1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References


Nodular regenerative hyperplasia (NRH) is increasingly being recognised in patients with inflammatory bowel disease (IBD). However, the pathogenesis and incidence of NRH in IBD, and the putative roles played by azathioprine (AZA), mercaptopurine (MP), or tioguanine (TG) remain unclear.


To summarise the data on the association between NRH and thiopurine therapy in patients with IBD.


A literature search was performed in PubMed and MEDLINE databases using the keywords ‘nodular regenerative hyperplasia AND (inflammatory bowel disease OR Crohn's disease OR ulcerative colitis) AND (azathioprine OR mercaptopurine OR tioguanine OR thioguanine).’ No time limit was placed on studies included.


Inflammatory bowel disease patients treated with AZA have a cumulative incidence of NRH of approximately 0.6% and 1.28% at 5 and 10 years, respectively, whereas those treated with high-dose TG (>40 mg/day) have a frequency of NRH of up to 62%, which is higher in patients with elevated liver enzymes and/or thrombocytopaenia than those without these abnormalities (frequency 76% vs. 33%). Conversely, low-dose TG therapy (<20 mg/day) is relatively safe, with no cases of NRH observed. NRH has also been found in 6% of operated thiopurine-naïve IBD patients. Male gender, older age, and stricturing disease/small bowel resection have been consistently identified as high-risk factors for NRH.


The pathogenesis of nodular regenerative hyperplasia in patients with IBD is complex and multifactorial involving disease-specific, genetic and iatrogenic risk factors. Clinicians should maintain a high index of suspicion for diagnosing nodular regenerative hyperplasia, especially in IBD patients with high-risk factors on thiopurine therapy, regardless of the presence of laboratory abnormalities.


  1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References

Nodular regenerative hyperplasia (NRH) is the commonest cause of non-cirrhotic idiopathic portal hypertension (PHT) in Western countries,[1-6] most commonly associated with haematological or autoimmune diseases[3, 4] (Table 1). Patients with NRH may be asymptomatic, with normal or only mild elevations of the liver function tests (LFTs), or may present with PHT (in 50–70%),[1, 7-11] usually in the background of well-preserved liver function.[3, 9, 12] Rarely, NRH might progress to end-stage liver disease necessitating transplantation.[9, 13] It has also been postulated that NRH is a pre-neoplastic condition, which might predispose some affected individuals to the development of hepatocellular carcinoma.[5, 14-16] Sinusoidal obstruction syndrome (SOS) is a related condition, characterised by a triad of weight gain/ascites, painful hepatomegaly and jaundice. Most cases (70–90%) resolve spontaneously, but a few may progress to NRH.[17] The incidence of NRH reported from autopsy studies is 0.5–2.6% and increases with age (5.6% in those >80 years).[2, 6, 14]

Table 1. Reported associations of SOS/NRH
  1. CREST, Calcinosis cutis; Raynaud's syndrome; oesophageal dysmotility; sclerodactyly; telangiectasia.

  2. a

    Factors associated with both SOS and NRH.

Autoimmune diseases
Rheumatoid arthritis (especially Felty's syndrome)a
Systemic lupus erythromatosusa
CREST syndrome
Progressive systemic sclerosis
Polyarteritis nodosum
Mixed connective tissue disease
Sjogren syndrome
Autoimmune haemolytic anaemia
Myasthenia gravis
Diabetes mellitus
Primary biliary cirrhosis
Mixed cryoglobulinaemia
Common variable immunodeficiency syndromea
Crohn's diseasea
Ulcerative colitisa
Coeliac disease
Multiple sclerosis
Liver transplanta
Renal transplanta
Heart transplant
Bone marrow transplanta
Infectious diseases
Miliary tuberculosis
Human immunodeficiency virus infectiona
Bacterial endocarditis
Hepatitis Ca
Haematological malignancies
Idiopathic thrombocytopaenic purpura
Myeloid metaplasia
Chronic myelogenous leukaemia
Chronic lymphocytic leukaemia
Hodgkin's lymphoma
Non-Hodgkin's lymphoma
Prothrombotic disorders
Polycythemia vera
Essential thrombocytosis
Sickle cell anaemia
Budd Chiari syndrome
Portal vein thrombosis
Anti-phospholipid syndrome
Factor V Leyden
Extrahepatic malignancies (especially metastatic)
Cytosine arabinoside
Toxic oil syndromea
Congestive heart failure
Pulmonary hypertension
Congenital (portal vein agenesis)

The primary event in the pathogenesis of NRH is believed to be sinusoidal endothelial injury and/or obliterative portal venopathy (OPV), resulting in alternating areas of hepatocyte hypoperfusion and atrophy and areas of adaptive hepatocyte hyperplasia.[18, 19] Liver biopsy is currently the only way of making a definitive diagnosis. To maximise diagnostic yield and avoid missing subtle NRH, adequate tissue samples (preferably using hepatic wedge or percutaneous biopsy) must be obtained, and examined by an experienced pathologist after staining with both haematoxilin and eosin and reticulin stains.3 However, liver biopsy is not perfect, with poor inter-observer agreement demonstrated for detecting NRH, even among expert histopathologists examining appropriately prepared samples (κ = 0.27 ± 0.05).[20] Histology typically shows diffusely distributed 1–3 mm nodules containing hypertrophied hepatocytes arranged in one to two-cell thick plates, with adjacent areas of atrophied single-plate thick hepatocytes, and little or no surrounding fibrous septae.[4, 6, 19, 21] Radiological appearances are often indistinguishable from cirrhosis.[4, 22] Contrast magnetic resonance imaging (MRI) has been proposed as an alternative to biopsy, but suffers from poor accuracy, (sensitivity and specificity of 77% and 72% respectively).[23-25] Portal pressure measurements reveal either an elevated or normal hepatic venous portal gradient (HVPG),[7, 8, 10, 11, 26, 27] postulated to depend on whether the primary insult is sinusoidal obstruction due to compression by regenerative nodules, or OPV respectively.[27]

In patients with inflammatory bowel disease (IBD), NRH has been predominantly associated with thiopurine [mercaptopurine (MP), azathioprine (AZA) or tioguanine (TG)] therapy, where it is postulated to be a dose-dependent adverse effect, mediated by 6-tioguanine nucleotides (6-TGN).[28-30] Following a study published over a decade ago that showed an alarmingly high incidence of NRH (62%) in IBD patients treated with high-dose TG,[31] formal guidelines were developed soon after on monitoring its use.[32] In more recent years, there have been increasing reports of NRH in IBD patients treated with AZA. NRH has also been reported in operated thiopurine-naïve IBD patients.[33] The purpose of this review is to summarise the data on the association between NRH, IBD and thiopurine therapy, and to highlight high-risk factors for this condition.

NRH in inflammatory bowel disease

  1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References

NRH in thiopurine-naïve patients with IBD

In the only study to date, de Boer et al. prospectively evaluated the prevalence of NRH in 433 thiopurine-naïve patients undergoing surgery for IBD-related complications.[33] NRH was found in 5 (6%) of 83 patients [68% with Crohn's disease (CD), 32% with ulcerative colitis (UC)] who underwent intraoperative liver biopsy after a median disease duration of 4 years (range 0–56). The authors concluded that IBD might be an independent risk factor for NRH, due to the higher prevalence seen compared to that in autopsy studies.[6, 14] However, this was a small non-controlled study, which enrolled only 19% of potentially eligible patients, and therefore, a sampling bias cannot be excluded. Furthermore, drug history was collected retrospectively, which made the study prone to recall bias. In addition, a high proportion of the patients (59%) were treated with corticosteroids prior to surgery. Prolonged therapy with contraceptive and androgenic steroids, as well as corticosteroids, has been implicated in the pathogenesis of NRH, hence the influence of these drugs in this study cannot be excluded.[5, 34-36]

NRH and/or SOS associated with AZA and MP

There are 16 case reports in the literature describing an association between NRH and AZA/MP therapy in IBD, involving a total of 19 patients.[37-52] Table 2 summarises the 10 reports published in English from which data could be extracted. In almost all cases, the diagnosis of NRH was made on liver biopsy following presentation with either laboratory abnormalities (mostly elevated LFTs or thrombocytopenia) or clinically evident PHT. Although resolution of laboratory abnormalities on cessation of AZA/MP was generally the rule, those with PHT had a variable clinical course; in some PHT persisted [38, 40, 42, 52] whereas in others it regressed/resolved on stopping AZA/MP.[39, 42, 47, 51] There are only two reports in the literature of SOS associated with AZA therapy in IBD patients.[49, 53]

Table 2. Summary of published case reports on nodular regenerative hyperplasia associated with AZA and MP therapy in inflammatory bowel disease
ReferenceStudy yearCountryIBD typeDisease duration (years)DrugDose (maximum)Therapy duration (months)SexAge (years)High-risk disease phenotypePresentationDiagnosis of NRHDisease course on stopping AZA/MP
  1. NRH, nodular regenerative hyperplasia; CD, Crohn's disease; UC, ulcerative colitis; IBD, inflammatory bowel disease; TI, terminal ileum; SB, small bowel; TJ, transjugular; PC, percutaneous; AZA, azathioprine; MP, mercaptopurine; LFTs, liver function tests; HVPG, hepatic venous portal gradient; PT, prothrombin time; PHT, portal hypertension; MRI, magnetic resonance imaging; CT, computed tomography; US, ultrasound; FBC, full blood count; IFX, infliximab; HCC, hepatocellular carcinoma; TIPS, transjugular intrahepatic portosystemic shunt.

León-Montañes et al.[47]2013SpainCD20AZA100 mg/day12Male47

Smoker (10/day)

Ileocolonic location (>100 cm TI resected)


Steroid dependent

Abnormal LFTs

PHT (thrombocytopaenia hepatosplenomegaly, ascites, varices)

HVPG normal

Liver biopsy (TJ)Treated with adalumimab with excellent response. Normal LFTs, endoscopy and normal CT/US at 1 year follow-up
Tuyama et al.[52]2013USACD11MP60Male23

Penetrating, obstructing

Ileocolonic location

Abnormal LFTs

PHT (thrombocytopaenia, splenomegaly, ascites, varices)


Liver biopsy (TJ) CD in remission, normal LFTs, PHT persists (ascites controlled on diuretics) at 16 months follow-up
Błogowski et al.[38]2011PolandCD3AZA100 mg/day36Female40Penetrating (partial SB resection)PHT (thrombocytopaenia, hepatosplenomegaly, varices, abdominal collaterals)Liver biopsy (PC)CD in remission, progression of PHT (on endoscopy and imaging) at 3 year follow-up
Calabrese et al.[40]2011USACD38MP200 mg/day60Male56

Ileal/perianal disease

Steroid dependent

Stricturing (previous TI resection)

PHT (thrombocytopaenia, splenomegaly, varices)

HVPG 20 mmHg

Liver biopsy (TJ)CD in remission, PHT persistent at 2 year follow-up
Bryant et al.[39]2010New ZealandUC24AZA3.5 mg/kg18Male54

Perianal/stenosing disease


PHT (thrombocytopaenia, ascites, splenomegaly)Liver biopsy (TJ)Thrombocytopaenia, ascites resolved (after paracentesis and diuretic therapy)
Seiderer et al.[51]2006GermanyCDAZA2.5 mg/kg5Male54Severe CD, with recurrent flare-ups



Liver biopsy

Contrast MRI (gadolinium chelate)

Normal LFTs and FBC, reversal of MRI changes of NRH at 12 months follow-up
de Boer et al.[43]2005The NetherlandsCDAZA2.5 mg/kg24Male35

Ileojejunal location

Perianal fistulae

TI stenosis, ileocecal resection

Elevated LFTs

PHT (thrombocytopaenia, varices, splenomegaly)

Liver biopsyNormalisation of all laboratory tests after cessation of AZA after 15 months
Daniel et al.[42] France           
Case 12005 UC1.5AZA200 mg/day6Male44Steroid dependent

Abnormal LFTs


Liver biopsy (PC)LFTs and platelets normalised after 19 and 34 months respectively) and CD remained in remission
Case 22005 UC6AZA175 mg/day6Male35Steroid dependentAbnormal LFTsLiver biopsy (PC)IBD in remission, normal LFTs and improved platelet count at 16 months
Case 32005 CD12AZA200 mg/day12Male46

Ileocolonic disease

Steroid dependency

Abnormal LFTs

PHT (thrombocytopaenia, varices)

Liver biopsy (PC)Treated with IFX, CD in remission, LFTs abnormal, platelets normal at 11 months follow-up
Case 42005 CD3AZA150 mg/day12Male26

Ileocolonic disease

Steroid dependency

Abnormal LFTs

PHT (thrombocytopaenia, hepatosplenomegaly, varices)

Liver biopsy (TJ)LFTs and platelets normalised after 3 years, but patient developed hypoalbuminaemia and ascites associated with CD flare up 1 month thereafter; resolved with remission
Russmann et al.[49]2001SwitzerlandUC12AZA200 mg/day24Male66Steroid dependency

Abnormal LFTs


Liver biopsy showed SOS. Repeat biopsy 1 month later showed NRH and HCC (despite stopping AZA)Underwent liver resection for HCC. Died 6 weeks later due to sepsis-related complications
Arnott and Ghosh[37]2000UKCD31AZA 216Female29

Anorectal disease

Multiple relapses, treated with steroids

Protocolectomy and TI resection >50 cm

PHT (variceal bleeding, treated by TIPPS -HVPG 15 mmHg)

Gastric varices

Low albumin (otherwise normal LFTs throughout treatment with AZA)

Liver biopsy (PC)CD in remission, varices eradicated by band ligation, HVPG 2 mmHg in TIPS at 2 years follow-up

In the largest case series to date, which recruited cases from 11 centres in Europe, Vernier-Massouille et al. reported 37 cases of biopsy-proven NRH following AZA therapy for a median duration of 48 months (range 6–187). The estimated cumulative incidence, calculated from one centre, was 0.50% at 5 years (95% CI, 0.11–0.89) and 1.25% at 10 years (95% CI, 0.29–2.21).[54] However, this was also a small observational study in which liver biopsy was performed entirely at the discretion of individual gastroenterologists from the centres involved, with no pre-defined criteria. In a subsequent study by the same group, Seksik et al. evaluated the incidence of NRH in 1888 IBD patients treated with AZA.[55] NRH was found in 15 patients (0.8%), after a median duration of therapy of 36 months (IQR 24–96). The estimated cumulative incidence was 0.64% ± 0.235 and 1.28 ± 0.45% at 5 and 10 years respectively. Of note, most of the data in this study were collected retrospectively over a period of 20 years, hence was subject to recall bias. The authors concluded that the incidence of NRH reported may have been underestimated as only symptomatic patients (with thrombocytopaenia, splenomegaly or clinical signs of PHT) underwent biopsy, and there was no systematic radiological or histological screening during follow-up.[54, 55] A more recent Spanish study, published only in abstract form, evaluated NRH in 538 patients treated with AZA/MP for over 15 years.[56] Overall, six patients (1.11%) were diagnosed with NRH, based on the presence of thrombocytopaenia, LFT abnormalities and/or ultrasonographic evidence of PHT. As liver biopsies were not done, the frequency of NRH reported is likely to have been underestimated.

Epidemiological studies describing AZA and/or MP-induced hepatotoxicity (defined as LFT elevations ≥2× the upper limit of normal) in IBD patients abound in the literature.[30, 57-59] Sporadic cases of NRH have been reported in some of the studies. In a long-term follow-up study evaluating the efficacy of AZA/MP in patients with CD, Bouhnik et al. found one case of NRH (0.3%) and one of peliosis on liver biopsy performed for persistent LFT derangements, after a median duration of therapy of 25 months (range 7–166).[60] A more recent study by López-Martin et al. found hepatotoxicity in 9 (2.4%) IBD patients treated with AZA, which resolved on discontinuing treatment in all except one patient subsequently diagnosed with NRH following liver biopsy.[61] In the largest cohort study to date Chaparro et al. found hepatotoxicity in 4% and thrombocytopaenia in 0.2%, but no cases of NRH, in 3931 IBD patients treated with AZA/MP for a median duration of 110 days (range 0–86).[57] In general, the majority of these studies were retrospective and primarily designed to detect the much more common AZA-induced hepatotoxicity (which has an incidence of up to 32%,[62] and is thought to be mediated by the methylation products of MP, 6-methylmercaptopurine and their ribonucleotides), rather than the comparatively rare NRH.

NRH and/or SOS associated with TG

Initial studies evaluating TG therapy in IBD patients intolerant or refractory to AZA/MP showed promising results, with good efficacy and acceptable short-term toxicity.[63, 64] However, this early optimism was short-lived as it was soon overshadowed by worrying reports of NRH developing in some patients. The first few cases of NRH in association with TG therapy in IBD patients were reported by Shastri et al., in three male IBD patients (two with CD, one with UC) who underwent liver biopsy due to elevated LFTs, after a mean duration of treatment of 16 months.[65, 66] Subsequently, a retrospective study by Dubinsky et al. involving 111 patients with IBD treated with TG found biopsy-proven NRH in 16 (62%) out of the 26 patients that underwent liver biopsy, which the authors speculated was due to an idiosyncratic mechanism.[31] NRH was found more frequently in symptomatic patients (with elevated LFTs and/or thrombocytopaenia) compared to asymptomatic ones (76% vs. 33% respectively). Although the dose of TG used in this study was not specified, measured median 6-TGN levels were approximately 1230 pmol/8 × 108 RBC (threefold higher than the normal upper limit following AZA/MP therapy) implying that doses used were in excess of 40 mg/day. A subsequent report by the same group, using the same patient cohort, reported NRH in 20 (53%) out of 37 biopsies performed after ≥1 year of TG therapy.[67] In a small prospective study investigating the use of TG for maintenance therapy in patients with chronic active CD who had failed previous steroid and/or AZA therapy, Herrlinger et al. found one case of PHT (presenting with elevated LFTs, splenomegaly and thrombocytopaenia), from presumed NRH, among 16 patients treated with TG (dose 20–40 mg/day) for ≥1 year.[68]

Seiderer et al. evaluated TG induced liver toxicity using a combination of MRI and liver biopsy in 45 patients with IBD treated with TG (dose 40–80 mg/day).[24] Histological findings consistent with NRH were described as ‘definite’ in 17.8% and ‘possible’ in a further 17.8%. On the other hand, Ferlitsch et al. performed liver biopsy and HVPG measurements in IBD patients treated with TG 40 mg/day (range 40–80).[7] Six (25%) patients with adequate biopsy samples obtained had NRH, and all had significantly increased HVPG compared to those without NRH (median HVPG 7 mmHg vs. 3 mmHg; P < 0.001). Using an internet-based online survey, Teml et al. systematically surveyed hepatotoxicity in IBD patients treated with TG over a period of 4 years, using a combination of LFT derangements, imaging (computed tomography and MRI) and liver biopsy.[69] NRH was diagnosed in 16 (27.1%) of patients who underwent liver biopsy, after a median duration of TG treatment of 118 weeks (range 21–169). A major limitation of all the above studies is that a significant proportion of the patients were treated with AZA/MP for prolonged periods of time before commencing TG, and liver biopsy was not performed prior to its initiation. For example, in the studies by Dubinsky et al. and Seiderer et al., patients included had been exposed to AZA or MP for a median duration of 12 months (range 0.5–96) and 11 months (range 0.5–60) respectively.[24, 31] In Dubinsky et al.'s study, 40% of the patients had evidence of preceding hepatotoxicity as a result of this.[31] Therefore, it is not possible to completely exclude the influence of AZA/MP in the pathogenesis of NRH cases found in these studies.

On the other hand, studies in which patients were treated with low doses of TG (20–40 mg) either found few or no cases of NRH in their cohorts. A retrospective study by de Boer et al. evaluating the tolerability and safety of low-dose TG (0.3 mg/kg) in 95 AZA/MP intolerant IBD patients found one case of splenomegaly due to PHT (among the 51 patients who underwent abdominal ultrasound) and two of thrombocytopaenia, after 1 year of therapy.[70] Gilissen et al. found no case of NRH and/or PHT on biopsy and MRI in 14 patients treated with TG (median dose 18.8 mg/day) for a minimum of 36 months.[71] Notably, in this study, AZA or MP was used for only a short duration prior to initiation of TG therapy. Similarly, Ansari et al. found no cases of NRH following biopsy/MRI in 11 IBD patients treated with TG (median dose and duration of 40 mg and 21.5 months respectively) for >1 year.[72] One patient developed both clinical and radiological evidence of PHT, which subsequently resolved after cessation of TG. All the patients in this study had 6-TGN levels in the intermediate range (805 pmol/8 × 108 RBC), apart from the patient who developed PHT (TGN level 1714 pmol/8 × 108 RBC). The authors postulated that this was due to inter-individual variation in the pharmacokinetics of TG. de Boer et al. found no case of NRH in 26 (93%) out of 28 IBD patients treated with TG (mean dose, TGN level, duration of use 19.5, 564 pmol/8 × 108 RBC, 38 months respectively).[73] Likewise, an observational Swedish study of 23 thiopurine-intolerant/resistant patients with severe CD treated with TG 40 mg/day (range 20–60) found no evidence of NRH on MRI (using a specific NRH protocol) in 13 patients treated for ≥6 months, and none among the only two patients who underwent liver biopsy.[74] van Asseldonk et al. evaluated 46 patients with UC treated with a median dose of TG of 20 mg/kg/day (range 18–24) over a median duration of 22 months (range 0.3–72.0), and also found no cases of NRH in the 12 patients who underwent liver biopsy.[75]

Taken as a whole, the available data appear to show a dose-dependent effect of TG in NRH causation, with no cases of NRH found in IBD patients treated with doses of ≤20 mg/day (corresponding to median 6-TGN levels of 600 pmol/8 × 108 RBC), whereas a frequency of NRH of 0–27% has been reported in those treated with TG 40–80 mg/day (corresponding to 6-TGN levels achieved >1000 pmol/8 × 108 RBC).[75] However, most of the data are from small, single-centre, prospective but non-controlled studies, in which only a few symptomatic patients underwent liver biopsy. Furthermore, to date, there have been no head-to-head (prospective) studies performed comparing low-dose vs. high-dose TG in comparable group of IBD patients, hence the case for dose-dependent TG toxicity in NRH is still inconclusive. Similar to the case with AZA/MP treated patients, the natural history of NRH in these patients is also unknown, with some studies showing regression/resolution of SOS/NRH on stopping TG,[68, 76, 77] while others showed progression regardless of this.[69] Table 3 summarises studies that have evaluated the occurrence of NRH in IBD patients. Surprisingly, there has been only one case report of SOS in a CD patient, reported following 14 months of treatment with TG (40 mg/day).[78]

Table 3. Studies reporting the prevalence of nodular regenerative hyperplasia in patients with inflammatory bowel disease
ReferenceYearStudy designSample size (n)IBD type Thiopurine dose, median (range)Thiopurine therapy before NRH diagnosis, duration (range)Liver biopsy requirement in study protocolCriteria for performing liver biopsyPatients biopsied, n (% of total)Prevalence/incidence rate of NRH
  1. NRH, nodular regenerative hyperplasia; CD, Crohn's disease; UC, ulcerative colitis; IBD, inflammatory bowel disease; AZA, azathioprine; MP, mercaptopurine; TG, tioguanine; 6-TGN, 6-tioguanine nucleotides; LFTs, liver function tests; PHT, portal hypertension; MRI, magnetic resonance imaging; RBC, red blood cells; IQR, interquartile range.

  2. a

    One patient with elevated LFTs diagnosed with PHT (splenomegaly, thrombocytopaenia) due to presumed NRH; patient declined liver biopsy.

  3. b

    One patient with elevated LFTs diagnosed with PHT (based on splenomegaly, among 51 patients that underwent abdominal ultrasound), and two developed thrombocytopaenia, due to presumed NRH.

Thiopurine-naïve patients
de Boer et al. [33]2008Prospective433

CD 68%

UC 32%

NoneNot applicableYesThiopurine-naïve patients undergoing gastrointestinal surgery for UC or CD83 (19%)6% (5/82)
Patients treated with TG
Herrlinger et al. [68]2003Prospective16CD 100%

20–40 mg/day

6-TGN levels 1067 pmol/8 × 108 RBC (470–1631)

44 weeksNoNot applicableNone6%a
Dubinsky et al.[31]2003Retrospective111

CD 69%

UC 27%

Up to 100 mg/day

6-TGN levels 1230 pmol/8 × 108 RBC (502–2310)

10 months (1–28)Yes

Patients with elevated LFTs and/or thrombocytopaenia (26%)

Routine screening in patients without blood abnormalities (74%)

26 (23%)62% (76% vs. 33% in patients with or without laboratory abnormalities respectively)
Geller et al.[67]2004Retrospective111

CD 69%

UC 27%

Not reported

6-TGN levels 1230 pmol/8 × 108 RBC (502–2310)

≥1 yearYes

Symptomatic patients with elevated LFTs and/or thrombocytopaenia

Routine screening in asymptomatic patients

38 (34%)53% (76% vs. 33% in patients with or without laboratory abnormalities respectively)
de Boer et al. [70]2005Retrospective95

CD 56%

UC 44%

Mean 24.6 mg/day (20–40). Mean 6-TGN level 540 pmol/8 × 108 RBC≥1 yearNoNot applicableNone2%b
Seiderer et al.[24] 2005Prospective45

CD 71%

UC 29%

40–80 mg/day (82% used maximum 40 mg/day)70 weeks (8–119)YesIBD patients treated with TG for ≥8 weeks45 (100%)36% (definite NRH in 18%; possible NRH in 18%)
Ferlitsch et al. [7]2007Prospective26

CD 89%

UC 8%

Median 40 mg (20–80)38 months (12–45)YesIBD patients treated with TG for ≥1 year26 (10%)25%
Teml et al.[69]2007Prospective304

CD 81%

UC 16%

20–40 mg118 weeks (21–169)YesUnspecified60 (20%)27%
Gilissen et al. [71]2007Prospective13

CD 62%

UC 38%

18.8 mg – mean (6.6–20)36 months (27–45)YesIBD patients treated with TG for ≥2 years13 (100%)0%
Ansari et al. [72] 2008Retrospective30CD 100%

40 mg (20–60)

Median 6-TGN 807 pmol/8 × 108 RBC (105–2545)

33 months (12–56)YesPatients treated with 6-TG for >1 year11 (37%)0%
de Boer et al. [73]2008Prospective28

CD 61%

UC 39%

Mean 19.5 mg/day (s.d. 5.5)38 months (30–53)YesPatients treated with TG for >30 months28 (100%)0%
Almer et al. [74] 2009Prospective23CD 100%40 mg (20–60) (dose adjusted to 6-TGN ≤1200 pmol/8 × 108 RBC)259 days (15–2272)NoNot applicable2 (9%)0% (56% of the patients underwent MRI using protocol for detecting NRH)
van Asseldonk et al.[75]2011Prospective cohort46UC 100%

0.3 mg/kg (s.d. 0.07)

Median 6-TGN level 278 pmol/8 × 108 RBC (68–492)

36 months (10–44)YesPatients treated with TG for ≥6 months12 (26%)0%
Patients treated with AZA and/or MP
Bouhnik et al.[60] 1996Prospective157CD 100%

AZA 2 mg/kg

MP 1.5 mg/kg

25 months (7–166)NoNot specifiedNot specified0.3%
Vernier-Massouille[54] 2007Case-control37

CD 84%

UC 16%

AZA 2 mg/kg/day (1.5–3)34 months (3–159)YesPHT (thrombocytopaenia, splenomegaly, clinical signs)37 (100%)

0.5% at 5 years

1.25% at 10 years

Seksik et al.[55] 2011Cohort study1888

CD 77.1%

UC 19.4%

AZA starting dose 2–2.5 mg/kg/day36 months (IQR 24–96)YesPHT (thrombocytopaenia, splenomegaly, clinical signs)Not specified

Cumulative incidence: 0.64% at 5 years

1.28% at 10 years

Lopez-Martin et al.[61] 2011Retrospective377

CD 55%

UC 42%

AZA 2.5 mg/kg

MP 1.5 mg/kg

9 months No

Persistent LFT elevation

PHT on imaging and endoscopy

Mesonero et al. [56]2013Retrospective538UnspecifiedUnspecified46 ± 39 monthsNoNot applicable2 (0.4%)1.11%

Risk factors for NRH in IBD patients

  1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References

Several risk factors have now been identified that may predispose IBD patients to NRH. The published case reports on NRH associated with AZA/MP involved predominantly male (87%), IBD patients (78% with CD), with a median age of 44 years (range 23–66). The majority had a phenotype indicative of severe/progressive IBD (terminal ileal location, stricturing, obstructing, penetrating, fistulating or perianal disease, extensive bowel resections, or steroid dependency)[79] (Table 1). Some of these factors were also identified in the prospective and retrospective studies, all of which predominantly enrolled patients with CD (proportion of patients with CD in the individual studies ranged from 55% to 100%). In the study by de Boer et al., older age at biopsy was significantly correlated with NRH in operated thiopurine-naïve IBD patients (mean age 56 vs. 37 years in those with and without NRH respectively; P = 0.015).[33] Similarly, in Teml et al.'s study, patients with NRH following TG therapy were significantly older compared to those without NRH (median age 44 vs. 35 years respectively; P = 0.002), and multivariate analysis confirmed age as the only independent (albeit weak) risk factor for NRH (OR 1.10, 95% CI 1.03–1.17, P = 0.005).[69] On the other hand, Dubinsky et al. and Geller et al. showed that male gender and preferential 6-methylmercaptopurine production in AZA-treated IBD patients were significantly associated with laboratory abnormalities [odd ratio 2.9 (95% CI 1.1–7.2; P < 0.03) and 3.0 (95% CI 1.2–7.4; P < 0.04) respectively], and hence NRH.[31, 67] Vernier-Massouille et al. showed that male gender and stricturing IBD were the only risk factors associated with NRH on AZA therapy.[54] Similarly, Seksik et al. showed that the only independent risk factors for NRH in AZA-treated IBD patients were male gender (hazard ratio 8.5, 95% CI 1.9–37.9) and small bowel resection of ≥50 cm (hazard ratio 6.6, 95% CI 2.2–20.0).[55] Male patients who had undergone previous extensive small bowel resection were estimated to have the highest cumulative risk for NRH (approximately 10% after 10 years).[55]

Emerging data suggest that genetic factors may also play a role in the pathogenesis of NRH. de Boer et al. hypothesised that NRH might only develop in patients treated with thiopurines carrying inactivating thiopurine S-methyltransferase (TPMT) single nucleotide polymorphisms (SNPs), resulting in high 6-TGN levels.[28] Consistent with this, Blogowski et al. recently reported a CD patient diagnosed with NRH on AZA therapy who was heterozygous for the TPMT*3A SNP and had concomitant factor V Leiden deficiency.[38] Similarly, Buster et al. described a liver transplant recipient with NRH on AZA therapy who was found to be heterozygous for both the TPMT*3C and inosine triphosphate pyrophosphatase (ITPA) (94C>A) SNPs.[80] Inactivating TPMT SNPs have also been shown in children with leukaemia treated with TG who subsequently developed NRH.[81] However, the evidence for a pharmacogenetic predisposition to NRH is still inconclusive, with some studies showing no association with inactivating TPMT SNPs or TPMT enzyme levels, in both IBD patients[42, 43, 54] and in children with leukaemia.[81-83] Due to the complexity of thiopurine metabolism, it is more likely that there is a complex interplay between SNPs in the genes encoding the multiple enzymes involved, including TPMT, ITPA, glutathione synthetase and methyltetrahydrofolate reductase. Based on current data, it can be concluded that IBD patients at highest risk of developing NRH are usually older males with complicated/severe IBD (predominantly CD), particularly those with persistent, unexplained laboratory abnormalities following exposure to thiopurines.

Postulated mechanisms of NRH in IBD patients

  1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References

The pathogenesis of NRH in IBD patients is poorly understood but is likely to be multifactorial. IBD is associated with several processes that increase in intensity with severity of inflammation, including immune activation, vasculitis, endothelial cell dysfunction and thrombogenesis.[84-87] High tumour necrosis factor alpha (TNF-α) levels, particulary in CD patients, and steroid therapy (independent of the underlying inflammation) might induce NRH by promoting thrombogenesis and OPV.[88] Similarly, extensive small bowel involvement or resection might promote OPV by causing malabsorption of vitamins B12, B6 and folic acid, with resultant hyperhomocystenaemia, which might promote thrombosis in the portal microvasculature. In addition, endothelial-mesenchymal transition (which has been shown to play an integral role in the pathogenesis of intestinal fibrosis and fistulating IBD),[89] persistent bacteraemia,[90] and gut-derived/genetic prothrombotic factors in IBD patients might promote OPV. Thiopurine therapy, particularly high-dose TG, might act synergistically with these factors in causing NRH, possibly mediated by high 6-TGN levels (>1000 pmol/8 × 108 RBC)[28, 29], in genetically predisposed individuals.[28]

Azathioprine has been postulated to damage hepatic sinusoidal endothelial cells and/or small hepatic and portal venules.[18] DeLeve et al. demonstrated that AZA was selectively toxic to murine sinusoidal endothelial cells in vitro (but not hepatocytes), by depleting cellular glutathione stores.[91] Conversely, studies in children with leukaemia treated with thiopurines,[92] as well as a recent study in a mouse model of SOS,[93] show that thiopurine-induced NRH might be specific to TG, but not AZA or MP. It has been hypothesised that this is due to the requirement for AZA and MP to undergo multiple enzymatic steps in their metabolism, which results in persistently low levels of 6-TGNs in hepatocytes, in contrast to TG which undergoes direct conversion into 6-TGN by hyoxanthine guanine phosphoribosyltransferase,[93-95] with resultant toxic levels of 6-TGNs. However, the assumption that NRH is mediated by high 6-TGN levels, as measured in peripheral blood RBCs, might be an oversimplification. RBC's are inert and metabolise individual thiopurines differently compared to other cells. Therefore, RBC 6-TGN levels might not necessarily reflect levels achieved in hepatic sinusoidal and endothelial cells, the presumed target cells in NRH.[95] Figure 1 summarises the potential pathogenetic mechanisms of NRH in IBD.


Figure 1. Summary of postulated potential mechanisms underlying the development of NRH in IBD. The pathogenesis of NRH in IBD is likely complex and multifactorial, involving host/disease-specific factors, as well as iatrogenic factors, acting through a final common pathway of OPV and/or SOS. Acute or chronic bowel inflammation promotes various immunological and prothrombotic mechanisms, proportional to the severity of inflammation. Hyperhomocystenemia may result from malabsorption of vitamin B12, B6 or folic acid due to extensive enteropathy or bowel resection. Recurrent severe exacerbations of IBD may cause cumulative injury, eventually culminating in NRH in the genetically predisposed. TNF-α, tumour necrosis factor alpha; VEGF, vascular endothelial growth factor; MMPs, matrix metalloproteinases; EndoMT, endothelial-mesenchymal transition; TGNs, tioguanine nucleotides; TG, tioguanine; AZA, azathioprine; MP, mercaptopurine; TPMT, thiopurine methyltransferase; SNPs, single nucleotide polymorphisms; OPV, obstructive portal venopathy; SOS, sinusoidal obstruction syndrome; NRH, nodular regenerative hyperplasia.

Download figure to PowerPoint


  1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References

The pathogenesis of NRH in patients with IBD is complex and multifactorial, involving disease-specific, genetic and iatrogenic risk factors. Despite accumulating evidence in support of a causative role of thiopurine therapy in its pathogenesis, the cellular and molecular mechanisms remain poorly understood. Severe, complicated IBD, especially CD, might be an independent predisposing risk factor for NRH. Clinicians should maintain a high index of suspicion for diagnosing NRH, particularly in patients with known high-risk factors, regardless of symptoms. Further studies are needed to clarify the incidence, pathogenesis, natural history and genetic predisposing factors to NRH, as well as the relative roles played by AZA and MP compared to TG in its causation.


  1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References

Guarantor of the article: Crispin Musumba.

Author contributions: Crispin Musumba performed the literature review, analysed the data, wrote the paper and approved the final version of the manuscript.


  1. Top of page
  2. Summary
  3. Introduction
  4. NRH in inflammatory bowel disease
  5. Risk factors for NRH in IBD patients
  6. Postulated mechanisms of NRH in IBD patients
  7. Conclusions
  8. Authorship
  9. Acknowledgement
  10. References