6-Tioguanine monitoring in steroid-dependent patients with inflammatory bowel diseases receiving azathioprine

Authors


Dr B. Bonaz, Département d'Hépato-Gastroentérologie, CHU de Grenoble, BP 217, 38043 Grenoble Cedex 09, France.
E-mail: bruno.bonaz@ujf-grenoble.fr

Summary

Background : 6-Thioguanine (6-tioguanine) nucleotides are the active metabolites of azathioprine.

Aim : The aim of the study was to evaluate the rate of clinical remission without steroids in steroid-dependent Crohn's disease and ulcerative colitis patients receiving azathioprine, the medium- and long-term efficacy and the predictive factors of clinical response when monitoring 6-tioguanine.

Methods : Steroid-dependent Crohn's disease and ulcerative colitis patients receiving either azathioprine or not (treated later with a daily dose of 2.5 mg/kg) were prospectively included. 6-Tioguanine was monitored at 1 and 2 months and every 3 months thereafter for 1 year. The azathioprine dose was adapted to reach a 6-tioguanine level of >250 pmol/8 × 108 red blood cells. Thiopurine methyltransferase genotype/phenotype was evaluated in some patients.

Results : A total of 106 patients were prospectively included (70 Crohn's disease, 36 ulcerative colitis). The clinical remission rate without steroids in patients receiving azathioprine, in intention-to-treat analysis, was 72% and 59% at 6 and 12 months, respectively. The remission rate was significantly higher in patients with 6-tioguanine >250 pmol/8 × 108 RBC (86% and 69% at 6 and 12 months, respectively; P < 0.01). No significant difference was observed between Crohn's disease and ulcerative colitis patients whether treated by azathioprine or not on inclusion. In the univariate analysis, the absence of Crohn's disease stenosis, a 6-tioguanine level >250 pmol/8 × 108 RBC, and an increase of erythrocyte mean corpuscular volume were the factors predictive of a favourable clinical response. In the multivariate analysis, only a 6-tioguanine level of >250 pmol/8 × 108 red blood cells was a predictive factor of favourable clinical remission.

Conclusions : Clinical remission without steroids is significantly more likely when monitoring 6-tioguanine so as to reach a level of >250 pmol/8 × 108 red blood cells in steroid-dependent Crohn's disease and ulcerative colitis patients receiving azathioprine (86% and 69% at 6 and 12 months, respectively).

Introduction

Steroids are efficient in the treatment of a flare of inflammatory bowel disease (IBD) in close to 90% of the cases in Crohn's disease (CD);1 approximately 80% of the patients have a favourable clinical response within 4 weeks.2 However, the percentage of steroid dependence, whatever its definition, is quite high (30–40%).3 Munkholm et al.4 reported 45% steroid-dependent patients but their definition of steroid dependence was restrictive (relapse following 30 days after steroid weaning). A 50% steroid dependence was observed in children by Markowitz et al.5 Immunosuppressive drugs, and especially azathioprine (AZA)/6-mercaptopurine (6-MP), are the treatment of choice for withdrawal of steroid dependence.1, 6 The response rate to steroids in patients with ulcerative colitis (UC) is about 70%;7 the rate of steroid-dependence is variable depending on the extent of the disease, the dose of steroids, and the definition of steroid dependence. Faubion et al.8 observed a 22% steroid dependence in the extended form of the disease. However, as recently discussed,6 few data are available on the remission rate without steroids in steroid-dependent CD and UC patients receiving AZA.

Azathioprine and 6-MP are pro-drugs requiring intracellular activation by a multi-enzymatic process involving hypoxanthine phosphoribosyl transferase (HGPRT), thiopurine methyltransferase (TPMT) and xanthine oxidase (XO).9 HGPRT converts AZA/6-MP into thioinosine monophosphate, which is metabolized into active cytotoxic 6-tioguanine nucleotides (6-TGN) that possess immune modifier activity.9 Tiede et al.10 found that AZA and its metabolites induced apoptosis of T cells from CD and control patients. Apoptosis induction required co-stimulation with CD28 and was mediated by specific blockade of Rac1 activation through binding of AZA-generated 6-tioguanine triphosphate (6-Tio-GTP) to Rac1 instead of GTP. The activation of Rac1 target genes such as mitogen-activated protein kinase, NF-κB, and bcl-xL was suppressed by AZA, leading to apoptosis via the mitochondrial pathway. 6-TGN largely contributes to the cytotoxicity of AZA/6-MP through their incorporation into the DNA and RNA.11 This process is in competition with the direct inactivation of AZA/6-MP by XO, or TPMT which is known to have a genetic polymorphism; 0.3% of the general population has low or no TPMT activity.12 In contrast, a high TPMT activity reduces AZA efficacy and increases AZA adverse effects, such as hepatotoxicity.13, 14 Recent studies have suggested that a therapeutic level of 6-TGN, defined by a red blood cell (RBC) concentration of 6-TGN >235–250 pmol/8 × 108 RBC, is required to obtain clinical remission in IBD patients.15–17 However, these data were not confirmed by others.18, 19

The aims of this study were to evaluate the rate of clinical remission without steroids in steroid-dependent IBD patients receiving AZA, the medium- and long-term efficacy of AZA (at 6 and 12 months, respectively) and the predictive factors of clinical response to AZA when monitoring 6-TGN so as to reach a level >250 pmol/8 × 108 RBC.

Materials and methods

Patient population

All steroid-dependent patients with ileo and/or colonic CD or rectocolonic UC were prospectively included between January 2001 and April 2002. For CD, phenotype was defined according to the Vienna classification.20 A steroid-dependent state was defined as a partial or complete clinical response to treatment with prednisone at 40–60 mg/day and relapse within 30 days after prednisone treatment was completed, or relapse with a dose reduction of prednisone resulting in the use of prednisone at doses leq15–25 mg/day for at least 6 months.3, 21 Steroid-dependent patients receiving maintenance treatment with AZA were also included. Inclusion criteria were: (a) age between 18 and 65 years, (b) patients given a stable dose of steroids during 2 weeks before inclusion, (c) patients receiving a stable dose of AZA for 6 months or more, and 6-TGN level (<250 pmol/8 × 108 RBC) for at least 3 months before inclusion or (d) patients receiving 5-aminosalicylates for ≥8 weeks, with a stable dose during the last 4 weeks before inclusion or (e) patients receiving antibiotics with a stable dose for more than 4 weeks. Patients were excluded from the study if they had received: (i) cyclosporin, methotrexate, infliximab or experimental agents within 3 months before inclusion; (ii) concomitant therapy known to interfere with AZA metabolism22, 23 initiated <4 weeks before inclusion, e.g. 5-aminosalicylates, diuretics or allopurinol. Patients with ano-perineal CD or CD limited to the upper digestive tract and hepatic, cardiac or renal failure were excluded. Patients with Epstein–Barr virus (EBV), cytomegalovirus (CMV), or Clostridium difficile infections were excluded. Finally, pregnant patients were excluded.

6-TGN nucleotide monitoring

Heparinized blood (5 mL) was obtained along with other routine blood samples at the time of a scheduled clinic visit or hospitalization. For the measurement of intracellular thionucleotides, the free base of each nucleotide was obtained by acid hydrolysis. The resulting purine, 6-thiopurine, was specifically extracted from the biological matrix, by phenyl mercury adduct formation, into toluene. Back extraction of this organic phase with 0.1 m HCl split the adduct and liberated the free 6-thiopurine into the acid layer, separated by reversed-phase high-performance liquid chromatography (HPLC) and quantified using UV detection.24–26 RBC concentration of 6-TGN was expressed as pmol/8 × 108 RBC.

6-Tioguanine level was evaluated on inclusion in patients already receiving AZA. AZA was introduced at a daily dose of 2.5 mg/kg in patients without AZA on inclusion. In both groups, 6-TGN was monitored 3 weeks later. If the 6-TGN level was >250 pmol/8 × 108 RBC, the AZA dose was not modified and a new dosage was formulated 1 month later to assess the 6-TGN kinetics. If 6-TGN level was <250 pmol/8 × 108 RBC, the AZA dose was increased to 25–50 mg/day depending on the 6-TGN level. A new 6-TGN dosage was then formulated 1 month later (i.e. 2 months after inclusion). At 2 months, if 6-TGN level remained <250 pmol/8 × 108 RBC, the AZA dose was increased to 25–50 mg/day and a new dosage was formulated 1 month later. After the second month of follow-up, a 6-TGN dosage was formulated every 3 months and the AZA dose was adapted according to the 6-TGN level in order to obtain 250 <6-TGN <400 pmol/8 × 108 RBC.

TPMT phenotype/genotype

In some cases, TPMT phenotype was evaluated by a reversed-phase HPLC assay, followed by UV detection (expressed as nmol/h/mL RBC),24 when TPMT deficiency (leucopenia <3000/mm3 in patients receiving AZA) or high TPMT activity (6-TGN <250 pmol/8 × 108 RBC with a daily AZA dose >2.5 mg/kg) was suspected. TPMT genotype was evaluated in case of leucopenia (<3000/mm3) and when the 6-TGN level was >600 pmol/8 × 108 RBC with a daily AZA dose <1.5 mg/kg. These patients were genotyped for two polymorphisms in TPMT transitions at positions 460 and 719 using a polymerase chain reaction allele-specific nucleotide hybridization assay.27, 28

Full blood count and liver function

Full blood count [especially leucocytes, haemoglobin concentration, platelets, mean corpuscular volume (MCV)] and liver function (alanine transaminase, aspartate transaminase, alkaline phosphatase, gamma-glutamyl transferase, bilirubins) were evaluated once a week during 4 weeks and then once a month during the rest of the study when AZA was initiated on inclusion. In patients already receiving AZA, the biological surveillance was performed on inclusion, every week during 1 month when increasing the AZA dose and every month thereafter. The serum concentration of vitamin B12 and folate was also determined.

Clinical outcome

Patients were followed up monthly during 1 year. Clinical activity was evaluated on inclusion and on each visit using the Crohn's Disease Activity Index (CDAI)29 for CD and the UCDAI30 (ulcerative colitis disease activity index) or the Lichtiger's score31 for UC. CDAI was evaluated monthly during three consecutive months and at 6 and 12 months thereafter. The UCDAI score was calculated on inclusion, 3, 6 and 12 months; The Lichtiger's score was evaluated at 1 and 2 months to reduce the number of endoscopic examinations and because it seemed to us too early to control endoscopy before 3 months for clinical response to treatment. In ileo-colonic CD, clinical remission was defined by a CDAI score <150 and clinical response assessed by a decrease of the CDAI score of at least 100 points. A clinical relapse was defined by a score >150 with an increase of the CDAI score of at least 100 points. An increase of the steroid dose for clinical worsening or a need for surgery was defined as a failure. In rectocolonic UC, a clinical remission was defined by a UCDAI score <2 and a clinical response by a decrease of the UCDAI score of at least two points. Clinical relapse was defined by a UCDAI score >3. A clinical remission on consultation without rectosigmoidoscopy was defined by a Lichtiger's score <10. A rectosigmoidoscopy was performed when the Lichtiger's score was >10, in favour of a clinical relapse; the UCDAI was then calculated.

Steroid tapering was performed according to the GETAID protocol,32 with some modifications, after 1 month of treatment in case of clinical response (a CDAI decrease of at least 100 points or a UCDAI decrease of at least two points after inclusion). Prednisone was reduced by 10 mg/week to 30 mg, then followed by a 5 mg/week decrease down to a weekly dose of 10 mg. Prednisone was then replaced by hydrocortisone at a dose of 20 mg (10 mg in the morning and at noon) to avoid adrenal deficiency caused by the long-term use of steroids by these patients. Plasma cortisol was evaluated after 2 weeks of 20 mg/day hydrocortisone which was interrupted if the plasma cortisol level was ≥300 nmol/L. If a clinical relapse was observed during steroid tapering, a two-step increase of prednisone, for at least 2 weeks, was made until clinical improvement, then steroid tapering was attempted again; these patients were considered as therapeutic failure.

Statistical analysis

Statistical analysis was performed using the SPSS 9.0 software. Quantitative variables were expressed as median and range, or as mean ± standard deviation when normally distributed. The parametric Student's test or the nonparametric Mann–Whitney's test were used, when appropriate, to compare quantitative variables between the two groups. Fisher's exact test was used to evaluate associations between the therapeutic response and categorical variables, such as type of IBD, disease location, type of complications, or smoking. A multivariable analysis by logistic regression was then conducted in order to identify independent predictive parameters of clinical response. All the variables which were significant in the univariate analysis were included in the multidimensional model. A P-value <0.05 was considered to be significant.

Results

One hundred and six patients (70 CD, 36 UC; 52 men, 54 women; sex ratio of men/women 0.9) with a median age of 42 years (range 18–65 years) were prospectively included between January 2001 and April 2002. The patient demographics are presented in Table 1. Among the 106 patients, (a) 42 patients (39.6%; 30 CD, 12 UC) were already receiving AZA on inclusion for a median of 2.8 years (range: 6 months to 8 years) with a median daily AZA dose of 1.92 mg/kg (range 1.67–2.45 mg/kg). On inclusion, the 6-TGN level of these patients was always < 250 pmol/8 × 108 RBC (median 167 pmol/8 × 108 RBC; range 91–240). The median steroid dose was 25 mg/day (range 15–40); (b) 64 patients (60.4%; 40 CD, 24 UC) were free of AZA on inclusion. The median steroid dose was 20 mg/day (range 15–40).

Table 1.  Patient demographics
  1. * Median (range).

Number106 (70 CD, 36 UC)
CD locationColitis: 20
Ileitis: 30
Ileo-colitis: 20
UC locationPancolitis: 25
Left-sided colitis: 11
Gender (M/F)52/54
Age (years)42 (18–65)*
Smoker (number)30
 CD30/70
 UC0/36
Steroid-dependent patients receiving AZA on inclusion
 Number42
 AZA dose (mg/kg/day)1.92 (1.6–2.35)*
 Duration of AZA (months)32 (6–96)*
 6-TGN levels (pmol/8 × 108 RBC)167 (91–240)*
 Steroid dose (mg/day)25 (15–40)*
 Duration of steroids (months)35 (8–98)*
Steroid-dependent patients without AZA on inclusion
 Number64
 Steroid dose (mg/day)20 (15–40)*
 Duration of steroids (months)16 (12–25)*

Clinical remission without steroids for at least 3 months

Only seven patients were excluded from the study because of side-effects caused by AZA, requiring treatment interruption. As the results of per protocol and intention-to-treat analysis were not statistically significantly different, only the latter group is presented; consequently, all our results are given in intention-to-treat analysis.

Clinical remission without steroids for at least 3 months was obtained in 77 patients (72%; Table 2). The median delay to obtain clinical remission without steroids in patients receiving AZA was 2.4 months (range 1.6–7). No clinical relapse was observed in these 77 patients 6 months after weaning of steroids. Furthermore, clinical remission was observed in 63 patients (59%; P < 0.01; Table 2) at 1 year. No significant difference, in terms of clinical remission, was observed at 6 and 12 months between CD and UC patients (P =0.64) (Table 3).

Table 2.  Percentage of clinical remission without steroids in patients receiving AZA
Percentage of clinical remission6 months (%)12 months (%)P-value
  1. Notes: Percentage of clinical remission without steroids in patients receiving AZA at 6 months and 1 year, whatever the type of IBD on inclusion, in either all steroid-dependent IBD patients (n = 106), or steroid-dependent patients free (n = 64) or receiving (n = 42) AZA on inclusion and in the sub-group of patients with a 6-TGN level >250 pmol/8 × 108 RBC or not. No significant difference was observed between the three groups at 6 months and 1 year while the difference was significant (**P < 0.01) between 6 months and 1 year for the total group of patients and between 6 months and 1 year for the sub-group with a 6-TGN level >250 pmol/8 × 108 RBC (P = 0.03).

Total group (n = 106)7259<0.01
 Patients without AZA on inclusion (n = 64)7058N.S.
 Patients receiving AZA on inclusion (n = 42)7762N.S.
Patients with 6-TGN of 250 pmol/8 × 108 RBC (n = 90)86690.03
 Patients without AZA on inclusion (n = 54)8873N.S.
 Patients receiving AZA on inclusion (n = 36)8366N.S.
Patients with 6-TGN <250 pmol/8 × 108 RBC (n = 16)00 
 Patients without AZA on inclusion (n = 10)00 
 Patients receiving AZA on inclusion (n = 6)00 
Table 3.  Percentage of clinical remission without steroids in patients receiving AZA
Percentage of clinical remission6 months (%)12 months (%)P-value
  1. Notes: Percentage of clinical remission without steroids in patients receiving AZA at 6 months and 1 year in either all steroid-dependent IBD patients (n = 106) or CD patients (n = 70) or UC patients (n = 36) receiving AZA according to the level of 6-TGN (> or <250 pmol/8 × 108 RBC). No significant difference was observed between the three groups at 6 months and 1 year while the difference was significant (**P < 0.01) between 6 months and 1 year for the total group and CD patients. Clinical remission was significantly higher (P < 0.01) in patients with a 6-TGN level >250 pmol/8 × 108 RBC. In these patients, the difference was significant (*P = 0.03) between 6 months and 1 year.

Total group (n = 106)7259<0.01
 6-TGN <250 pmol/8 × 108 RBC (n = 16)00 
 6-TGN = 250 pmol/8 × 108 RBC (n = 90)86690.03
CD patients (n = 70)7458<0.01
 6-TGN <250 pmol/8 × 108 RBC (n = 10)00 
 6-TGN = 250 pmol/8 × 108 RBC (n = 60)88670.05
UC patients (n = 36)6861N.S.
 6-TGN <250 pmol/8 × 108 RBC (n = 6)00 
 6-TGN = 250 pmol/8 × 108 RBC (n = 30)8373N.S.

In the group of patients receiving AZA on inclusion (n = 42; 30 CD, 12 UC) with a 6-TGN level <250 pmol/8 × 108 RBC (median 167 pmol/8 × 108 RBC; range 91–240), 32 patients (77%) were in remission without steroids at 6 months (Table 2). The 6-TGN level of the 32 patients in remission was >250 pmol/8 × 108 RBC. A 6-TGN level >250 pmol/8 × 108 RBC was reached in 36 patients (87.5%). In six patients, despite an increase of the AZA dose (median 4.1 mg/kg/days; range 3.7–5.2), the 6-TGN level remained <250 pmol/8 × 108 RBC (median 72 pmol/8 × 108 RBC; range 35–225) at 6 months. These six patients had a high TPMT activity (median 34 nmol/h/mL RBC; range 28–36). In three of them, the increase of the AZA dose >4.2 mg/kg/day was followed by clinical remission at 9 months; these results show that in case of TPMT activity >32 nmol/h/mL RBC, a 50% remission rate in patients receiving AZA is observed in intention-to-treat analysis. Indeed, the median daily AZA dose on inclusion was 1.92 mg/kg with a median 6-TGN level on inclusion of 167 pmol/8 × 108 RBC (range 91–240) while the median daily AZA dose in patients in clinical remission without steroids was 3.4 mg/kg with a median 6-TGN level of 355 pmol/8 × 108 RBC (range: 251–487). No difference was observed between CD and UC responder patients when comparing either the clinical remission rate without steroids in patients receiving AZA at 6 and 12 months (P = N.S.; Table 3) or the AZA dose and the 6-TGN level necessary to obtain a clinical remission (Table 3).

In the group of patients free of AZA on inclusion (n =64; 40 CD, 24 UC), clinical remission without steroids was obtained in patients receiving AZA in 45 of 64 (70%) patients and 37 of 64 (58%) patients at 6 and 12 months, respectively (P = N.S.; Table 2). All these patients had a stable 6-TGN level >250 pmol/8 × 108 RBC (median 325; range 252–454); the median delay to obtain clinical remission was 2.8 months (range 2–7). No difference was observed for the rate of clinical remission and for the 6-TGN level between CD and UC patients (P > 0.10) (Table 3). Among the 19 patients still steroid-dependent (median dose 20 mg/day; range 15–30), 10 (52.6%) patients presented a 6-TGN level <250 pmol/8 × 108 RBC despite a progressive increase of the daily AZA dose (median 3.7 mg/kg; range 3.4–4.6); six of these patients had high TPMT activity (median 31 nmol/h/mL RBC; range 26–34). The last four patients had median TPMT activity of 22 nmol/h/mL RBC (range 18–24). None of the CD patients still steroid-dependent receiving AZA had stopped smoking while 50% of the patients in remission had stopped smoking. A 6-TGN level >250 pmol/8 × 108 RBC allowed a clinical remission without steroids in 77 (86%) and 62 (69%) patients respectively at 6 and 12 months while none of the patients with a 6-TGN level <250 pmol/8 × 108 RBC was in remission (P < 0.01; Table 3). In patients with a clinical response, the median delay of steroid weaning was 2.3 months (range 1.5–3.6).

The 106 patients were followed-up for at least 1 year with a median of 1.6 year (range 1–2.6). Twenty-eight (20%) patients relapsed at 1 year under AZA despite a 6-TGN level >250 pmol/8 × 108 RBC in the course of follow-up. Sixty per cent of these patients were treated with infliximab administered intravenously at a dose of 5 mg/kg at week 0, 2 and 6 with a 80% clinical remission rate at 3 months. Patients receiving AZA-infliximab with a transient 6-TGN level increase 2 weeks after the first infusion responded to infliximab, as previously described.26 Twenty per cent of the patients intolerant to AZA (as for example pancreatitis) were switched to tioguanine (TG) at a daily dose of 20 mg25 and 50% were weaned at 6 months. None of these patients had liver test abnormalities under TG. Finally, six (20%) patients were treated with methotrexate (25 mg/week intramuscularly for 3 months and then 15 mg/week as a maintenance treatment) with a clinical remission without steroids in one-third (n = 2) of the patients at 12 months.

The median delay of clinical relapse in patients receiving AZA was 200 days (range 102–304). No difference was observed between CD and UC patients with or without AZA on inclusion (data not shown).

Age, longevity of the disease, level of colonic involvement, presence of stenosis, and smoking was not a discriminative factor in this subgroup of patients. In contrast, a high TPMT activity >26 nmol/h/mL RBC was a factor of AZA resistance.

Influence of 6-TGN levels on clinical remission in steroid-dependent IBD patients

Among the 106 patients included in this study, we identified a group of 90 patients (61 CD, 29 UC) with a 6-TGN level >250 pmol/8 × 108 RBC. In this group, the percentage of remission at 6 and 12 months was significantly greater than in the whole population or in the group with 6-TGN <250 pmol/8 × 108 RBC [86% vs. 72% vs. 0% at 6 months (P < 0.05); 69% vs. 59% vs. 0% at 12 months (P < 0.05)]. The intake of AZA before inclusion did not influence the rates of clinical remission. Indeed, the clinical remission rate at 6 months was 86% (77 patients) i.e. in the same order of magnitude in patients receiving (88%, n = 31) AZA on inclusion or not (83%, n = 46; P = N.S.). The remission rate at 12 months was 69% (n = 62) for the total group, 73% (n = 26) in the group of patients with AZA on inclusion and 66% (n = 36) in the group of patients without AZA on inclusion (P = N.S.; Table 2).

In our study, a 6-TGN level >250 pmol/8 × 108 RBC was the only predictive factor to wean steroid dependence in the multivariate analysis (Table 4). Indeed, 86% of the patients with a 6-TGN level >250 pmol/8 × 108 RBC were in remission without steroids at 6 months while only 14% of the patients with a 6-TGN level >250 pmol/8 × 108 RBC was still steroid dependent. At 12 months, the percentage of patients in clinical remission without steroids with a 6-TGN level >250 pmol/8 × 108 RBC decreased to 69%. In contrast, none of the patients with a 6-TGN level <250 pmol/8 × 108 RBC reached clinical remission without steroids (P < 0.01; Table 3). The threshold of a 6-TGN level >250 pmol/8 × 108 RBC was discriminative to reach a clinical remission without steroids 6 months after inclusion (CD: 88%; n = 52 vs. UC: 83%; n = 25) whatever the type of IBD. In contrast, only 13 CD patients (14%) with a 6-TGN level >250 pmol/8 × 108 RBC were still steroid dependent at 6 months vs. 17% in the UC group (P = N.S.). In both UC and CD, none of the patients with a 6-TGN level <250 pmol/8 × 108 RBC were in clinical remission without steroids at 6 months (Table 3). Location of the disease and smoking were not factors of steroid dependence with failure of AZA response whatever the type of IBD. In contrast, the clinical remission rate was significantly higher in the group of patients with a TPMT activity <26 nmol/h/mL RBC [89% (n = 48) vs. 31% (n = 3); P < 0.01] and in the group of patients without CD stenosis (76% vs. 32%; P = 0.04). In addition, in univariate analysis, erythrocyte MCV was significantly higher in the group of patients in clinical remission without steroids than in the group which did not respond to AZA (94 ± 3 vs. 87 ± 4; P < 0.01).

Table 4.  Multivariable logistic regression analysis for inflammatory Crohn's disease
VariableResponse OR (95% CI)P-value
6-TGN = 250 pmol/8 × 108 RBC11.2 (2.6–24.2)<0.001
Strictures1.2 (0.4–4.2)0.51
TPMT activity <26 nmol/h/mL RBC2.67 (0.54–4.56)0.75

Logistic regression analysis was performed to adjust for age on inclusion, disease duration and sex (Table 4). A 6-TGN level >250 pmol/8 × 108 RBC was a predictive factor of AZA response in intention-to-treat analysis [odds ratio (OR) 11.2; 95% confidence interval (CI) 2.6–24.2]. In contrast, predictive factors observed in univariate analysis (CD stenosis, MCV) were not observed in multivariate analysis.

Adverse effects

Leucopenia was observed in 10 patients (eight CD, two UC; 9.4%). Two of them had a 6-TGN level >600 pmol/8 × 108 RBC (680 and 710 pmol/8 × 108 RBC); one patient was heterozygote deficient for TPMT while no TPMT deficiency was observed in the second patient. Leucopenia (<1500/mm3) lead to interruption of AZA treatment in two patients who completely recovered. Three patients presented viral infection (EBV: n = 2; CMV: n = 1), while the 6-TGN level ranged between 250 and 400 pmol/8 × 108 RBC, and lead to interruption of AZA. One patient presented high TPMT activity (34 nmol/h/mL RBC) requiring a high daily AZA dose (5.3 mg/kg) to reach a 6-TGN level >250 pmol/8 × 108 RBC; even if 6-TGN levels did not reach a high value, other metabolites may possibly be involved as well as a possible concomitant inosine triphosphate pyrophosphatase (ITPase) gene mutation (not performed). Four patients presented leucopenia because of a folate and/or vitamin B12 deficiency; vitamin supplementation brought the leucocyte count back to normal without the need to stop AZA.

One of our patients with a high TPMT activity (34 nmol/h/mL RBC) presented liver test abnormalities (fivefold increase of alanine aspartate/transaminase); interruption of AZA treatment restored normal liver tests. Two patients developed pancreatitis within the first month of inclusion contraindicating the use of AZA. Five patients presented digestive intolerance (nausea, vomiting) to AZA requiring a switch to 6-MP without adverse effects and identical 6-TGN monitoring; in two cases, the persistence of adverse effects required interruption of AZA/6-MP treatment with a successful switch to TG (20 mg/day).

Discussion

This study shows that clinical remission without steroids is obtained by AZA treatment in 73% and 59% of steroid-dependent IBD patients at 6 and 12 months, respectively. The remission rates are significantly increased in patients with a 6-TGN level >250 pmol/8 × 108 RBC (86% and 69% at 6 and 12 months, respectively; P < 0.05).

AZA efficacy in steroid-dependent patients is not well documented in the literature.6, 21 Indeed, various authors have tried to analyse the association AZA steroids but they did not take into account steroid-dependent patients. Ewe et al.33 compared the association of steroids (60 mg/day) with AZA (2.5 mg/kg/day) vs. steroids alone; unfortunately, the trial did not address steroid dependence and the authors aimed to taper steroids at a dose of 10 mg/day. After 4 months of treatment, 76% of the patients receiving AZA steroids were in clinical remission (with a steroid dose of 10 mg/day) vs. 38% in the group of patients receiving steroids alone (P = 0.03). In this study, patients were considered as weaned of steroids when reaching a dose of 10 mg/day. Candy et al.34 included 63 CD patients either receiving steroids alone or combined with AZA (same dose as above) with an attempt to stop steroids at 3 months. The remission rate without steroids at 15 months in patients receiving AZA steroids was 42% vs. 7% in the steroid group. Data on steroid dependence were not presented in this study either and no significant difference was observed between the two groups at 3 months. Markowitz et al.5 reported similar results in children. They included 55 children for a first CD flare randomized in two groups (steroids vs. steroids–6-MP). The rate of clinical remission without steroids at 1 year was significantly high in the 6-MP group (81% vs. 41%; P < 0.05); none of the patients was steroid dependent. In a double-blind cross-over study, Present et al.35 compared 6-MP vs. placebo in 83 CD patients. For 75% of the patients receiving 6-MP, steroid dosage could be reduced vs. 36% in the placebo group. In this study, the authors gave neither the number of steroid-dependent patients nor the remission rate. In the meta-analysis performed by Pearson et al.,36 the authors showed that AZA therapy allowed steroid tapering in 65% of the patients (95% CI 56–74), significantly higher than in the placebo group (36%; 95% CI 27–45). No data were available regarding steroid-dependent patients. Mantzaris et al.37 recently compared AZA vs. AZA-olsalazine in steroid-dependent UC patients. In the group of 34 patients receiving AZA, 27 patients (81%) were in remission without steroids at 2 years. These results are higher than ours; however, patients needed to be in remission under AZA when included and without steroids. Indeed, included patients had previously responded to AZA. Dubinsky et al.,16 using the 6-TGN level in patients receiving AZA, did not study the clinical response rate in steroid-dependent patients receiving AZA. Sixty-eight patients were included in this study but no data were available even if, in a multivariate analysis, only the 6-TGN level was a predictive factor of AZA response. In Cuffari et al.’s15 study, the number of steroid-dependent patients was not documented; 18 patients with a 6-TGN level <250 pmol/8 × 108 RBC were AZA resistant. The increase of AZA dosage in this subgroup lead to clinical remission in 12 of 18 (66.7%) patients; this rate is comparable with that observed in our study. In contrast, Dubinsky et al.,13 in a study performed on 51 patients resistant to 6-MP, showed that an increase of the 6-MP dose was followed by clinical remission in only 14 (27.5%) patients with a significant increase of 6-TGN (183–306 pmol/8 × 108 RBC; P = 0.03). Non-responding patients had high TPMT activity as observed by a significantly (P < 0.001) higher 6-methyl mercaptopurine ribonucleotides)/6-TGN ratio. Our study confirms that high TPMT activity is unfavourable for response to AZA. However, in our study, we observed high TPMT activity in 10% of the population as defined by a TPMT activity >26 nmol/h/mL in RBC, commonly reported in the literature,12 while this percentage was higher in Dubinsky's study.13 In a more recent study, Cuffari et al.17 analysed the influence of TPMT activity and 6-TGN levels on clinical response to AZA. In multivariate analysis, a TPMT activity <15.3 U/mL was correlated with a six times higher clinical response rate; a 6-TGN level >292 pmol/8 × 108 RBC had a positive predictive value of clinical response to AZA reaching around 86%. Our results are close to those of Cuffari et al.17 as the clinical remission rate was significantly higher in the group of patients with a TPMT activity <26 nmol/h/mL RBC (89% vs. 31%; P < 0.01). However, our evaluation technique (HPLC) of TPMT activity is different from the radioactive assay used by Cuffari et al..17 As TPMT activity evaluation was restricted to 54 of 106 (51%) patients, it was not possible to evaluate this parameter in a uni- and multivariate analysis. In addition, because of the study design, it was not possible to confirm a correlation between TPMT activity and 6-TGN levels.

To our knowledge, this is the first time that AZA has proved effective in a group of steroid-dependent patients with a clinical remission rate of 72% at 6 months. This rate decreased over time but was still 59% at 12 months. These remission rates are higher than in studies on nonsteroid-dependent patients, possibly because of the use of 6-TGN monitoring in our study with a threshold of therapeutic efficiency >250 pmol/8 × 108 RBC. Indeed, recent studies suggested that a therapeutic level of 6-TGN, defined by a RBC concentration of 6-TGN >235–250 or even 292 pmol/8 × 108 RBC, needed to be reached to obtain clinical remission in IBD patients.15, 16 In the present study, we clearly show that none of the steroid-dependent patients with a 6-TGN level <250 pmol/8 × 108 RBC reached a clinical remission without steroids while 86% of the patients with a 6-TGN level >250 pmol/8 × 108 RBC were in remission without steroids. At 1 year, 69% of the patients were still in remission without steroids while maintaining a 6-TGN level >250 pmol/8 × 108 RBC. Consequently, the threshold of 250 pmol/8 × 108 RBC needs to be reached in steroid-dependent patients. In multivariate analysis, this is the principal factor of AZA response without steroids. A high TPMT activity is a factor of AZA resistance (P < 0.05). In the subgroup of CD patients, stenosis and/or smoking did not appear as a factor of AZA failure, most likely because these subgroups were too small. However, in Munkholm's study,4 steroid dependence was not correlated with any clinical parameters and was independent of the location of the disease. When a threshold of 235 pmol/8 × 108 RBC was chosen as described by Cuffari et al.,15 the remission rate decreased to 78% vs. 86% (P = N.S.) and 63% vs. 69%; (P = N.S.) at 6 and 12 months respectively (data not shown). Even if this difference was not significant, the results were nevertheless more favourable with a threshold of 250 pmol/8 × 108 RBC. Furthermore, the aim of this study was not to look for the more discriminative threshold of 6-TGN. Results with the group of UC patients were similar to those in the CD group with a remission rate of 86% at 6 months without steroids. Very recently, Lopez-Sanroman et al.38 found a clinical remission rate close to ours at 1 year, in steroid-dependent UC patients without steroids (n =36), receiving AZA/6-MP. The authors did not use 6-TGN monitoring but they determined TPMT activity before introducing AZA/6-MP and they adapted the dose accordingly. They found that erythrocyte MCV was higher in AZA/-6MP responders. Our results are concordant as the increase of erythrocyte MCV was a factor of clinical response in the univariate analysis (but not in the multivariate analysis). In fact, we think that erythrocyte MCV could simply be a surrogate marker of the 6-TGN concentration, as demonstrated by Thomas et al.39 In addition, 6-TGN monitoring enabled faster steroid weaning compared with Lopez-Sanroman et al.'s study,38 i.e. 2.4 months in our study vs. 4.6 months respectively, and we had less adverse effects (20% vs. 38%). However, this is based on a comparison with previous published studies and thus does not allow us to draw any conclusions that 6-TGN monitoring reduces the delay of steroid withdrawal.

We found the threshold of 6-TGN, >250 pmol/8 × 108 RBC, useful for UC but this was not observed by Cuffari et al.15 where only 19 UC patients were included vs. 36 in ours. Our results are of interest as 83% of the patients with a 6-TGN level >250 pmol/8 × 108 RBC were in clinical remission while only 14% were not. Conversely no patients with 6-TGN <250 pmol/8 × 108 RBC were in clinical remission without steroids. 6-TGN monitoring in steroid-dependent patients receiving AZA is of interest as in 48 steroid-dependent patients with AZA failure, an increase of the AZA dose to reach a 6-TGN level >250 pmol/8 × 108 RBC was followed by a clinical remission without steroids in 80% of the cases. Cuffari et al.15 have shown, in a subgroup of 18 patients, that a 6-TGN monitoring to reach a 6-TGN level >250 pmol/8 × 108 RBC allowed a clinical remission in 12 of 18 (60%) patients.

Limitations of the study

We are aware, that a prospective randomized study vs. placebo would be necessary to confirm our results in particular for the UC group (n = 36). We did not have any control group in double-blind vs. placebo; this could have influenced our results. Indeed, the placebo response rate may reach 40%. However, the remission rate at 6 and 12 months seems much higher than that in studies with AZA used in the active form of the disease vs. placebo (43% of remission at 15 months in Candy et al.'s study34). These results are far better than those obtained with other immunosuppressants. Feagan et al.40 studied the effect of methotrexate in steroid-dependent patients vs. placebo. Thirty-nine per cent of their patients were in remission at 16 weeks vs. 19% in the placebo group. Our clinical results were much more favourable when compared either with those of the methotrexate or the placebo group. Recent data published by the GETAID show infliximab, administered at week 0, 2, and 6, in association with AZA in steroid-dependent patients without AZA or receiving AZA on inclusion.41 The remission rates were 75% vs. 38% and 57% vs. 29% at 6 and 12 months in patients receiving AZA and placebo respectively. The results were not different whatever the AZA status on inclusion; 6-TGN was not monitored. These data are close to ours even if we did not use infliximab but a 6-TGN monitoring. Consequently, we suggest use of infliximab in patients for whom AZA therapy has failed despite 6-TGN and/or TPMT monitoring. Despite a high median AZA dose (3.4 mg/kg/day), the adverse effect rates requiring interruption of the AZA treatment are concordant with those of the literature (20% and 9.5% respectively) as reported by Fraser et al.,42 who observed a 28% side-effect rate. Recent data showed that some AZA adverse effects, independent of the 6-TGN level and TPMT activity, were due to ITPase gene mutation.43 These data could account for the comparable adverse effect rate despite the high AZA dose. The use of AZA metabolic markers (TPMT genotype/phenotype, 6-TGN) should allow to reduce adverse effects but results are contradictory at this time.14, 44 Finally, the genetic mutation of ITPase may explain some leucopenia;43 in our study, two patients had an unexplained leucopenia perhaps because of this type of mutation.

Our data on TPMT activity >26 nmol/h/mL RBC as an AZA resistance factor show certain limitations. Indeed, the choice of a particular group of patients with an inadequacy between the AZA dose and the 6-TGN level, suggesting an abnormal TPMT activity, introduces a bias which does not enable us to directly relate clinical remission and TPMT activity. In addition, TPMT activity evaluation was performed in patients receiving AZA which is known to increase TPMT activity.22 Finally, if our results are in favour of a cut-off level of 6-TGN at about 250 pmol/8 × 108 RBC in the management of IBD patients, and the study design amplifies this phenomenon. It is important to mention that, on inclusion, some patients were already receiving AZA with a 6-TGN level <250 pmol/8 × 108 RBC thus excluding responders receiving AZA before inclusion despite 6-TGN level <250 pmol/8 × 108 RBC and patients resistant to AZA with a 6-TGN level >250 pmol/8 × 108 RBC. In addition, the aim of 6-TGN monitoring was to reach a 6-TGN level >250 pmol/8 × 108 RBC from 2 months and to evaluate the clinical remission rate at this threshold but not to look for a cut-off level. Our study cannot exclude that a lower 6-TGN level could induce the same results in some patients. In cases of AZA failure, it seems necessary to verify the 6-TGN level and, if this level is <250 pmol/8 × 108 RBC, then the AZA dose should be increased to reach this threshold which, according to our study, lifts the AZA resistance in about 80% steroid-dependent patients.

Conclusion

AZA/6-MP is the gold standard in the treatment of steroid-dependent IBD patients. The clinical remission rate without steroids in patients receiving AZA is 72% and 59% at 6 and 12 months respectively. Therapeutic monitoring of AZA metabolites to reach a 6-TGN level >250 pmol/8 × 108 RBC is of particular interest as a predictive factor of clinical response in these patients (OR 11.2; 95% CI 2.6–24.2). Switching from AZA to another treatment, or a combination with infliximab, is indicated when there is no clinical remission without steroids if 6-TGN level is >250 pmol/8 × 108 RBC or if this threshold is impossible to reach.

Acknowledgements

The authors would like to thank Pr. Pierre-Emmanuel Colle for copy-editing this document.

Ancillary