6-TGNs affect the immune response via several potential mechanisms. They are incorporated into DNA, thereby inhibiting its synthesis and causing DNA strand breakage—and thus suppressing cell proliferation [10]. Other mechanisms include direct cytotoxicity, inhibition of de novo purine biosynthesis, and suppression of cytokine synthesis [8, 11–13].

Another important mechanism is by inducing T-cell apoptosis by modulating cell Rac 1 signaling. Rac 1 bounds to the metabolite 6-Thio-GTP instead of GTP, thus blocking the Rac1 activation pathway. Thus, Rac1 targeted genes as mitogen activated protein kinase, NF-kB, and bcl-x (L) are suppressed—leading to mitochondrial pathway of apoptosis [14, 15]. Defects in T cell apoptosis were identified as one of the triggers of gut inflammation in CD [16, 17].

The use of thiopurines for IBD treatment was initiated in the early 1960s [18, 19]. Since then, the effect of the medication on disease activity and symptoms has been intensively studied. AZA AND 6-MP were both shown to be effective in induction of remission of active CD with an odds ratio of up to 3.1 compared with placebo [20, 21].

Therapeutic onset after thiopurine initiation was shown to be between 12 and 17 weeks, the time needed for TGNs to be incorporated into DNA [22]. This Cochrane meta-analysis published in 2010 included eight randomized placebo-controlled trials in adults with active CD. The outcome measure was the proportion of patients with clinical improvement or remission (as defined by the Crohn’s disease activity index (CDAI) , the Harvey–Bradshaw Index , subjective evaluation, or steroid sparing effect ). The pooled response rate was 54 % for the group with thiopurine analogs versus 34 % for the placebo treated patients.

The OR of response to azathioprine or 6-mercaptopurine therapy compared with placebo in active Crohn’s disease was 2.43 (95 % CI 1.62–3.64). The number needed to treat (NNT ) was 5 to observe an effect of therapy in one patient. Treatment of >17 weeks resulted in an OR of 2.61 (95 % CI 1.69–4.03). A steroid sparing effect was seen with an OR of 3.69 (95 % CI 2.12–6.42), with NNT of 3 to observe steroid sparing in one patient [22].

Another Cochrane meta-analysis including seven maintenance CD trials found that the OR for maintenance of remission was 2.32 with an NNT of 6. The OR for maintenance of remission with 6-MP was 3.32, with an NNT of 4 [2]. Overall remission rate was 71 % for AZA compared with 55 % for placebo.

However, a recent Cochrane database systematic review published in 2013 showed contradicting results [23]. This meta-analysis included 13 randomized controlled trials of thiopurine treatment compared to placebo or active therapy involving adult patients with active CD. One thousand two hundred and eleven patients were included in the study. The study found no statistically significant difference in clinical remission rates between thiopurines and placebo. Forty-eight percent (95/197) of patients receiving antimetabolites achieved remission compared to 37 % (68/183) of placebo patients (five studies, 380 patients; RR 1.23, 95 % CI 0.97–1.55). No statistically significant difference in clinical improvement rates were found between azathioprine or 6-mercaptopurine and placebo treated patients. Forty-eight percent (107/225) of patients receiving antimetabolites achieved clinical improvement or remission compared to 36 % (75/209) of placebo patients (eight studies, 434 patients; RR 1.26, 95 % CI 0.98–1.62). There was a statistically significant difference in steroid sparing (defined as prednisone dose < 10 mg/day while maintaining remission) between azathioprine and placebo . Sixty-four percent (47/163) of azathioprine patients were able to reduce their prednisone dose to <10 mg/day compared to 46 % (32/70) of placebo patients (RR 1.34, 95 % CI 1.02–1.77).

These results challenge the accepted concept regarding thiopurine treatment in CD patients that was accumulated in 6 decades of usage, and should be taken with excessive cautious.

Nevertheless, the second European evidence-based Consensus on the diagnosis and management of Crohn’s disease from the European Crohn’s and Colitis Organization (ECCO) states that AZA 1.5–2.5 mg/kg/day or 6-MP 0.75–1.5 mg/kg/day may be used in active CD as adjunctive therapy or steroid-sparing agent, and is also effective for the maintenance of remission in CD and has a steroid-sparing effect [24].

Mucosal healing is an important indicator to evaluate the efficacy of treatment and serves as a predictor of delayed onset complications and a decreased need for surgery [25]. Therefore, mucosal healing is used as a therapeutic end-point in many clinical trials [26–28].

Mucosal healing as a therapeutic goal in patients treated with thiopurines as a single agent was assessed in several studies [29–32]. A recent study from China assessed long-term mucosal healing in 36 patients with small bowel CD naïve to biologic therapy and to thiopurines using double balloon enteroscopy . After 6 months of treatment with an average dose of 61.8 ± 17.2 mg/day of AZA 26 patients (72.2 %) achieved clinical remission and the ten patients (27.8 %) had a clinical response. After 12 months of treatment, complete mucosal healing was achieved in seven patients (19.4 %), near-complete healing (defined as a marked endoscopic improvement, possible aphthous ulcers (<0.5 cm) or erosions in the absence of stenosis, the affected segment is less than 50 %) in 2 (5.6 %), partial healing (defined as less than 50 % affected areas and the size of the biggest ulcer of less than 2 cm, considerable numbers of ulcers still persisted and single luminal narrowing observed, but was passable by DBE) in 10 (27.8 %) and no healing in 17 (47.2 %). After 24 months of treatment complete healing was observed in 11 (30.6 %), near-complete healing in 9 (25.0 %), partial healing in 12 (33.3 %) and no healing in 4 (11.1 %) [29]. Another study compared AZA (2.0–2.5 mg/kg a day, n = 38) to Budesonide (BUD) (6–9 mg /day, n = 39) for 1 year [30]. Mucosal healing was assessed using ileocolonoscopy with regional biopsies. At the end of the study 32 and 25 patients in the AZA and BUD groups, respectively, were in clinical remission (P = 0.07). The Crohn’s Disease Endoscopic Index of Severity (CDEIS) score fell significantly only in the AZA group (P < 0.0001). Complete or near complete healing of colonic mucosa was achieved in 83 % of AZA-treated patients compared to 24 % of BUD-treated patients (P < 0.0001). Histologic activity as assessed by an average histology score (AHS) fell significantly only in the AZA group (P < 0.001 versus baseline) and was significantly lower than in the BUD group at the end of the study (P < 0.001). In the terminal ileum complete healing was achieved in (59 %), near-complete healing in 4 of 19 (21 %) patients, partial healing in 16 %, and no change in 5 % of AZA-treated patients compared with 12 %, 18 %,24 %, and 35 % of BUD-treated patients, respectively (P = 0.001 and P = 0.04, for complete healing and worse/no change, respectively).

The SONIC trial [26] (The study of biologic and immunomodulation naïve patient in Crohn’s disease) included 508 CD immunosuppressive and biologic naïve patients. At week 26, colonic mucosal healing was achieved in 16.5 % of patients on azathioprine monotherapy . Further colonoscopies after longer treatment period were not performed. However, week 26 is probably too early to assess mucosal healing on thiopurine therapy since as written above, therapeutic effect after treatment initiation only begins after 12–17 weeks [22].

Thiopurines were shown to be more effective than placebo for perianal fistula closure. In a prospective study conducted on 83 CD patients, the efficacy of 2 years 6-MP treatment compared to placebo was assessed. Thirty one percent of patients treated with 6-MP achieved closure of their fistula compared with only 6 % in the placebo group (P < 0.001) [1]. Another observational study showed complete fistula closure in 39 % of patients and improvement in 26 % after >6 months of treatment [33].

In a meta-analysis of five clinical trials in which fistula closure served as a secondary end point, 54 % of patients treated with thiopurines had an improvement of their fistula status compared to 21 % of the placebo group (OR = 4.4 CI 1.5–13.2) [20].

Postoperative endoscopic recurrence of CD has been reported to be as high as 73 % 1 year post surgery [34], and clinical relapse rates have been reported to range from 22–55 % 5 years post surgery [35]. Facing the high recurrence rate, preventive treatment post-surgery for maintaining remission is usually warned.

The role of thiopurines in postsurgical maintenance of disease remission was assessed in various clinical trials [36–41]. A recent Cochrane database meta-analysis summarizing the results of seven randomized controlled studies (584 CD patients post bowel resection) was published in 2014 [36]. The studies included in the analysis compared treatment with thiopurines to placebo, 5-ASA and anti-TNF agents. A pooled analysis of two studies (n = 168 patients) showed 48 % relapse in thiopurine treated patients compared to 63 % relapse in the placebo group (RR 0.74, 95 % CI 0.58–0.94). A pooled analysis of five studies (n = 425 patients) showed 63 % clinical relapse in patients treated with thiopurines, compared to 54 % of 5-ASA patients (RR 1.15, 95 % CI 0.99–1.34). One study (n = 33) found decreased clinical (RR 5.18, 95 % CI 1.35–19.83) and endoscopic relapse (RR 10.35, 95 % CI 1.50). In summary, thiopurines are probably better than placebo for maintaining postsurgical remission. They might be less efficient than anti-TNFa agents—but data is scarce. Comparing their efficacy to 5-ASA agents is inconclusive.

Further research to assess their role in postoperative maintenance is warranted.

Early data favors early commencing of immunomodulatory treatment in order to alter disease progression and avoid complications. Data from two studies in pediatric patients [42, 43] showed better outcome in patients treated with thiopurines within 3 months from diagnosis. Punati et al. [42] compared outcome in newly diagnosed CD pediatric patients treated within 0–3 months from diagnosis (n = 150) (early) compared with patients treated within 3–12 months from diagnosis (n = 49) (late). Twelfth months from diagnosis only 22 % of the early group had received corticosteroids in the preceding quarter, compared to 41 % from of late group (P = 0.013). The number of hospitalizations per patient was significantly lower in the early group over the 2-year follow-up (P = 0.03). No difference was noted in the rates of remission, infliximab use over time, or surgery.

A recent study from the Groupe d’Etude Thérapeutique des Affections Inflammatoires du Tube Digestif [44] compared early administration of AZA (within 6 months from diagnosis) with conventional management (AZA only in cases of corticosteroid dependency, chronic active disease with frequent flares, poor response to corticosteroids, or development of severe perianal disease). Patients were prospectively followed for 3 years. Primary end point was steroids and/or anti-TNFa-free remission during follow-up. During follow-up, 61 % of the conventional treatment group needed AZA initiation, at median time of 11 months. Remission rates did not differ between the groups (67 % in the early treatment group vs 56 % in the conventional) (P = .69). Among secondary outcomes, higher percentage of early treatment group was free of perianal surgery than in the conventional management group (96 % ± 3 % and 82 % ± 6 % at month 36, respectively; P = .036). The cumulative proportion of patients free of intestinal surgery and anti-TNF therapy did not differ between groups. The study conclusion was that early treatment with AZA had no benefit over conventional therapy in increasing time of clinical remission. A recent Spanish multicenter study reported similar results [45].

The conflicting results between these studies might be explained by the different patients’ population (pediatric vs adults). However, the results of this prospective study challenge the common opinion regarding the yield of early treatment for all patients. Notably, almost 40 % of patients in the conventional treatment group were immunomodulators free for 3 years after diagnosis. Considering the drug adverse effects this is important data that should be taken into account while making decisions on treatment initiation. These data emphasize the importance of risk stratification and personalized—tailored—treatment while taking therapeutic decisions.

The question of treatment duration after achieving long-term clinical remission was addressed in two multicenter randomized placebo-controlled trials. The first was published in 2005 [46] by Lemann et al. included 83 CD patients in clinical remission on AZA therapy for at least 42 months. Patients were randomized to maintenance therapy or placebo and were followed for 18 months. The primary end point was clinical relapse. At 18 months 8 % at the AZA group compared to 21 % in the placebo group relapsed. The difference did not reach statistical significance (P = 0.195). C-reactive protein level >20 mg/L, time without steroids <50 months, and hemoglobin level <12 g/dL at baseline were found to be predictive of relapse in their multivariate analysis. The follow-up study published in 2009 [47] included 66 patients who stopped AZA treatment after median duration of 68 months treatment and 63 months remission. Thirty-two patients had a relapse after median follow-up of 54 months. The cumulative probabilities ± standard error of relapse at 1, 3, and 5 years were 14.0 % ± 4.3 %, 52.8 % ± 7.1 %, and 62.7 % ± 7.2 %, respectively. C-reactive protein concentration of 20 mg/L or greater, hemoglobin level less than 12 g/dL, and neutrophil count 4 × 10(9)/L or greater were associated increased risk of relapse. Retreatment with azathioprine alone achieved remission in 22/23 of the patients. A large retrospective multicenter study with 1176 IBD patients treated with AZA showed decreased disease exacerbations and need for steroid treatment within the first 4 years of AZA treatment (P < 0.001). Treatment discontinuation after 3–4 years did not lead to immediate disease exacerbation; however continuation beyond 4 years decreased disease activity and steroid consumption [48]. A recent multicenter retrospective cohort study from the UK [49] included patients with at least 3 years thiopurine use that were in sustained clinical remission and with a minimum of 1 year of follow-up post drug withdrawal. Median duration of thiopurine use prior to withdrawal was 6.0 years. Relapse occurred in 23 % of patients at 12 months and 39 % at 24 months. Elevated CRP at withdrawal was again associated with higher relapse rates at 12 months.

According to the European Crohn’s and Colitis Organization (ECCO) guidelines discontinuation of thiopurine maintenance therapy should be considered when the patient is 4 years in remission [50].

The conclusion driven from these studies is that if thiopurine treatment is well tolerated it should usually be continued for a long period. Risk factors for disease relapse as well as adverse events with long-term use as serious infections and risk of malignancy should be considered and discussed with the patient especially elderly patients, and decision should be individualized.

The yield of combination therapy of thiopurines with anti-TNFa was assessed in several multicenter prospective studies. Baert et al. [51] showed that concomitant immunosuppressive therapy significantly lowers the titers of anti-infliximab antibodies , thus leading to improved pharmacokinetics as demonstrated by significantly higher concentrations of infliximab 4 weeks after drug infusion. The SONIC study that compared biologic therapy versus thiopurines alone or in combination in biologically and immunomodulation naïve and relatively newly diagnosed patients [26], included 508 CD patients. Fifty-seven percent of patients receiving combination therapy were in corticosteroid-free clinical remission at week 26 compared with 44.4 % of patients on infliximab monotherapy and 30.0 % of patients on azathioprine monotherapy . (P < 0.001 for the comparison with combination therapy and P = 0.006 for the comparison with infliximab.) Mucosal healing at week 26 was achieved in 44 % of patients receiving combination therapy, compared with 30 % of patients on infliximab monotherapy and 16.5 % of patients on azathioprine monotherapy (P < 0.001 for the comparison with combination therapy and P = 0.02 for the comparison with infliximab). At week 50 remission rates were 72 %, 61 % and 55 %, respectively.

Another study by Lémann et al. [52] assessed steroid-free remission at week 24 after combination therapy of thiopurines with infliximab or thiopurine monotherapy in patients naïve to immunomodulators and in patients that failed on a stable dose thiopurines In all patients group remission rates were higher on combination therapy; thus combination therapy was shown to be more effective than thiopurine monotherapy for induction of remission in steroid-dependent CD patients. However, at week 52 only 27 % of patients on combination therapy in the thiopurine failure group were still in remission off steroids, compared with 52 % in the naive group.

A recent meta-analysis assessed the yield of combination therapy of thiopurine with Adalimumab (ADA) [53]. Results showed that ADA monotherapy was inferior to combination therapy (OR = 0.78 CI 0.64–0.96, P = 0.02) for induction of remission. However, there was no additional benefit of its addition for maintenance of remission or need for dose escalation over ADA monotherapy.

These evidence favors combination therapy in patients with moderate to severe CD, especially in immunosuppressive naïve patients. Combination therapy achieves higher rate of rapid disease control and limited tissue damage.

The superiority of combination therapy over monotherapy adds another therapeutic dilemma—whether and when to switch to monotherapy, and which of the treatments should be stopped. Van Assche et al. randomized 80 patients with at least 6 months of remission on combination therapy to infliximab monotherapy versus combination therapy [54]. After 104 weeks 60 % of the patients in the combination therapy group needed a change in infliximab doses or stopped therapy compared to 55 % of the infliximab monotherapy group (nonsignificant). CRP level were higher and the drug trough levels were lower, in the infliximab monotherapy group. They concluded that combination therapy beyond 6 months offers no clear benefit over scheduled infliximab monotherapy. The long-term implications of higher infliximab trough levels and decreased CRP levels in the combination group should be further studied. Predictors of relapse after azathioprine withdrawal were identified in a study by Oussalah et al. [55]. In this observational study patients were treated with combination therapy for at least 6 months before azathioprine was discontinued. At last follow-up (week 104), 35 out of 48 (73 %) patients were infliximab failure free. The disease-free probabilities were 85 % (±5 %) at 12 months and 41 % (±18 %) at both 24 and 32 months. Risk factor for disease relapse were in patients with duration of combination therapy of less than 27 months and/or the presence of biological inflammation (CRP > 5 and platelets count > 280,000).

Taken together, the decision whether and when to stop one of the medications on combination therapy should be taken individually, on the basis of disease history, side effects, personal preferences, and cost-effectiveness. Current data clearly indicates that combination therapy is more effective than monotherapy, and that thiopurine co-therapy is associated with higher infliximab trough levels.

Over 6 decades of thiopurine use in IBD provides a wide and very long-term safety profile. Generally, the medication is well tolerated. However, adverse events that require disease withdrawal occur in 10–18 % of patients. These include bone marrow toxicity, pancreatitis, and various allergic reactions as fever, flu-like illness, rash, and arthralgias [2, 20, 56–59]. Drug reaction may be related to drug metabolism and TPMT and TGN measurements can guide treatment (see below), but in 1–6.5 % of patients idiosyncratic reactions may occur [60]. Most of the nonspecific allergic reactions occur at treatment initiation, within 2–3 weeks, and tend to improve with time. Initiating treatment with 50 % of the optimal therapeutic dose can reduce the severity of the reaction and allow early intervention. Once tolerated, dose can be gradually increased with close monitoring of blood count, liver function tests, and TGN levels. Recent data suggests that dividing the optimal dose into two small daily doses can reduce some side effects as nausea [61, 62]. This treatment modification was also shown to improve long-term remission rate [63]. Patients who are intolerant to AZA may benefit from a switch to a metabolic descendant as 6-MP or TGN [64, 65]. Notably, post AZA/6-MP induced pancreatitis further treatment with 6-MP or AZA is contraindicated. However, tolerance and response to TGN was shown in these patients [66].

Hepatotoxicity may occur in 10–17 % of patients, and may be related to increased concentration of 6-MMPR [66]. Co-treatment with allopurinol was shown to shift the metabolic pathway to 6-TGN with normalization of liver enzymes [66, 67].

Rare disorders of the liver vasculature were described, including sinusoidal dilatation, nodular regenerative hyperplasia (NHR), fibrosis, peliosis hepatitis, and veno-occlusive disease (VOD) [68–70].

All immunomodulatory treatment carries an increased risk of infections. Thiopurines specifically increased susceptibility to viral infections [71]. In case of an acute infection, treatment with thiopurines can be withdrawn and renewed after recovery.

Risk of infections is accentuated in patients on combination therapy [71].

Mild leukopenia is observed in 5–25 % of patients treated with thiopurines and is associated with higher concentrations of 6 TGN metabolites. Leukopenia is most common at treatment initiation, but can occur at any time during treatment [57, 72].

Severe pancytopenia was reported in patients with TPMT deficiency . TPMT activity differs according to allelic polymorphism. In Caucasian population, 0.3 % is TPMT deficient with none or very little activity, 12.4 % have intermediate activity and 87.3 % have normal activity. Therefore, TPMT activity should be measured prior to therapy [73]. However, between 50 and 75 % of thiopurine related leukopenia occurs in patients with normal TPMT activity [74, 75]. Therefore, habitual monitoring during therapy is mandatory.

Most cases of myeloproliferative disorders in IBD patients on thiopurines are associated with EBV [76]. Recently, a few fatal cases have been reported in thiopurine treated young IBD EBV seronegative males associated with post-mononucleosis lymphoproliferation with or without hemophagocytic lymphohistiocytosis (HLH) [77]. Though the risk is very low, the ECCO Pathogenesis Scientific Workshop [78] recommends considering avoiding thiopurine treatment in IBD male patients less than 35 years who are EBV seronegative.

Increased risk of malignancy , specifically lymphoproliferative diseases is another concern during thiopurine therapy. The risk was shown to be ×4–×5 increased compared with the background population and is even more pronounced in elderly patients. Still, the absolute risk is low [79, 80].

Exposure to ultraviolet radiation was shown to increase the risk of non-melanoma skin cancer in thiopurine treated patients. A cohort study from South Africa [80] demonstrated an odds ratio of 5 for non-melanoma skin cancer. Among Caucasian patients the odds ratio was 12.4, and in non-Caucasians the risk was negligible.

Therefore, Caucasian patients on thiopurine therapy should be advised to avoid sun exposure and perform regular skin screening.

Recently, a large retrospective study assessed malignancy risk in patients treated with thiopurines and/or with anti-TNFa [81]. Data showed an increased risk of malignancy in IBD patients treated with thiopurines compared with patients treated with anti-TNF antibodies (hazard ratio 4.15; 95 % CI 1.82–9.44; P = 0.0007; univariate Cox regression). Patients aged over 50 years treated with thiopurines had 18.2 % tendency to develop malignancy, compared with 3.8 % of patients under 50 years of age (P = 0.0008). Treatment duration of more than 4 years was associated with an increased risk for skin cancer.

A recent prospective observational study from the CESAM study group [77] included 19,486 IBD patients with a total follow up of 49,736 patient-years, found that IBD patients with past exposure to thiopurines have a sevenfold increased risk to develop myeloproliferative disorders. The risk was not increased in IBD patients that stopped thiopurine treatment.

In the last decade a rare and aggressive extra-nodal form of non-Hodgkin’s lymphoma—hepatosplenic T-cell lymphoma (HSTCL) —was reported in IBD patients, in association with immunosuppressive therapy [82–84]. Out of over 200 cases reported in the literature, less than 40 are IBD patients. The disease is often fatal.

All of the IBD patients were all treated with thiopurines alone or in combination therapy with anti-TNF a (more that 50 % of cases). Median thiopurine exposure was 6 years. Most patients were young (90 % were under 35 years) males (90 % of the cases). The ECCO Pathogenesis Scientific Workshop from 2014 [78] concluded that though the overall risk of HSTCL is very low, it is to be taken into account in young (<35 years) men treated with combination therapy with more than 2 years of thiopurine intake.