Article Text

Download PDFPDF

Azathioprine and the neurologist
  1. Matthew McWilliam,
  2. Usman Khan
  1. Neurology, St George's University Hospital, London, UK
  1. Correspondence to Dr Usman Khan, Neurology, St George's University Hospital, London SW17 0QT, UK; usman.khan4{at}


Neurologists are very familiar with using corticosteroids and are aware of their considerable risk of adverse effects with prolonged use. Thus, we frequently consider alternative immunosuppression or corticosteroid sparing agents. However, unlike other specialties, such as rheumatology, there are few indications for corticosteroid-sparing agents in neurology and so our experience is less extensive; even these indications may reduce further as more disease-modifying treatments become available for neurological conditions. Azathioprine is perhaps the most commonly used corticosteroid-sparing agent in neurology. This review aims to remind neurologists of important aspects of azathioprine prescribing, focussing on enhancing patient safety and clinician confidence in its prescribing.

  • biochemistry
  • clinical neurology
  • liver disease
  • medicine
  • myasthenia

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is Azathioprine?

azathioprine belongs to the thiopurine drug group. It is a widely used antimetabolite, immunosuppressant prodrug that inhibits purine synthesis, slowing cell proliferation within the immune system. It was originally developed in transplant surgery to prevent graft rejection. It is widely used in rheumatology—where there are three conditions that have the only UK-licensed indications: dermatomyositis, polymyositis and systemic lupus erythematosus—and in gastroenterology for managing inflammatory bowel disease. Neurologists do see inflammatory myopathy and lupus cases but often also involve rheumatologists early in their management. Neurologists may also use azathioprine as unlicensed treatment in various autoimmune and other immune-mediated neurological conditions, either alone with corticosteroids, and with varying success (box 1).1–12 In neurology, it is most commonly used for treating myasthenia gravis.

Box 1

Licenced use and some examples of unlicensed use of azathioprine in neurology




Systemic lupus erythematosus.3


Myasthenia gravis.4

Lambert-Eaton myasthenic syndrome.5

Chronic inflammatory demyelinating polyneuropathy.6

Multifocal motor neuropathy.7

Multiple sclerosis.8

Neuromyelitis optica.9


Cerebral vasculitis.11

Autoimmune encephalitis.12

azathioprine metabolism

azathioprine is quickly absorbed from the gut and does not cross the blood–brain barrier. It has a short half-life (about an hour) due to rapid non-enzymatic conversion to 6-mercaptopurine, the active compound, and other imidazole derivatives. 6-mercaptopurine has a longer half-life (3–5 hours) and is metabolised by three separate pathways (figure 1), each with clinical relevance:

Figure 1

Metabolism of azathioprine. HPRT, hypoxanthine-guanine phosphoribosyl-transferase; TPMT, thiopurine methyltransferase; XO, xanthine oxidase; 6-MP, 6-mercaptopurine; 6-MMP, 6-methyl mercaptopurine; 6-TG=6-thioguanine.

  • First, 6-mercaptopurine is metabolised to 6-thioguanine nucleotides, which govern the immunosuppressive effect through incorporation into DNA. Mercaptopurine and tioguanine have been marketed as individual immunosuppressive drugs for different clinical scenarios.

  • Second, 6-mercaptopurine is competitively degraded by xanthine oxidase to 6-thiouric acid, the basis of an important drug interaction with allopurinol (see later).

TPMT activity depends on genotype: 89% of people have the wild-type and so normal TPMT enzyme activity; 11% are heterozygous for TPMT mutations and have reduced enzyme activity; 0.3% are homozygous for mutations giving a complete deficiency of TPMT, and have the highest risk of the most severe side effects with standard azathioprine doses.13 Very high serum TPMT levels can lead to the preferred conversion of 6-mercaptopurine to 6-methylmercatopurine, a process called hypermethylation, putting the patient at risk of complications (see below).

TPMT phenotyping by enzyme assay is preferred to genotyping, as the assay can detect all patients who are completely deficient, irrespective of whether they have a rare or common variant TPMT genotype. Although clinicians appropriately focus on measuring serum TPMT levels before starting azathioprine, TPMT polymorphisms probably account for only 10% of thiopurine toxicity14; 50%–75% of all patients developing leucopenia on azathioprine have normal serum TPMT levels. In Asian and some other population groups, there are recommendations to genotype for nudix hydrolase (NUDT15) mutations, especially in those with recurrent severe myelosuppression with azathioprine.15 This enzyme probably dephosphorylates the active metabolites of thiopurines; mutations can cause toxicity through build-up of these metabolites.

Metabolite monitoring in Azathioprine prescribing

Almost all the recommendations for neurology have been extrapolated from the gastroenterology literature.

Measurements of 6-thioguanine nucleotides, along with 6-methylmercaptopurine concentrations, may serve as a surrogate marker of epigenetic and genetic factors influencing thiopurine metabolism; in this setting they can help with the practical application of therapeutic drug monitoring of azathioprine. Monitoring of these metabolites (to help balance efficacy and toxicity) is well established in patients with inflammatory bowel disease, and positively impacts on outcome in this setting. 16

A serum concentration of 6-thioguanine above 400 pmol/8×108 correlates with myelosuppression; values between 235 and 450 pmol/8×108 correlate with clinical efficacy. A serum concentration of 6-methylmercaptopurine above 5700 pmol/8×108 suggests an increased risk of hepatotoxicity.17 6-methylmercaptopurine causes a spectrum of liver injury, ranging from an asymptomatic rise in aminotransferases, to cholestatic hepatitis and vascular liver disease. This resembles thiopurine-associated liver injury, although with less clear understanding of the underlying mechanisms18; thiopurine-associated cholestatic hepatitis most likely occurs as part of an idiosyncratic hypersensitivity reaction (see box 2).

Box 2

A case of azathioprine-induced cholestatic hepatitis

A 71-year-old man took treatment for both myasthenia gravis (serum anti-acetylcholine receptor antibody positive with no evidence of thymoma) and type 2 diabetes mellitus. His corticosteroid treatment, although very effective, upset his glycaemic control culminating in a foot drop. His serum thiopurine methyltransferase concentration was low at 60 mU/L (normal 68–150) and so he started low-dose azathioprine at 0.5 mg/kg (50 mg) daily. On review at a month, he reported malaise, anorexia, nausea, insomnia and generalised pruritus. He had leucopenia and significantly deranged liver function tests although with preserved liver synthetic function (serum albumin and clotting). After immediately stopping the azathioprine, he slowly improved clinically and biochemically. On full recovery of his liver function, we reduced his prednisolone to a maintenance dose and started mycophenolate mofetil (figure 2).

Hepatotoxicity is a rare (annual incidence 0.1%–2%) and heterogeneous adverse effect of azathioprine. Cholestatic reactions, as here, comprise three-quarters of cases; the remainder have either mild, transient and asymptomatic rises in serum aminotransferase concentrations, or chronic hepatic injury marked by peliosis hepatis, veno-occlusive disease or nodular regenerative hyperplasia.28 Azathioprine-induced acute cholestatic hepatitis typically presents within a few months of starting or increasing the drug dose, but sometimes as early as 2 weeks; it is self-limiting once the azathioprine is stopped. The mechanism is uncertain but is probably an idiosyncratic hypersensitivity reaction to the nitroimidazole component of azathioprine.

Figure 2

Liver function test changes in (bilirubin normal range (NR) 0–12 µmol/L), albumin (ALB, NR: 33–55 g/L), alanine transaminase (ALT, NR: 0–52 U/L), alkaline phosphatase (ALP, NR: 30–160 U/L), gamma-glutamyl transferase (GGT, NR: 0–64 U/L) and lymphocyte count (NR: 1.1–4.0×109 /L)) over time on azathioprine.

Hypermethylation, usually but not always occurring with a very high TPMT serum level, causes shunting of 6-mercaptopurine to 6-methylmercaptopurine, in preference to its conversion to 6-thioguanine.19 This reduces the serum concentration of 6-thioguanine nucleotides, reducing azathioprine’s myelosuppressive effect, and also increases the serum concentration of 6-methylmercaptopurine, thereby increasing hepatotoxicity risk. Hypermethylation is defined as a serum concentration ratio of 6-methylmercaptopurine: 6-thioguanine of more than 11; a high ratio correlates with azathioprine treatment failure.

Azathioprine metabolite concentrations (and hence the risk of myelosuppression and hepatotoxicity) vary significantly between individuals. However, although it is important to measure these metabolites to guide azathioprine prescribing, this does not replace blood monitoring. Nevertheless, neurologists should perhaps more often consider measurement of azathioprine metabolites.20

Practicalities and the governance around Azathioprine use in neurology

Table 1 highlights some important clinical and governance-related points for neurologists to consider when starting patients on azathioprine.

Table 1

Recommendations for the safe use of azathioprine in neurology patients

Dose and efficacy

Most neurologists aim for an eventual therapeutic dose of azathioprine of 2.5 mg/kg/day. Thus, the British National Formulary recommends an initial azathioprine dose of 0.5–1.0 mg/kg/day for myasthenia gravis, increased over 3–4 weeks to 2.0–2.5 mg/kg/day. It is essential to measure the serum TPMT level before starting azathioprine; anyone completely deficient in this enzyme should not receive the drug. Dosing by TPMT level can be extrapolated from gastroenterology literature, which suggests a starting dose of 1 mg/kg/day for patients with intermediate levels of TPMT, and 2.0–2.5 mg/kg/day for patients with normal or high TPMT levels.21 In all such cases, careful blood monitoring remains essential (see below). It can take up to 18 months for the therapeutic and corticosteroid-sparing effect of the drug to manifest.

Patients at the extremes of body weight (body mass index below 18 or above 30 kg/m2) are at risk of toxicity, in both situations due to inadvertent overdosing of azathioprine. In these setting, most regimens suggest a lower dose and increasing incrementally every 2–4 weeks towards a target weight-based dose, guided by careful blood monitoring. Interestingly, body weight does not correlate with azathioprine metabolite concentrations; this suggests that clinicians should use metabolite concentrations, rather than body weight, to guide azathioprine efficacy and to avoid toxicity22; this may be especially relevant to patients with extremes of body weight.

Blood and metabolite monitoring

Regular monitoring of bloods is recommended when starting azathioprine in order to identify the common adverse effects of bone marrow suppression (leucopenia and thrombocytopenia)—these may occur even in patients with normal TPMT levels—and hepatotoxicity including the less common, but not rare, adverse effect of cholestatic hepatitis (box 2). The British Society for Rheumatology guideline for prescribing and monitoring of non-biological disease-modifying antirheumatic drugs recommends fortnightly blood monitoring (full blood count, creatinine/calculated glomerular filtration rate, alanine transaminase and/or aspartate transaminase and albumin) on starting azathioprine or on changing dose, until on a stable dose for 6 weeks; patients on a stable dose need monthly blood monitoring for 3 months and thereafter at least every 12 weeks.23 More frequent monitoring (eg, initially weekly) is appropriate for patients at higher risk of toxicity. Once established on azathioprine, it is important to set up shared-care protocols to ensure its safe prescribing in the community.

Patients taking azathioprine may develop dose-related increases in mean corpuscular volume (MCV) owing to megaloblastic bone marrow changes. The rise in MCV appears to correlate with azathioprine efficacy.24 However, this change and also a fall in lymphocyte count, develop over many months; they are best regarded as indicators of drug adherence rather than as tools to monitor therapeutic effect (see box 3). A lymphopenia of <0.5×109/L suggests the need to reduce the dose; treatment should be withheld if the total white cell count falls below 1.5×109/L.

Box 3

The practical application of azathioprine metabolite testing

A 30-year-old woman with acetylcholine receptor antibody-positive and thymoma-negative generalised myasthenia gravis did not wish to take corticosteroids as she was worried about side effects. We discussed her case with Professor Hilton-Jones (Oxford, UK) who recommended starting azathioprine (she had a normal serum thiopurine methyltransferase concentration) and up-titrating pyridostigmine. Despite increasing the azathioprine to 3 mg/kg/day (150 mg daily) she still had symptoms and signs of mild fatigue over a year after starting azathioprine. Her liver function remained normal (figure 3). She requested a further increase in azathioprine dose.

We measured her azathioprine metabolite concentrations: her 6-thioguanine concentration was 424 pmol/8×108 RBC (235–450) and 6-methyl mercaptopurine concentration was 17 034 pmol/8×108 RBC (0–5700). Because of this we advised her that it would be unsafe to increase the azathioprine further. All her symptoms eventually resolved but she developed macrocytosis and leucopenia 3 years after starting azathioprine (figure 3).

Although her 6-thioguanine concentration was therapeutic, the very high serum 6-methylmercaptopurine concentration and its ratio to 6-thioguanine (at 40) suggested hypermethylation, which might explain the longer than expected latency to therapeutic effect. She would potentially be at risk of hepatotoxicity which might be prevented by split-dosing or reducing the azathioprine dose and adding allopurinol (see text) .

Figure 3

Changes in lymphocyte count (normal range (NR): 1.1–4.0×109/L), mean corpuscular volume (MCV) (NR: 78–97 fl) and liver function (bilirubin normal range (NR) 0–12 µmol/L), albumin (ALB, NR: 33–55 g/L), alanine transaminase (ALT, NR: 0–52 U/L), alkaline phosphatase (ALP, NR: 30–160 U/L) and gamma-glutamyl transferase (GGT, NR: 0–64 U/L)) over time on azathioprine.

While the role of azathioprine metabolite testing remains understudied in neurological disease and the impact on outcome uncertain, some guidelines suggest it is clinically useful.20 It is still unclear if all neurology patients taking azathioprine should undergo metabolite testing or only those at higher risk, for example, patients at extremes of body weight, the elderly, those with hepatic or renal impairment, or those with intermediate levels of TPMT.

Extensive use of azathioprine in inflammatory bowel disease has allowed protocolisation of azathioprine metabolite testing (serum metabolite concentrations and the ratio of 6-methylmercaptopurine:6-thioguanine). These are measured in patients with active inflammatory bowel disease 4 weeks after starting azathioprine, or after a change in dose, then 12–16 weeks after starting, and then annually.25 In our opinion, it is sensible to apply this protocol to higher-risk neurology patients. It may be reasonable to consider metabolite testing for most neurology patients taking azathioprine, and certainly for those who respond poorly to the drug. This simple blood test may help identify patients at risk of toxicity (necessitating a reduction of azathioprine dose and slower titration), patients who have therapeutic serum concentrations of 6-thioguanine but still with active disease (raising consideration of changing to another immunosuppressive agent), patients with subtherapeutic 6-thioguanine serum concentrations (which usually responds to an increased azathioprine dose), patients who are poorly adherent to treatment (who have very low 6-thioguanine and 6-methylmercaptopurine serum concentrations), and patients who are hypermethylating (box 3 and table 1). Patients who are hypermethylating risk hepatotoxicity. One option is dose-splitting, equating reducing the 6-mercaptopurine dose below the optimum substrate affinity for TPMT, and potentially reducing the 6-methylmercaptopurine serum concentration without affecting that of 6-thioguanine; the other option is reduce the azathioprine dose (to 25%–50% of the standard dose) and add allopurinol 100 mg. Allopurinol is a xanthine oxidase inhibitor that prevents the breakdown of 6-mercaptopurine into thiouric acid (see figure 1). The resulting increased bioavailability of 6-mercaptopurine corrects hypermethylation by shunting back in favour of 6-thioguanine production.

Pregnancy and breast feeding

The British National Formulary and other drug information sites suggest that azathioprine can be given in pregnancy.26 27 Azathioprine is considered lower risk in pregnancy when compared with other immunosuppressive agents. The recorded modest increase of low birth weight or atrial or ventricular septal defects is mostly derived from the inflammatory bowel disease literature and it is uncertain to what extent the risk to the fetus is driven by the azathioprine or the underlying disease. Moreover, we do not know whether this risk translates to the neurology population. Nevertheless, the use of azathioprine during pregnancy necessitates an informed discussion with the patient, weighing the risks against the benefits, and it is essential to involve maternal medicine services. Azathioprine should not be started de novo in pregnant women; its long latency to therapeutic effect would make this an illogical choice for immunosuppression in this setting. Overall, azathioprine concentration in breast milk is very low and mothers taking azathioprine may breast feed; any detectable concentration of azathioprine in breast milk drops significantly 4–6 hours after a dose.

Drug interactions

We do not propose to discuss azathioprine drug interactions in detail. Severe leucopenia may occur when azathioprine is given together with allopurinol, owing to inhibition of xanthine oxidase by allopurinol prolonging azathioprine’s action (see figure 1). Patients taking allopurinol require a reduced dose of azathioprine to 25%–50% of the standard dose. Coadministration of azathioprine with angiotensin-converting enzyme inhibitors may also induce severe leucopenia. Note that azathioprine also reduces the anticoagulant effect of warfarin.

Risk of infection

Patients either taking azathioprine alone or combined with other immunosuppressants, particularly corticosteroids, have increased susceptibility to viral, fungal and bacterial infections.27 The actual risk is difficult to quantify but may be higher when azathioprine is combined with corticosteroids.

Various infection screens are recommended before starting azathioprine (table 1). Varicella zoster virus infection (chickenpox and shingles) may manifest, sometimes severely, when taking azathioprine and other immunosuppressants. Before starting azathioprine, patients should be asked about their history of varicella zoster and considered for serological testing to determine previous exposure. Those with no history of exposure or serological evidence of exposure to varicella zoster should be advised to avoid contact with people who have chickenpox or shingles. Patients inadvertently exposed to varicella zoster should seek specialist advice (and be considered for passive immunisation with varicella zoster immunoglobulin); those with confirmed infection may need antiviral treatment and supportive care, together with specialist input.

The immunosuppressive activity of azathioprine could, in theory, result in an atypical and potentially dangerous response to live vaccines; thus, people taking azathioprine should not receive live vaccines. Furthermore, they may also be at risk of a less severe response to killed vaccines. Nevertheless, patients are recommended to receive the pneumococcal vaccine before starting azathioprine, as well as an annual influenza vaccination.

Risk of malignancy

Although azathioprine use has been associated with the risk of developing non-Hodgkin’s lymphomas and other malignancies (notably skin cancers), it is not clear to what extent this risk is driven by the underlying disease for which the azathioprine is being given. It may be more relevant in those people needing more intensive immunosuppression, such as transplant recipients, where other immunosuppressive agents may contribute to the risk. Overall, the long-term risk of azathioprine-related malignancy in the neurology population is likely to be small.27


Neurologists are already fairly familiar with using azathioprine. Nevertheless, it is not always easy to remember the various factors impinging on the safe and effective administration of the drug in neurology patients. Neurologists need to be aware of the potential problems with its use and how best to tailor its administration, although most of the relevant guidance is extrapolated from the field of gastroenterology. We hope that adherence to these suggestions will ensure that azathioprine remains a safe and effective drug for neurology patients.

Key points

  • Serum thiopurine methyltransferase (TPMT) concentration measurement is essential before starting azathioprine, even though high serum concentrations account for only a small proportion of overall azathioprine toxicity.

  • Fortnightly white cell counts and liver function tests are essential on starting azathioprine to identify myelosuppression and hepatotoxicity; these can develop despite a normal serum TPMT concentration.

  • Azathioprine metabolite measurements can help to tailor the azathioprine dose, especially in higher-risk neurology patients or those with a poor clinical response.

  • Infection screening before starting azathioprine is an important safety measure, but easily overlooked.

  • Shared-care agreements with primary care help to ensure safe and effective use of azathioprine in the community.

  • Azathioprine is safe in pregnancy and breast feeding, and probably has a low long-term malignancy risk.



  • Contributors All authors contributed to the final manuscript and its subsequent revision. UK conceived the idea around the manuscript and both authors contributed to the generation of the manuscript and researching of the ideas contained therein.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Commissioned; externally peer reviewed by Jackie Palace, Oxford, UK.