SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

Background:  Thioguanine nucleotides (TGNs) are the active product of thiopurine metabolism. Levels have been correlated with effective clinical response. Nonetheless, the value of TGN monitoring in clinical practice is debated. We report the influence of introducing TGN monitoring into a large adult inflammatory bowel disease (IBD) clinic.

Patients and methods:  Patients with IBD undergoing TGN monitoring were identified from Purine Research Laboratory records. Whole blood TGNs and methylated mercaptopurine nucleotides were hydrolysed to the base and measured using HPLC. Clinical and laboratory data were obtained retrospectively.

Results:  One hundred and eighty-nine patients with 608 available TGN results were identified. In non-responders, TGNs directed treatment change in 39/53 patients. When treatment was changed as directed by TGN, 18/20 (90%) improved vs. 7/21 (33%) where the treatment decision was not TGN-directed, p < 0.001. Where treatment change was directed at optimisation of thiopurine therapy, 14/20 achieved steroid-free remission at 6 months vs. 3/10 where the TGN was ignored, (p = 0.037).

Six per cent of patients were non-adherent, 25% under-dosed and 29% over-dosed by TGN. Twelve per cent of patients predominantly methylated thiopurines, this group had low TGN levels and high risk of hepatotoxicity. In responders, adherence and dosing issues were identified and TGN-guided dose-reduction was possible without precipitating relapse.

Mean cell volume (MCV), white blood cell count (WBC) and lymphocyte counts were not adequate surrogate markers. MCV/WBC ratio correlated with clinical response, but was less useful than TGN for guiding clinical decisions.

Conclusions:  Monitoring TGNs enables thiopurine therapy to be optimised and individualised, guiding effective treatment decisions and improving clinical outcomes.


What’s known

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References
  •  Thioguanine nucleotides (TGNs) are the active product of thiopurine metabolism.
  •  Levels have been correlated with effective clinical response in inflammatory bowel disease, but this relationship is not perfect.
  •  As a result, the value of TGN monitoring in clinical practice is debated and many centres still do not use them.

What’s new

We demonstrate that using TGNs in clinical practice helps to:

  •  Guide more effective decision making.
  •  Individualise treatment.
  •  Improve clinical outcomes.

Introduction

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

Azathioprine (AZA) and mercaptopurine (MP) are the first-line immunomodulators for inflammatory bowel disease (IBD) and have proven efficacy in induction of remission in Crohn’s disease (CD) and maintenance of disease remission in both ulcerative colitis (UC) and CD. In addition, thiopurine drugs have a role in fistula healing, as steroid-sparing agents (1–4) and as concomitant immunosuppression to prevent loss of response to biologic therapy (5). Indeed, as many as 60% of those diagnosed with CD receive AZA at some point in their disease course (6) and this figure is likely to increase as more emphasis is put on mucosal healing and earlier intervention (7–9).

Lack of response and toxicity are significant issues with thiopurine therapy. Concern over toxicity leads to cautious dosing strategies (10), reducing the chance of complete remission and prolonging the time taken to achieve remission, resulting in prolonged symptoms, impaired quality of life and disease complications. Furthermore, patients risk being switched to other medications unnecessarily (corticosteroids, alternative immunomodulators and/or biologics) or the requirement for surgery. Measures, which permit individualisation of thiopurine dosing to avoid these outcomes are therefore clearly desirable.

AZA is a prodrug, which enters the endogenous purine salvage pathway to be converted to its active end-product, thioguanine nucleotides (TGNs), see Figure 1, which are incorporated into DNA (11–14), trigger apoptosis of dividing cells by binding Rac-1 receptors (15) and inhibit T-cell activation signals (16). Other metabolic pathways compete with this anabolic process (see Figure 1), most importantly the production of methylated metabolites (MeMP), which can be measured alongside TGNs. Some patients over-produce MeMP resulting in low TGN levels/non-response (17–19) and hepatotoxicity (20,21) from an accumulation of MeMP. Detection of predominant methylation is important as it can be easily circumvented by prescription of low-dose AZA alongside allopurinol (22–24), retaining thiopurines as a suitable treatment option for these patients. There is significant inter-individual variation in these pathways, meaning that even weight-adjusted dosing leaves a significant proportion of patients under or over-dosed.

image

Figure 1. Drugs: AZA, azathioprine; MP, mercaptopurine. Metabolites: tIMP, thioinosine monophosphate; tXMP, thioxanthine monophosphate; TGNs, thioguanine nucleotides; tITP, thioinosine triphosphate; 6-MeMP, 6-methylmercaptopurine; 6-Me-tIMP, 6-methyl thioinosine monophosphate; 8-OH 6-MP, 8-hydroxy mercaptopurine; 6TU, 6-thiouric acid. Enzymes: TPMT, thiopurine methyltransferase; XDH, xanthine oxidase/dehydrogenase; AO, aldehyde oxidase; ITPase, inosine triphosphatase; IMPDH, inosine monophosphate dehydrogenase; HGPRT, hypoxanthine-guanine phosphoribosyltransferase; GMPS, guanosine monophosphate synthetase; GST, glutathione s-transferase

Download figure to PowerPoint

As a result of these problems, TGN level monitoring has been proposed as a way of optimising and individualising thiopurine treatment. A consensus is emerging that TGN levels do correlate with response to thiopurine treatment (25), inversely correlate with Harvey–Bradshaw Index (26) and are lower in those with active disease (27). In addition, TGNs have been shown to be a useful tool to detect non-adherence to treatment, suboptimal dosing and biochemical resistance (predominant methylation) in patients who are non-responders to thiopurine treatment (27–32). TGN levels may also predict dose-dependent toxicity and detect non-adherence in those who are in remission (29,33). A recent article addressing the utility of measuring TGNs in paediatric practice found that this altered management in 36% of cases (34).

Surrogate markers of therapeutic TGN concentrations have been proposed, including mean cell volume (MCV), change in MCV, white blood cell (WBC) and lymphocyte counts. However, whereas these markers correlate with TGN concentrations (1,35–39), there is considerable inter-individual variability, limiting their value in clinical practice (34,39–42). Likewise, deliberately pushing up doses of AZA until white cell counts reduce has been shown to cause toxicity (34,43–45). Interestingly, Waljee et al. reported that a computer algorithm incorporating basic blood biochemistry, full blood count data and patient age was strongly associated with response to thiopurines (46), but whether this could be used to guide clinical decision-making remains unclear (47). Likewise, a recent Spanish study has cast doubt on the ability of TGN monitoring to accurately predict remission, but does not address how the results would affect clinical management (48).

Although there is no reference to TGN measurement in the British Society of Gastroenterology guidelines on management of IBD (49), the American Gastroenterological Association supports their use for all the above indications (50) and British paediatric (51) and ECCO guidelines (52) recommend their use when patients relapse on thiopurine treatment. A perception nevertheless persists, that TGN measurements are only useful for detecting non-adherence and, alongside a lack of data demonstrating their usefulness in clinical practice, this has led many centres to continue to manage their thiopurine patients without TGN monitoring. The aim of this study was therefore to determine the impact of introducing clinical TGN monitoring into a large UK IBD service studying, in particular, the influence on clinical management decisions made as a consequence of TGN measurement.

Methods

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

Patients

Patients attending the specialist IBD clinic at Guy’s and St Thomas’ NHS Foundation Trust, who had TGNs measured to monitor either AZA or MP treatment, were identified from the Purine Research Laboratory (PRL) database. In each case, clinical records and laboratory results were reviewed retrospectively to record data on demographics, type of IBD, indication for treatment, thiopurine dose, toxicity and outcome as determined by the treating physician. Clinical response was defined in two ways. First, by a global assessment of outcome by the treating physician and secondly, by 6-month steroid-free remission (patients being off all steroid and in clinical remission by 6 months with no recourse to alternative treatment escalation).

Thiopurine methyl transferase (TPMT) activity was checked in all cases prior to treatment. Patients receiving tioguanine or thiopurine alongside allopurinol at the index TGN result were excluded. TGN was measured not earlier than 4 weeks after starting treatment. In a proportion of patients, multiple TGN measurements had been taken.

Changes in management based on the TGN level were noted and the proportion of TGN measurements, which prompted a change established. Treatment outcomes were compared between the patients in whom treatment decisions were made as directed by the TGN results and those for whom treatment decisions were different from that dictated by the TGN result. As some guidelines restrict the use of TGNs to non-responders, a separate analysis of the impact of the TGN result on the treatment of non-responders was made. TGN values were also compared with full blood count indices to assess value added by TGN measurement vs. traditional monitoring.

Laboratory methods

TGN and methylated thiopurine metabolite (MeMP) concentrations were measured by the PRL (GSTS Pathology at St Thomas’ Hospital, London), as the hydrolysed base in whole blood according to the PCA hydrolysis method reported by Dervieux and Boulieu (53). TGN concentrations were categorised according to the PRL therapeutic range, as sub-therapeutic (< 200 pmol/8 × 108 RBC), therapeutic (200–400 pmol/8 × 108 RBC) and high (> 400 pmol/8 × 108 RBC).

Statistical analysis was performed using Instat version 3.0a for Macintosh, (Graphpad Software, San Diego, CA, USA http://www.graphpad.com). For analysis of contingency tables, Chi-squared tests were used, unless cells contained values less than 5 in which instance a Fisher’s exact test was performed. Variance in means was tested using ANOVA where results were normally distributed, or alternatively using Kruskal–Wallis test.

Results

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

One hundred and eighty-nine patients were identified for whom a total of 608 TGN results were available for analysis. The age range was 12–83 years, (median 38 years), 103 were female subjects, 134 had CD, 50 UC and 5 IBD-unclassified. At the time of first TGN measurement, 15 patients were on concomitant biologic therapy with either infliximab or adalimumab.

An overview of the initial TGN results is shown in Figure 2. Of the 189 patients, only 75 (40%) were dosed correctly according to TGN levels, 12 patients (6%) were non-compliant with zero- detectable metabolites, 47 (25%) were under-dosed and 55 (29%) had high TGN levels.

image

Figure 2.  Distribution of patients, according to their initial TGN levels relative to our therapeutic range of 200–400 pmol/8 × 108 RBC. ‘Zeros’ are patients with no detectable TGNs or methylated metabolites

Download figure to PowerPoint

TGN levels predicted response to thiopurine treatment, with response rates varying from 84% in those who were appropriately dosed, to 18% in those who were non-adherent, see Figure 3. The therapeutic range appears appropriate, as the remission rate was equivalent in those with TGNs in the lower quartile (200–250 pmol/8 × 108 RBC) and upper quartile (350–400 pmol/8 × 108 RBC): 9/13 vs. 14/17, p = 0.4.

image

Figure 3.  Rate of non-response as a percentage according to TGN result. TGNs classified as zero (non-adherent), low < 200 pmol/8 × 108 RBC, therapeutic (200–400 pmol/8 × 108 RBC) or high (> 400 pmol/8 × 108 RBC)

Download figure to PowerPoint

Non-adherence

Twelve patients (6%) were non-adherent as defined by undetectable TGN levels at the time of the first TGN measurement. Overall, including subsequent testing, 8% of all TGN results were zero. The majority of non-adherent patients were non-responders (9/11 where clinical response can be determined). The issue of non-adherence was discussed with patients who generally admitted this, facilitating useful conversations about, for example the importance of continuing or the appropriateness of withdrawing therapy, drug safety, alternative treatment approaches and the importance of disease control at conception and during pregnancy.

There was evidence of adherence varying over time, with patients initially demonstrating therapeutic TGN, later having undetectable TGN concentrations. This was a particular issue in patients starting infliximab (4/15) and in those attempting to conceive (4/15).

TGN concentrations below and above the therapeutic range

Forty-seven of 189 patients (25%) were under-dosed by TGN level and 55 (29%) patients had TGN concentrations above the therapeutic range reflecting potential over-dosing, of whom two patients had extremely high TGNs > 1000 pmol/8 × 108 RBC (Table 1).

Table 1. First recorded thioguanine nucleotide concentrations and haematological indices in patients receiving azathioprine or mercaptopurine for the treatment of IBD
TGN concentration pmol/8 × 108 RBCNumber (%)Mean MCV, fl (range)Mean WBC (range)Mean lymphocyte count (range)
  1. MCV, mean cell volume; TGN, thioguanine nucleotides; WBC white blood cell count.

  2. Significant differences: *p = 0.03, ANOVA.

  3. **p = 0.02, Kruskal–Wallis.

Zero12 (6)89* (74–98)6.7 (4.9–12.4)1.6** (0.9–2.6)
Low (< 200)47 (25)93 (69–118)6.5 (2.7–14.9)1.3 (0.4–2.9)
Normal (200–400)75 (40)94 (74–112)6.7 (3.5–12.0)1.3 (0.3–5.8)
High (> 400)55 (29)96* (78–105)6.0 (2.4–14.3)1.1** (0.2–2.1)

In those with low TGNs, the result led to 18 dose increases, nine of which were in patients in complete remission who would therefore not have been otherwise dose-escalated. An additional two patients were switched to combination low dose AZA/allopurinol therapy as a result of accompanying high MeMP levels, indicating preferential methylation.

In those with TGNs above the upper limit of the therapeutic range of 400 pmol/8 × 108 RBC, dose reductions were, in practice, triggered by a TGN > 550 pmol/8 × 108 RBC. None of the 30 patients with TGN in the range 400–550 pmol/8 × 108 had a dose reduction. In contrast, dose reductions were made in 14 of 24 patients with TGN greater than 550 pmol/8 × 108 RBC. Where dose-reduction was an isolated treatment change, 7/7 patients in remission at the time of dose-reduction remained so at review, one patient remained steroid-dependent.

Non-responders

There were 53 patients unresponsive to thiopurine therapy at the time the first TGN concentration was measured (excluding all patients for whom response to thiopurines could not be determined in isolation). In these patients, 39 (74%) documented treatment changes were prompted by the TGN result. Non-adherence was addressed in 12 patients, 3 patients were switched to azathioprine/allopurinol combination therapy and 13 patients had doses increased or decreased on the basis of the TGN result. If a blind policy of dose escalation in non-responders was applied, 11 of these dose-escalations would have happened anyway. However, 11 non-responders with existing high TGNs would have been dose-escalated inappropriately. Seven of the eight non-responders with preferential methylation (TGN/MeMP ratio > 11) avoided possible toxicity associated with dose-escalation, whereas the one patient where dose-escalation was attempted despite preferential methylation developed hepatotoxicity. Eleven non-responders went directly to a change in treatment plan (to methotrexate, biologic or surgery), including six undergoing surgery for stricturing disease. Blind-dose escalation in this group could have delayed effective therapy, whilst thiopurine dosing was altered inappropriately.

In four patients, treatment changes were delayed until a repeat TGN concentration was measured at the next clinic visit. In each of these patients, there was a treatment change on the basis of the second result (two dose changes and two switched to AZA/allopurinol), bringing the total treatment changes in non-responders based on TGN concentrations to 43/53 (81%). A proportion of the non-responders in whom no change was made were lost to follow-up and several had entered remission at the time of their review.

For non-responding patients in whom the response to treatment adjustment could be assessed in isolation: where appropriate action was taken as directed by the TGN results, 18/20 (90%) patients had an improved clinical outcome vs. 7/21 (33%) where the treatment decision was counter to that indicated by the TGN level (p < 0.001, Chi-squared test). For the analysis of steroid-free remission, only those patients where the outcome of the TGN was directed at optimisation of thiopurine therapy (rather than abandoning thiopurines in favour of surgery or biologic) have been included (i.e. non-adherence, under-dosing, hyper-methylation). In this group, where thiopurines were appropriately optimised 14/20 were in steroid-free remission at 6 months, and a further three patients were continuing to slowly wean off long-term steroids, this compares to 3/10 remissions in those patients where the TGN was ignored, (p = 0.037 or 0.003 if slow steroid-weaners included).

Methylated metabolites

At the first TGN check, 21/177 (12%) patients taking their medication demonstrated preferential thiopurine methylation with a ratio of MeMP/TGN > 11. If an absolute cut off of 5700 pmol/8 × 108RBC is used instead, five patients were detected, four of these also had a ratio > 11. Predominant methylation, by ratio, was associated with subtherapeutic TGN concentrations, p < 0.001 (Chi-squared test), but not with response to treatment (p = 0.47). In predominant methylators who were responders, no action was taken in 6/11 cases. However, one patient had their treatment stopped, two had dose adjustments and two switched to azathioprine and allopurinol combination treatment. Amongst non-responders, 2/8 were switched to allopurinol/azathioprine combination treatment, (one after an inappropriate dose-increase had induced abnormal liver function tests), three patients went to surgery and three continued on thiopurine monotherapy. Two patients could not have a clinical response to thiopurine determined in isolation from other treatment changes. Abnormal liver function tests were documented in 5/21 (24%) of patients demonstrating predominant methylation by ratio, (and 1/5 (20%) with MeMP > 5700 pmol/8 × 108 RBC).

Predicting TGN and clinical outcome from other results

Table 1 shows the results of the first available TGN result for each patient alongside the relevant full blood count parameters (and Table 2 shows the same data for all their TGN results). Table 3 and Figure 3 show the rates of non-response to therapy at the first TGN check in the four TGN subgroups. Response data excludes those receiving concurrent infliximab/adalimumab or those where levels had been taken when clinical response data was still awaited (i.e. at less than 4 months of treatment).

Table 2. Summary of all TGN results and full blood count indices
TGN concentration pmol/8 × 108 RBCNumber (%)Mean MCV, fl (range)Mean WBC (range)Mean lymphocyte count (range)
  1. MCV, mean cell volume; TGN, thioguanine nucleotides; WBC white blood cell count.

Zero46 (8)88 (67–117)9.0 (3.6–14.3)1.5 (0.5–2.6)
Low (< 200)136 (22)91 (69–118)6.7 (2.7–14.9)1.3 (0.2–5)
Normal (200–400)260 (43)94 (69–113)6.5 (1.5–17.4)1.2 (0.2–5.8)
High (> 400)166 (27)97 (78–119)6.1 (1.7–19.7)1.1 (0.2–2.4)
Table 3. The relationship between first TGN concentration taken for each patient and clinical response at that time to azathioprine/mercaptopurine
TGN concentration pmol/8 × 108 RBCNon-responders (%)
  1. TGN, thioguanine nucleotides.

Zero9/11 (82)
Low (< 200)22/45 (49)
Normal (200–400)11/67 (16)
High (> 400)11/46 (24)

A comparison of MCV results between those with zero, low, normal and high initial TGN results (Table 1) revealed one significant difference, (p = 0.03, ANOVA), specifically, those with TGNs above the therapeutic range had a greater MCV than those who were non-adherent (zero TGN). There was no significant difference in MCV between those with TGN in the target range and any other group. Likewise, lymphocyte counts were only significantly different when those with TGNs exceeding the target range were compared with non-adherent individuals, (p = 0.02, Kruskal–Wallis test). There was no difference in total white cell counts across all groups (p = 0.4, ANOVA). The haematological indices were remarkably similar across all TGN groups (Table 2).

Waljee et al. suggest that the ratio MCV/WBC, using a cut-off value of 12, predicts response as accurately as TGN. In our group, there was a separation in response rate with this cut off, 22/40 with a ratio less than 12 being non-responders and 33/124 with a ratio above p = 0.001 (Fisher-exact test).

Using WBC < 4 × 109 cells/l to predict therapeutic or high TGN levels gives a specificity of 91.4%, but a sensitivity of only 8.7%. Using an MCV cut off of 95 fl to predict therapeutic or high TGN levels gives a sensitivity of 53% and specificity of 58%. The ratio of MCV/WBC predicted therapeutic or high TGN with a sensitivity and specificity of 75% and 25% using the published cut-off value of 12. (It performed slightly better under our analysis using a cutoff of 20, which gave a sensitivity and specificity of 72% and 79% respectively). Full blood count indices were not therefore considered an adequate surrogate marker of TGN level.

Discussion

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

Our results show that TGNs are useful in clinical practice, identifying non-adherence, preferential methylation, under- and over-dosing and facilitating timely treatment change in those non-responsive to thiopurines despite therapeutic TGN concentration.

In non-responders to single agent thiopurine therapy, the TGN concentrations provided a clear benefit, demonstrating the reason for an individual’s non-response and consequently guiding treatment decisions. If TGNs had not been measured in this cohort and non-responders had been treated by blind dose-increments, this would have been likely to benefit 14/53 patients where TGN results demonstrated under-dosing. However, in 22/53 individuals with therapeutic or high TGN concentrations and 8/53 with hypermethylation, it would have been potentially harmful. In the nine non-adherent individuals, dose-increments would have been unlikely to have had an effect.

In our experience, the TGNs also provided useful information in the management of patients classified as responders, detecting non-adherence, under- and over-dosing, including cases with extremely high TGN concentrations, where we anticipate toxicity was avoided.

An interesting issue is what to do with patients given a historical label of non-response in whom treatment with thiopurines was abandoned. Our results would suggest that a significant number of these would have been non-adherent and a further proportion under-dosed or predominantly methylating. If such patients are running out of therapeutic options, thiopurine agents should be revisited.

In our study, methylated metabolites were useful in identifying a group of patients that are biochemically resistant to thiopurines. This group suffered a high incidence of abnormal liver function tests and had lower TGN levels. In several instances, hyper-methylation appears to have gone unnoticed by the treating clinician, and there may have been opportunities to optimise thiopurine treatment in this group that were missed. This has been one of the drivers behind the development of a virtual clinic in our centre (which reviews, amongst other aspects, all TGN and MeMP results to ensure they are appropriately acted upon), and the production of a handbook for everyone working in the IBD clinic, which gives practical advice on interpretation of the TGN and MeMP results. An overview of the guidance on interpretation of TGN/MeMP results is given in Table 4.

Table 4. Use of TGN and MeMP levels in clinical practice
TGN level pmol/8 × 108 RBCInterpretationAction required
  1. MeMP, methylated metabolites; TPMT, thiopurine methyl transferase; TGN, thioguanine nucleotides.

ZeroNon-adherenceDiscussion with patient
Low (< 200)MeMP:TGN < 11 under-dosedDose increment & recheck TGN in 4 weeks
MeMP:TGN ≥ 11 hypermethylationSwitch to low dose AZA (25% standard TPMT-guided target dose) & Allopurinol 100 mg
Normal (200–400)Adequate doseIn responders, no action required. For non-responders – recheck TGN, seek alternative explanation for symptoms, consider switch to alternative therapy, e.g. methotrexate
High (> 400)Over-doseIn responders, consider dose-reduction, particularly if TGN > 550. In non-responders, recheck TGN, seek alternative explanation for symptoms, consider switch to alternative therapy, e.g. methotrexate

Although several studies have demonstrated successful dose-titration based on TGN monitoring (27,28,33,54), only one study has been conducted prospectively comparing outcome in patients whose AZA treatment was adjusted using TGN monitoring and those treated in the traditional manner (55). This failed to show an increased remission rate in the group with TGN monitoring. However, this study has several shortcomings, in particular, the short-study period (outcome measured at 16 weeks), the extensive co-prescription of steroid (in all patients until 12 weeks with 56% still on them when the primary outcome was measured at 16 weeks) and the small likelihood that a statistically significant difference in response rate could be identified between a control group (that received on average 2.7 mg/kg) and a treatment group where dose was not allowed to exceed 3 mg/kg. This study did show that TGN monitoring could predict toxicity and was able to demonstrate a biochemically resistant group in whom dose-escalation did not increase TGN concentrations (23,55,56). A recent report of the use of TGNs in paediatric practice supports the fact that they aid clinical decision-making and improved patient outcomes (34), whereas another recent study in adults with non-response to traditionally dosed thiopurines confirms the utility of TGNs in practice. This study had very similar results to our own, and demonstrated that treatment decisions guided by TGN/MeMP profile produced a markedly superior outcome to those which were discordant with the TGN, although outcome analysis was restricted to a physician’s global assessment of ‘clinical improvement’ (57).

In this study, we have used 200–400 pmol/8 × 108 RBC as the therapeutic range, which is the range in clinical use at our centre and the PRL. The upper limit of the range is consistent with that used in other trials (55) although it lacks an evidence base, whereas the lower limit of the range derives from the literature and our previous experience of TGN monitoring (58). The figure of 200 is lower than that used in some other studies, but analysis of our results demonstrated that there was no significant difference in remission rate between those with a TGN between 200 and 250 pmol/8 × 108 RBC and those between 350 and 400 pmol/8 × 108 RBC. Although the therapeutic range used in this study is clearly clinically useful, how to respond to moderately high or low TGN concentrations in a patient with a good response to treatment remains unknown. It would be reasonable to suppose that those with therapeutic TGN concentrations are more likely to maintain their response, but there is little evidence to support dose-adjustment in this situation. There is some evidence to suggest that long-term complications of thiopurines, in particular malignancy, may relate to higher TGN concentrations (59), but the upper limit of the therapeutic range has not been the subject of much investigation. Our experience suggests that if TGNs are greater than 550 pmol/8 × 108 RBC, dose-reduction can be undertaken without compromising clinical benefit or reducing TGN levels below the therapeutic range. The effect of dose-reduction with TGN between 400 and 550 pmol/8 × 108 RBC could not be assessed as no dose-reduction was attempted on patients with TGNs within this range.

There was a wide inter-individual variation of MCV, WBC and lymphocyte count in all TGN ranges and some individuals with completely normal full blood counts despite therapeutic or even high TGN levels. There was no statistically significant difference in blood count indices between those with therapeutic TGN concentration and any other group. In addition, TGN concentrations reach a steady state 4 weeks (58,60) after the drug is initiated, which allows dosing to be adjusted and treatment optimised without having to wait 3–4 months to assess clinical response. This is not true of surrogate markers of response, such as the full blood count indices, which evolve over the same time-frame as clinical response.

TGN measurements have a nominal cost of approximately £30 in the UK. If checked only in non-responders, each TGN result should direct a change in therapy. This cost compares favourably with the measurement of TPMT activity, which costs £300 per change in treatment (halving target dose in the 1 : 10 patients that will be detected to be TPMT heterozygotes). In our real-world analysis, 39/53 non-responders had a change in their treatment made on the basis of the TGN result. This would still only give a cost of £41 per treatment change, rising to £56 if those who would have benefited from blind dose-increments are also excluded.

Not checking TGN measurements also has potential costs. Costs to the patient are considerable, in terms of the possible unnecessary loss of one of their established treatment options, the possibility of avoidable toxicity, the consequences to their health of relapse or progressive disease and the impact of treatment escalation to biologic or surgery. There are also significant potential costs to the health service provider, including the financial implications of a patient requiring biologics or surgery, avoidance of serious toxicity and the costs of monitoring, clinic visits and drug prescriptions for non-adherent individuals.

The main limitation of the study is the retrospective collection of clinical outcome data. However, this was not our primary endpoint and we have tried to make this assessment as robust as possible by reporting a 6-month steroid-free remission rate alongside the treating physician’s assessment.

In summary, therefore we have successfully introduced TGN monitoring into clinical practice where it has proven useful in guiding clinical decision-making and appears to improve clinical outcomes. Checking TGN concentrations in non-responders is cost-effective and in each case should prompt a change in treatment. Testing TGNs in all patients on thiopurines will confirm adherence, identify under- and over-dosing and predominant methylation. Responders with TGN concentrations > 550 pmol/8 × 108 RBC can safely have their dose of thiopurine reduced. We would advocate measuring TGN and MeMP concentrations in all thiopurine-treated patients.

Acknowledgements

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References

We acknowledge funding support from Guy’s and St. Thomas’ Charity. We have no additional acknowledgements.

References

  1. Top of page
  2. Summary
  3. What’s known
  4. Introduction
  5. Methods
  6. Results
  7. Discussion
  8. Acknowledgements
  9. References