33%) 3 (25%) 4 (33%) 8 (66%) 9 (75%) 3 (25%) 2 (16%) 35 ⫾ 26

.8 .7 .3 .3 .2 .5 ⬎.9 .4 .5 .5 .8 ⬎.9 .2 ⬎.9 ⬎.9 ⬎.9 .13 .3 ⬎.9 .2 .5 ⬎.9 .01 .7 .5 .01 .09 ⬍.01 ⬍.01 ⬎.9 .13 .3 .001

AB, antibiotic; BM, bowel movement; CD, Crohn’s disease; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; H2B, H2 blockers; Hb, hemoglobin; MTX, methotrexate; TNF, tumor necrosis factor. aPercentage is of the 92 patients in the benign outcome group who were treated with AB⫹IM. bPercentage noted is out of the number of endoscopies done. cPercent of patients sums to more than 100% because some patients received more than 1 IM.

Although an increased incidence of CDI among IBD patients receiving IM was recently reported,1 the impact of ongoing IM therapy on the outcome of CDI-associated IBD is yet undefined. In several studies of other patient populations, immunosuppression was not associated with mortality or with a more severe CDI colitis compared with matched controls without IM treatment,5,21,22 although steroids did increase the risk of relapse in 1 of these works.21 Moreover, a single case report of a dramatic response to corticosteroids in a 5-year-old child with severe nonresponsive CDI further underscores this therapeutic dilemma.23 In contrast, in other series, immunosuppression predisposed to development of severe or fulminant CDI in hospitalized patients,24,25 and was found to be associated with increased risk for 30 days mortality in patients with severe CDI admitted to an intensive care unit.26 Our results agree with the later works, by incriminating antibiotic plus IM as possibly related to a worse outcome compared with antibiotic alone in flaring IBD patients with concomitant C difficile. Arguably, this adverse outcome could merely stem from more severe flares among the patients who

were treated with IM. Nevertheless, the patients with adverse outcome were not found to have more severe presentation compared with those with a favorable outcome, as judged by an array of acceptable disease severity parameters (Table 3). Moreover, in multivariate analyses, only the AB⫹IM treatment arm showed a strong trend for independent association with the adverse outcome, and a more profound immunosuppression by multi-IM regimen showed a statistically significant independent association with this outcome (Tables 4 and 5). Collectively, these findings may therefore suggest a trend toward a detrimental effect of the combination antibiotic plus IM treatment in these patients, especially when more than one IM is used. However, prospective controlled studies validating these observations are mandatory before management recommendations can be endorsed. Interestingly, in 3 out of the 9 colectomized patients in the present cohort, pseudo membranes indicating ongoing CDI were unexpectedly found on histology of the resected specimen. This may indicate that in at least some of the nonresponsive patients, relentless infection may be the culprit responsible for the adverse outcome. Conversely, these data can be interpreted

986

BEN-HORIN ET AL

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 7, No. 9

Table 4. Parameters Entered Into the Multivariate Bias-Reduced Logistic Regression Model for the Whole Cohort, and the Respective Results of Their Independent Effect on the Occurrence of the Primary Adverse Outcome Parameter

Likelihood ratio

Confidence interval

P value

AB⫹IM treatment Female sex Having UC Severe diarrhea (⬎6 BM/d) Blood in stool Abdominal tenderness

11.9 0.9 1 0.5 1.6 0.8

0.9–157 0.3–3.1 0.2–4.3 0.13–1.65 0.3–8 0.2–2.8

.06 .9 .9 .2 .6 .7

AB, antibiotic; BM, bowel movement.

Table 6. A Multivariate Analysis of the Independent Parameters’ Effect on the Occurrence of the Primary Adverse Outcome Among Patients Who Were Not Treated With Immunomodulators Prior to the Index Hospitalization (n ⫽ 102) Parameter

Odds ratio

Confidence interval

P value

Female sex Having UC Hypoalbuminemia (⬍2.5 mg/dL) Abdominal pain Severe diarrhea (⬎6 BM/d) Blood in stool Abdominal tenderness IM⫹AB treatment

1.05 1.4 1.5 1.3 4.8 4.1 1.4 10.1

0.9–1.2 0.8–2 0.2–10 0.2–8 0.9–26 0.3–22 0.2–9 0.8–330

.3 .7 .7 .7 .07 .3 .6 .08

AB, antibiotic; BM, bowel movement.

as suggesting that most adverse outcomes (ie, the remaining 6 out of 9 colectomies) were due to unresponsive underlying IBD. Further works are pertinent in order to elucidate whether the main cause of unfavorable outcome in these patients is a relentless infectious process, uncontrolled IBD with mere colonization of C difficile, or the interplay between the 2 etiologies. Several other factors have been previously shown to be associated with a worse outcome of CDI in non-IBD populations.8 –11 Our findings replicated some of these previous reports by showing that acid-suppressing medications are associated with increased CDI relapse risk, and that hypoalbuminemia and anemia correlate with a more prolonged hospitalization. However, none of these factors were predictive of the more stringent and grave events comprising the composite primary outcome in our study population. There are several limitations to our study. Although most patients were identified by a systematic search of microbiology and clinical databases, it is impossible to exclude biased identification of some patients. In addition, despite the finding that the coadministration of IM with antibiotics was associated with a worse outcome, the retrospective nature of our study precludes the ascertainment of causality. As such, a definite cause and effect can only be determined by a prospective blinded controlled study comparing antibiotic alone versus antibiotic plus IM regimens. Nevertheless, the present study may also be valuable for providing indispensable data required for the rational design of such future prospective study.

Table 5. Parameters Entered Into the Multivariate Logistic Regression Model to Examine the Independent Effect on the Primary Outcome of Double/Triple IM Versus a Single IM Within the AB⫹IM Group Parameter

Odds ratio

Confidence interval

P value

Double or triple IM treatment Having UC Hypoalbuminemia (⬍2.5 mg/dL) Leukocytosis (⬎16 K/␮L) Severe diarrhea (⬎6 BM/d) Blood in stool Abdominal tenderness

17.1 0.9 8.1 4 0.4 3.4 0.4

3.2–91 0.1–6 0.93–71 0.5–27 0.1–2 0.4–37 0.07–2

⬍.01 .9 .06 .3 .4 .2 .2

AB, antibiotic; BM, bowel movement.

Another limitation of this study is due to possible inaccuracies of CDI diagnosis. In most of the participating centers CDI diagnosis relied on enzyme immunoassay-based assays for toxin presence in stool (Figure 1), which are considered to be diagnostically inferior to cell cytotoxicity assays.27 However, recent works documented high accuracy of several commercially-available assays when compared with cell cytotoxicity assays.28,29 Moreover, these assays remain the most commonly used method for diagnosis of CDI in clinical practice.27 Thus, we believe the present results reflect the outcome of patients as treated in real life situations of CDI-associated IBD flare-up, and therefore provide valuable insight on the best management approach for these patients. In conclusion, many physicians coadminister antibiotics plus IMs to patients with CDI-associated IBD exacerbation. This cotreatment may tend to be associated with a worse clinical outcome compared with administration of antibiotics alone. However, prospective studies are direly needed in order to validate these observations and to elucidate the best management approach for flaring IBD patients with concomitant CDI. References 1. Issa M, Vijayapal A, Graham MB, et al. Impact of Clostridium difficile on inflammatory bowel disease. Clin Gastroenterol Hepatol 2007;5:345–351. 2. Rodemann JF, Dubberke ER, Reske KA, et al. Incidence of Clostridium difficile infection in inflammatory bowel disease. Clin Gastroenterol Hepatol 2007;5:339 –344. 3. Nguyen GC, Kaplan GG, Harris ML, et al. A national survey of the prevalence and impact of Clostridium difficile infection among hospitalized inflammatory bowel disease patients. Am J Gastroenterol 2008;103:1443–1450. 4. Meyer AM, Ramzan NN, Loftus EV Jr, et al. The diagnostic yield of stool pathogen studies during relapses of inflammatory bowel disease. J Clin Gastroenterol 2004;38:772–775. 5. Mylonaki M, Langmead L, Pantes A, et al. Enteric infection in relapse of inflammatory bowel disease: importance of microbiological examination of stool. Eur J Gastroenterol Hepatol 2004; 16:775–778. 6. Ananthakrishnan AN, McGinley EL, Binion DG. Excess hospitalization burden associated with Clostridium difficile in patients with Inflammatory Bowel Disease. Gut 2008;57:205–210. 7. McFarland LV. Update on the changing epidemiology of Clostrid-

September 2009

8.

9.

10.

11.

12. 13.

14.

15.

16.

17.

18.

19.

20. 21.

ium difficile-associated disease. Nat Clin Pract Gastroenterol Hepatol 2008;5:40 – 48. Cadle RM, Mansouri MD, Logan N, et al. Association of protonpump inhibitors with outcomes in Clostridium difficile colitis. Am J Health Syst Pharm 2007;64:2359 –2363. Ramaswamy R, Grover H, Corpuz M, et al. Prognostic criteria in Clostridium difficile colitis. Am J Gastroenterol 1996;91:460 – 464. Greenstein AJ, Byrn JC, Zhang LP, et al. Risk factors for the development of fulminant Clostridium difficile colitis. Surgery 2008;143:623– 629. Pépin J, Valiquette L, Gagnon S, et al. Outcomes of Clostridium difficile-associated disease treated with metronidazole or vancomycin before and after the emergence of NAP1/027. Am J Gastroenterol 2007;102:2781–2788. Heinze G, Schemper M. A solution to the problem of separation in logistic regression. Stat Med 2002;21:2409 –2419. Wessa P. Bias Reduced Logistic Regression (v1.0.2) in Free Statistics Software (v1.1.23-r2), Office for Research Development and Education. Available at: http://www.wessa.net/rwasp_ logisticregression.was/. Accessed February 10, 2009. Keighley MR, Youngs D, Johnson M, et al. Clostridium difficile toxin in acute diarrhoea complicating inflammatory bowel disease. Gut 1982;23:410 – 414. van de Beek D, de Gans J, McIntyre P, et al. Corticosteroids for acute bacterial meningitis. Cochrane Database Syst Rev 2007; CD004405. Thwaites GE, Nguyen DB, Nguyen HD, et al. Dexamethasone for the treatment of tuberculous meningitis in adolescents and adults. N Engl J Med 2004;351:1741–1745. Mayosi BM, Ntsekhe M, Volmink JA, et al. Interventions for treating tuberculous pericarditis. Cochrane Database Syst Rev 2002;CD000526. Hato N, Yamada H, Kohno H, et al. Valacyclovir and prednisolone treatment for Bell’s palsy: a multicenter, randomized, placebocontrolled study. Otol Neurotol 2007;28:408 – 413. Castagliuolo I, Karalis K, Valenick L, et al. Endogenous corticosteroids modulate Clostridium difficile toxin A-induced enteritis in rats. Am J Physiol Gastrointest Liver Physiol 2001;280:G539 – G545. Bartlett JG. Treatment of antibiotic-associated pseudomembranous colitis. Rev Infect Dis 1984;6 Suppl 1:S235–S241. Gellad ZF, Alexander BD, Liu JK, et al. Severity of Clostridium difficile-associated diarrhea in solid organ transplant patients. Transpl Infect Dis 2007;9:276 –280.

COMBINATION TREATMENT OF IBD AND C DIFFICILE

987

22. Hardt C, Berns T, Treder W, et al. Univariate and multivariate analysis of risk factors for severe Clostridium difficile-associated diarrhoea: importance of co-morbidity and serum C-reactive protein. World J Gastroenterol 2008;14:4338 – 4341. 23. Cavagnaro C, Berezin S, Medow MS. Corticosteroid treatment of severe, non-responsive Clostridium difficile induced colitis. Arch Dis Child 2003;88:342–344. 24. Rubin MS, Bodenstein LE, Kent KC. Severe Clostridium difficile colitis. Dis Colon Rectum 1995;38:350 –354. 25. Dallal RM, Harbrecht BG, Boujoukas AJ, et al. Fulminant Clostridium difficile: an underappreciated and increasing cause of death and complications. Ann Surg 2002;235:363–367. 26. Lamontagne F, Labbé AC, Haeck O, et al. Impact of emergency colectomy on survival of patients with fulminant Clostridium difficile colitis during an epidemic caused by a hypervirulent strain. Ann Surg 2007;245:267–272. 27. Barlett JG. Clostridium difficile– old and new observations. J Clin Gastroenterol 2007;41:s24 –s29. 28. Miendje Deyi VY, Vandenberg O, Mascart G, et al. Diagnostic value of five commercial tests for the rapid diagnosis of Clostridium difficile-associated disease. Clin Lab 2008;54:9 –13. 29. Musher DM, Manhas A, Jain P, et al. Detection of Clostridium difficile toxin: comparison of enzyme immunoassay results with results obtained by cytotoxicity assay. J Clin Microbiol 2007;45: 2737–2739.

Reprint requests Shomron Ben-Horin, MD, Gastroenterology Department, Sheba Medical Center, Tel Hashomer 52621, Israel. e-mail: [email protected]; fax: (972) 3-5303160. Acknowledgments The authors thank Prof Epameinondas V. Tsianos of University Hospital of Ioannina, Prof Severine Vermeire of the University hospital in Leuven, Prof Walter Reinisch from the Medical University of Vienna, Dr Iris Dotan from Tel-Aviv Medical Center, and Prof Martin Zeitz from Universitätsmedizin in Berlin for their valuable support of this work. In addition, the authors would like to thank Dr Benjamin Avidan and Mrs Angela Chetrit, both from the Sheba Medical Center in Israel, and Prof Patrick Wessa from the K. U. Leuven Association, Belgium, for their valuable assistance and advice in the statistical analyses. Conflicts of interest The authors disclose no conflicts.