This document updates UK Haemophilia Centre Doctors Organization (UKHCDO) guidelines on the management of factor VIII/IX (FVIII/IX) inhibitors in congenital haemophilia (Hay et al, 2000, 2006). Acquired haemophilia is excluded and will be covered separately. Most data apply to FVIII inhibitors and the recommendations for FIX inhibitors are sometimes extrapolated from this. Low titre inhibitors are defined as <5 Bethesda units (BU)/ml and high titre ≥5 BU/ml. These guidelines are targeted towards haemophilia treaters in the UK. Not all recommendations may be appropriate for other countries with different health care arrangements and resources. The writing group reviewed publications known to them supplemented with papers identified through Pubmed, using index terms h(a)emophilia, factor VIII and IX, inhibitors, alloantibodies, rFVIIa, NovoSeven, FEIBA, aPCC, rituximab, management. The writing group produced the draft guideline which was reviewed and revised by members of the UKHCDO Advisory Board. The guideline was finally reviewed by a sounding board of approximately 50 UK haematologists, the British Committee for Standards in Haematology (BCSH) and the British Society for Haematology Committees and comments incorporated where appropriate. The ‘GRADE’ system was used to quote levels and grades of evidence, details of which can be found at http://www.bcshguidelines.com/BCSH_PROCESS/EVIDENCE_LEVELS_AND_GRADES_OF_RECOMMENDATION/43_GRADE.html. The objective of this guideline is to provide healthcare professionals with pragmatic guidance on the management of patients with FVIII/FIX inhibitors although individual patient circumstances may dictate an alternative approach. Patients with FVIII/IX inhibitors must be registered with, and have their treatment co-ordinated by a Comprehensive Care Haemophilia Centre (CCC) experienced in the management of inhibitors (National Service Specification available at www.ukhcdo.org). Centres must provide 24-h access to senior clinicians with experience in inhibitor management and laboratory services for the measurement of factor levels and inhibitor titres. Patients should be offered inclusion in appropriate clinical trials and reported to registries. UK patients must be registered with the National Haemophilia Database and details of their inhibitor reported as soon as they are confirmed. Recognized and potential risk factors for inhibitor formation are listed in Table 1. Clinicians are advised to check all mutations in mild and moderate haemophilia A on the Haemophilia A Mutation, Structure, Test and Resource Site (HAMSTeRs) database (www.hadb.org.uk) to establish whether any association with inhibitor formation has been reported. The International Study on aetiology of inhibitors in patients with moderate or mild forms of haemophilia A, influences of Immuno-Genetic and Haemophilia Treatment factors (INSIGHT) is expected to report imminently (Eckhardt et al, 2012). This study will provide the largest, unselected cohort of patients with mild/moderate haemophilia A, including exposure data, from which clinicians can gauge prevalence of inhibitors for a given mutation to inform future treatment decisions. Recognition that treatment-related factors can influence inhibitor development offers the possibility that treatment might be modified in order to reduce this risk. Desmopressin (DDAVP) should always be considered when appropriate in mild/moderate haemophilia A (Eckhardt et al, 2009, 2011). At the present time in severe haemophilia A, only prophylaxis has been shown to possibly reduce the risk of inhibitor development and whether this will be applicable to all patients remains to be defined (Gouw et al, 2007a; Auerswald et al, 2012). Severe haemophilia A: Highest risk: null mutations, large deletions and mutations inducing stop codons Medium risk: Intron 1 and 22 inversion, splice site Lower risk: small deletion/insertion and missense mutation Mild/moderate haemophilia A: Usually low risk but specific mutations associated with increased risk Severe haemophilia B: Increased risk with major deletions and null mutations Inhibitors are rare with point mutations Oldenburg and Pavlova (2006), Gouw et al (2011), Boekhorst et al (2008), HAMSTeRs database (www.hadb.org.uk) Gouw et al (2012) Severe haemophilia A: 2- to 5-fold increased risk associated in patients of Hispanic and African origin compared with Caucasians Viel et al (2009), Maclean et al (2011), Astermark et al (2001) Severe haemophilia A: Increased risk with first degree family history Incidence with family history 48% (95% confidence interval: 35–62), incidence without family history 15% (11–21) Severe haemophilia A: Age at first exposure does not appear to have an effect Risk is highest below the age of 5 years and increases after age of 60 years Mild haemophilia A: Risk increases with age Gouw et al (2007a), Chalmers et al (2007), Santagostino et al (2005) Hay et al (2011) Mauser-Bunschoten et al (2012) Severe haemophilia A: Lower risk if human immunodeficiency virus positive Severe haemophilia A: Risk highest during early exposures with a median time of inhibitor presentation at about 10–15 EDs Risk lower after 150 EDs but may occur throughout life. Severe haemophilia B: Risk highest during early exposures with a median time of inhibitor presentation at about 11 EDs. Inhibitors are reported up to 180 EDs. Late inhibitors have not been reported Bray et al (1994), Lusher et al (1993), Lusher et al (2003), Rothschild et al (1998), McMillan et al (1988), Shapiro et al (2005) Hay et al (2011) Severe haemophilia A: Risk increased with 5 or more EDs at first treatment Haemophilia B: No data Mild haemophilia A: Risked increased with intense exposure Gouw et al (2007a), Gouw et al (2007b), Chalmers et al (2007), Maclean et al (2011) Mauser-Bunschoten et al (2012) Severe haemophilia A: Early prophylaxis is associated with a decreased risk in some retrospective studies Severe haemophilia A: No evidence of any difference in inhibitor risk between recombinant and plasma-derived concentrates There is no convincing evidence for a difference in inhibitor risk between different types of rFVIII Franchini and Lippi (2010), Iorio et al (2010b), Aledort et al (2011a), Aledort et al (2011b), Iorio et al (2011), Severe haemophilia A: Risk increased if surgery combined with an intensive first exposure (>4 ED) compared to first exposure without surgery OR 4 (95% confidence interval, 2–8·4) Mild moderate haemophilia A: Risk increased by intensive exposure at the time of surgery especially associated with high-risk mutation Eckhardt et al (2009), Eckhardt et al (2011), Mauser-Bunschoten et al (2012) FVIII/FIX mutation analysis should be undertaken in all patients with haemophilia A and B, especially newly diagnosed patients (Grade 2C). Previously untreated and minimally treated patients with severe haemophilia A who have received an intensive FVIII exposure [≥5 exposure days (EDs)] should be closely monitored for inhibitor formation (Grade 1B). Some consideration may be given to starting early prophylaxis (Grade 2C). All patients who require replacement therapy with concentrate, including previously untreated patients, should be treated with recombinant FVIII/IX (Grade 1C). Inhibitor testing should be performed if a patient has a poor clinical response to concentrate or lower FVIII/IX levels than expected after concentrate infusion. Early detection of an inhibitor is crucial to minimize anamnesis and, if the inhibitor does not rise above 10 BU/ml, allow immune toleration induction (ITI) to be started without delay. Early detection will also limit exposure to sub-optimal treatment. Inhibitor testing is required before elective invasive procedures, when the clinical or laboratory response to concentrate is sub-optimal, before and after a switch of concentrate and 2–3 weeks after intensive treatment (≥5 EDs) or surgery in mild or moderately affected patients. An inhibitor screen should be performed in patients with severe haemophilia at least every third ED or every 3 months if concentrate exposure has occurred (whichever is sooner) until 20 EDs have been achieved. Thereafter, inhibitor testing should be undertaken every 3–6 months until 150 EDs. Most boys with severe haemophilia are established on prophylaxis by the 20th ED and the pragmatic approach is to measure trough levels at least every 3–6 months: if FVIII/IX is measurable further testing is not necessary but if <1 iu/dl, inhibitor testing is required. This strategy, however, may miss some low titre inhibitors and should not be regarded as a definition for the presence or absence of an inhibitor. Inhibitors may occur at any age and incidence increases again after the age of 60 years (Hay et al, 2011), therefore testing should continue 1–2 times a year indefinitely. Patients with mild or moderate haemophilia should be tested annually if exposed to concentrate and after any intensive exposure (≥5 EDs) or surgery (Eckhardt et al, 2009, 2011; Mauser-Bunschoten et al 2012). Patients with mild/moderate haemophilia with a mutation judged to have an increased prevalence of inhibitor formation (see section 4) should be considered for testing after every exposure. FIX inhibitors are associated with allergic reactions to FIX, including life-threatening anaphylaxis (Warrier et al, 1997; Warrier & Lusher, 1998; DiMichele, 2007; Recht et al, 2011), especially in those with gene deletions (Thorland et al, 1999). Any reaction should prompt inhibitor testing before further FIX exposure as even low-level FIX inhibitors may cause anaphylaxis. Patients should have their mutation established as soon as the diagnosis is made to identify the minority of patients with major gene deletions, since FIX inhibitor risk is almost completely confined to this group. The first 20 exposures in patients with severe haemophilia B should be given in hospital with access to paediatric resuscitation facilities, although this does not necessarily need to be in a haemophilia centre. An inhibitor screen should be performed in patients with severe haemophilia B at least every third ED or every 3 months if concentrate exposure has occurred (whichever is sooner) until 20 EDs have been achieved, irrespective of the FIX mutation. Thereafter, inhibitor testing should be undertaken every 3–6 months until 150 EDs. FIX inhibitors have not been reported in patients with more than 150 EDs and so further testing is not required unless clinically indicated. In this guideline, the term ‘inhibitor testing’ includes Bethesda assays, inhibitor screens, enzyme-linked immunosorbent assays (ELISAs) and a 48-h trough measurement on standard dose prophylaxis (20–50 iu/kg alternate day). Any positive test must be confirmed on a repeat sample as soon as possible. Inhibitor tests are most sensitive following a washout period where the factor level has returned to baseline for 24 h, because residual infused concentrate may mask or quench a low-titre inhibitor. FVIII inhibitors are time- and temperature-dependent. FIX inhibitors are not time-dependent. Inhibitor testing is commonly performed using a Nijmegen-Bethesda assay although this is relatively insensitive (Verbruggen et al, 1995). Activated partial thromboplastin time (APTT)-based methods are described (Ewing & Kasper, 1982) and, if used, each laboratory will need to standardize the assay to define an abnormal result. A screening method that determines inhibitory activity against the patient's factor concentrate has been described (Keeling et al, 2005). It is a useful screening test and appears to be more sensitive than a Bethesda assay, however, it may be less specific and associated with false-positive results (UK National Haemophilia Database, unpublished data). In patients using standard prophylaxis (20–50 iu/kg alternate days), a measureable FVIII trough level at 48 h can pragmatically be interpreted as a negative inhibitor screen because this is likely to be associated with a half-life >7 h. In patients with mild or moderate haemophilia A the sensitivity of an inhibitor test may be improved by heating the plasma at 58°C for 90 min to inactive residual FVIII (Kitchen et al, 2009 and Miller et al, 2012). FVIII/IX inhibitors should be quantified with a Bethesda assay (Kasper et al, 1975) with Nijmegen modification for FVIII (Verbruggen et al, 1995). When recombinant B-domain-deleted porcine FVIII concentrate becomes available, appropriate quantification of cross-reactive inhibitors will also be required. An ELISA method may be useful if a lupus anticoagulant is present or for inhibitors which increase clearance rather than inhibiting activity (Sahud et al, 2007). However, an ELISA test may also detect non-inhibitory antibodies. FVIII in vivo recovery (IVR) is calculated by subtracting the pre-infusion from the post-infusion level; it should be reported as the increase in iu/dl divided by the infused dose in iu/kg. Inaccuracies in measuring IVR (Björkman et al, 2007) will be exacerbated by the presence of an inhibitor. In haemophilia A, samples taken 1-h post-infusion underestimate the IVR in most patients lacking an inhibitor (Björkman et al, 2010). The optimal sampling time in the presence of a low titre inhibitor is unknown but it should be standardized so that serial results can be compared. By consensus, we recommend that the post-infusion sample be taken at 15 min. IVR is of limited use for monitoring the strength of an inhibitor but is important for guiding replacement therapy when treating bleeding episodes. The most sensitive way to detect and quantify an inhibitor is to measure the clearance of FVIII/IX. An expert consensus has suggested that a FVIII inhibitor be considered to be present in very young children when the elimination half-life was <6 h. However, the only available published study in children aged 1–6 years, who had no detectable or past history of an inhibitor, (n = 54) reported a median (95% confidence interval) FVIII half-life of 9·4 (7·4–13·1) h (Blanchette et al, 2008) and methodological considerations related to the reduced blood sampling schedule suggest that this half-life may be an underestimate (Björkman et al, 2010). When these data were re-analysed using a population pharmacokinetic method the shortest half-life was about 7 h, even in children aged 1 year (Björkman et al, 2012). In addition, the normal half-life for an individual who has an inhibitor is very unlikely to be known because this will not have been measured prior to inhibitor development. In view of these emerging data we suggest that a FVIII inhibitor should be considered to be present if the half-life is <7 h. There are no consensus criteria for recognition of a FIX inhibitor because normal FIX half-life is uncertain; reports vary widely with values for plasma-derived (pd) FIX ranging from 29 to 43 h and for recombinant FIX (rFIX) from 18 to 24 h (Björkman, 2011). The half-life of rFIX in infants and young children is unknown (Shapiro et al, 2005). For pharmacokinetic studies, 50 iu/kg of FVIII or 75 iu/kg of FIX are infused after a 3-d washout period. The International Society on Thrombosis and Haemostasis FVIII/FIX Scientific and Standardization Committee (ISTH SSC) recommendations state that samples should be taken; pre-dose, and at 10–15 min, 30 min, 1, 3, 6, 9, 24, 28, 32 and 48 h post-infusion for FVIII and an additional sample taken 72 h post-infusion for FIX (Lee et al, 2001). This is very difficult to achieve in young children and times of 1, 3, 6–8, 24 and 48 h are suggested for FVIII in these patients, although this results in an apparently shorter half-life compared to full sampling in non-inhibitor patients (Björkman et al, 2010).This may, therefore, suggest the presence of an inhibitor when none exists. The effect of reduced sampling time points is virtually eliminated by the use of population pharmacokinetic models (Björkman et al, 2012) but no population model is available for patients with low titre inhibitors at present. International Society on Thrombosis and Haemostasis FVIII/FIX Scientific and Standardization Committee guidelines recommend that a series of simple linear regression models can reduce calculations involved in half-life estimation but they emphasize that rigorous statistical analysis is required in order to assign the correct regression function (Lee et al, 2001). A number of computer programs can be used to estimate FVIII half-life but none have been validated for the measurement of half-life in the presence of an inhibitor. The calculation of an accurate half-life in the presence of a low titre inhibitor is a highly specialized procedure, beyond the ability of most centres. If half-life needs to be measured, the sampling schedule and methodology for calculation (Win-Non-Lin software; Pharsight, St Louis, MO, USA) used in the International ITI (I-ITI) study should be used so that patient outcomes can be compared to the results of that study (Hay & DiMichele, 2012). It should be recognized, however, that this sampling schedule (pre, 0·25–0·5, 1, 2, 4, 6, 24 and 48 h) may underestimate the half-life (Björkman et al, 2010). A definition of an inhibitor based on half-life is the hardest and most sensitive measure but is difficult to apply in most routine clinical circumstances. In view of this, and because of the challenge of measuring FVIII half-life in patients with low titre inhibitors in routine practice, we will use a pragmatic and clinically relevant surrogate measure of normal FVIII pharmacokinetics in this guideline, as a FVIII level ≥1 iu/dl at 48 h in an individual receiving standard prophylaxis (20–50 iu/kg on alternate days). An inhibitor test should be performed in severely affected patients with haemophilia A or B at least every third ED or every 3 months until the 20th ED (Grade 2C). After the 20th ED an inhibitor test should be done every 3–6 months up to 150 EDs. For haemophilia A, inhibitor testing should continue 1–2 times a year indefinitely (Grade 1C). For haemophilia B, testing after 150 EDs is only required if clinically indicated. An inhibitor test should be performed in all patients with haemophilia A before any change in concentrate and at least twice in the first 6 months after the change or if there is any change in bleeding pattern or response to FVIII (Grade 2C). An inhibitor test should be performed in mild and moderate haemophilia A yearly (if they have been exposed to FVIII) or after intensive exposure (≥5 EDs) or after surgery (Grade 1C). Patients with mild/moderate haemophilia A and a mutation with high inhibitor prevalence and/or family history of inhibitors should undergo inhibitor testing after all exposures (Grade 1C). Patients with haemophilia B should be tested after an allergic reaction to replacement therapy before any further FIX exposure occurs (Grade 1B). Tests to detect the presence or titre of an inhibitor should be done after a washout that ensures that the baseline factor level has been reached (Grade 1B). With currently available methodology it is difficult to accurately monitor FVIII half-life in patients with low titre inhibitors in routine clinical practice. If required, half-life should be measured by the methods described in the International Immune Tolerance study (Grade 2C). The current consensus definition for a FVIII inhibitor is an elimination half-life of <6 h, but this is likely to be an underestimate (Grade 2B) and the definition suggested in this guideline is <7 h (Grade 2B). We suggest that a pragmatic and clinically relevant surrogate measure of normal pharmacokinetics is a FVIII level ≥1 iu/dl at 48 h in an individual receiving standard prophylaxis (20–50 iu/kg on alternate days) (Grade 2C). There is no criterion for recognition of a FIX inhibitor other than the presence of a positive Bethesda assay (Grade 2C). IVR is a relatively inaccurate method to assess the strength of an inhibitor but is useful for guiding replacement therapy (Grade 2B). Inhibitor treatment involves the control and prevention of bleeds and strategies to eradicate the inhibitor. Immune tolerance induction (ITI) must be viewed as a long-term investment and the high initial cost compared with the cost of life-long treatment in the presence of a persistent inhibitor. Patients with a FVIII inhibitor, measured on more than one occasion, that interferes with prophylaxis or treatment of bleeds at standard doses of FVIII should undergo ITI to eliminate the inhibitor and restore normal clinical responsiveness to FVIII. Factors that potentially affect the outcome of ITI are listed in Table 2. Good risk patients are defined as having an inhibitor titre <10 BU/ml and an historic peak titre <200 BU/ml. Mariani et al (1994), Kroner (1999), DiMichele and Kroner (2002) Mariani et al (1994), Kroner (1999), DiMichele and Kroner (2002) Mauser-Bunschoten et al (1995) Poor risk patients 100–200 iu/kg/d probably more effective Good risk patients No difference between 200 iu/kg/d and 50 iu/kg 3 times a week but tolerance achieved more rapidly with higher dose regimens and with less intercurrent bleeds Mariani et al (1994) Kroner (1999), Mauser-Bunschoten et al (1995), Brackmann et al (1996), DiMichele and Kroner (2002) DiMichele (2003) Hay and DiMichele (2012) Poor risk if >250 BU/ml The best indicator of success or failure of ITI on multivariate analysis of the International ITI study No evidence of difference in first line ITI between plasma-derived (pd) and recombinant FVIII Uncontrolled reports of responses to pdFVIII after failure of first-line ITI Kreuz et al (1996), Gringeri et al (2007), Kurth et al (2011) Kroner (1999), Brackmann et al (1996) Mauser-Bunschoten et al (1995) Kreuz et al (1995), DiMichele (2003) It is important to avoid interruption to ITI and to follow a protocol closely because the first attempt at ITI carries a considerably greater chance of achieving long-term tolerance than rescue therapy (Lenk, 1999). ITI regimens should be reviewed by a haemophilia clinician every month, and more formally reviewed every 3 months by a clinician with expertise in ITI. Previous reports have suggested that most patients achieve tolerance within 6–12 months and a minority may take 1–3 years or more (Kreuz et al, 1995; Brackmann et al, 1996). The International ITI Study, however, found that, in good risk patients, the median time on ITI in the low-dose arm was 16·4 months and in the high-dose arm 14·2 months (Hay & DiMichele, 2012). Patients who are super-high responders (inhibitor titre rises to >500 BU/ml after starting ITI) usually have a poor outcome (DiMichele & Kroner, 2002; Hay & DiMichele, 2012). ITI can be abandoned in such patients after 6–9 months and second-line therapy considered (see section 6.1.7) unless there is evidence of a significant ongoing decline in inhibitor titre (at least a 20% fall in inhibitor titre in each 6 month period). Uncontrolled data have suggested that tolerance may be more readily achieved using low-purity pdFVIII than with recombinant FVIII (rFVIII) (Kreuz et al, 1996; Gringeri et al, 2007; Kurth et al, 2011). This remains controversial and there are a number of studies showing that the reported success-rates for ITI do not appear to be influenced by the product-type (Mauser-Bunschoten et al, 1995; Brackmann et al, 1996; Batlle et al, 1999; Rocino & de Biasi, 1999; Smith et al, 1999). A randomized comparison of the efficacy of high-dose pdFVIII or rFVIII for ITI in poor-risk patients is in progress (Gringeri, 2007). First-line ITI should be conducted using rFVIII concentrate, unless as part of a clinical trial, and is usually performed with the product used by the patient at the time of inhibitor development. Starting titre is the most powerful predictor of ITI success (Mariani et al, 1994; DiMichele & Kroner, 2002) and regimens that delay treatment until the inhibitor has fallen below 10 BU/ml show very high success-rates (Mauser-Bunschoten et al, 1995; Rocino & de Biasi, 1999; Smith et al, 1999). It took a median of 5 months from diagnosis for titres to fall to <10 BU/ml in the I-ITI study (Hay & DiMichele, 2012). Rate of response to ITI did not decline until ITI had been delayed for 5 years from the time of diagnosis in the North American ITI registry (NAITR) (DiMichele & Kroner, 2002). Inhibitors that fail to fall to <10 BU/ml over 12–24 months often respond less well to ITI. Bleeds should be treated with activated recombinant FVII (rFVIIa) during this time to avoid an anamnestic response. Immune toleration induction should therefore be delayed until the inhibitor titre has fallen below 10 BU/ml. If the inhibitor is <10 BU/ml when first detected, and does not rise above 10 BU/ml in the subsequent 1–2 weeks, ITI should be started promptly, highlighting one of the benefits of good surveillance and early inhibitor detection. Commencement of ITI should also be considered if the inhibitor titre has not fallen below 10 BU/ml within 1 year, or if there has not been a downward trend in titre during these first 12 months. A central venous access device (CVAD) is usually inserted to facilitate ITI. Some centres attempt ITI without the use of a CVAD, because infection has been suggested to adversely affect the outcome of ITI, especially in poor risk patients. The I-ITI study observed that, in good-risk patients, infection or CVAD placement had no effect on either the proportion achieving tolerance or the time taken to become tolerant (Hay & DiMichele, 2012). Implantable CVADs are significantly less likely to become infected during ITI than external lines, such as Hickman or Broviac catheters (Hay & DiMichele, 2012). The choice of ITI regimen remains problematic. The I-ITI and NAITI suggest that poor-risk patients (peak titre >200 BU/ml, starting titre >10 BU/ml) are best tolerized using a high-dose regimen (100–200 iu/kg/d FVIII) (Mariani et al, 1994; DiMichele & Kroner, 2002). These registries and the I-ITI study suggest that high dose and low-dose (50 iu/kg three times weekly) regimens are equally effective in inducing tolerance in good risk patients (Mariani et al, 1994; Mauser-Bunschoten et al, 1995; DiMichele & Kroner, 2002; Hay & DiMichele, 2012). By implication, therefore, 200 and 100 iu/kg/d can be assumed to be equally efficacious for inducing tolerance in good risk patients (Hay & DiMichele, 2012) but the relative effect of these high dose regimens on bleeding is unknown. Low-dose ITI (50 iu/kg three times a week or on alternate days) takes longer to achieve a negative Bethesda titre (DiMichele & Kroner, 2002; Hay & DiMichele, 2012) and is associated with significantly more intercurrent bleeding before the Bethesda titre becomes negative, a period during which 85% of intercurrent bleeding on ITI takes place (Hay & DiMichele, 2012). Low-titre inhibitors (historic peak titre <5 BU/ml) are usually readily tolerized using a low-dose regimen (50 iu/kg alternate day) (Ter Avest et al, 2010). Most published ITI regimens, with occasional exceptions (Smith et al, 1999), maintain the same dose of FVIII until the patient is considered tolerant. Dose tailoring, however, has been used on an empirical basis by some clinicians and three observations from the I-ITI study are relevant to this: (i) the outcome of ITI is unrelated to dose in good risk patients; (ii) high-dose ITI is associated with a statistically significant reduction in bleeding only in the early phase of ITI; (iii) although high-dose patients achieve a negative Bethesda titre three times faster than low-dose, the time taken to achieve the subsequent milestones of normal recovery and half-life were similar (Hay & DiMichele, 2012). These findings suggests that it might be possible, having started with high-dose ITI, to reduce the dose of FVIII during the course of ITI without affecting the time taken to achieve tolerance, as long as intercurrent bleeding is minimized and joint function preserved. It is important to note that, in the I-ITI study, the number of bleeds occurring between the time of the first negative Bethesda assay and tolerance was 56 in the low-dose arm and 7 in the high-dose. This did not reach statistical significance as there were relatively few patients available for analysis during these phases of the trial and the study may not have had sufficient power to detect a difference. The effect on long-term joint outcome of this difference in bleed number is not known. Most clinicians consider that all patients with a historic peak titre >5 BU/ml requiring ITI should be started with high-dose to minimize inter-current bleeding. However, in good risk patients, dose reduction in stages can reduce costs as long as there is no subsequent increase in break-through bleeds that require by-passing therapy. This strategy would not be expected to increase the time taken to achieve tolerance (Hay & DiMichele, 2012). Although not tested in a controlled trial, a similar regimen has been reported to be successful in a small series of patients (Smith et al, 1999). There is limited evidence to guide recommendations for the conduct of ITI and the following is a pragmatic, practical consensus method. It recognizes the fact that the accurate measurement of FVIII half-life in patients with low-titre inhibitors is difficult for most haemophilia centres, that the normal FVIII half-life of an individual patient is unknown and that a FVIII half-life of 6 h is likely to be too short to be a suitable criterion for tolerance. The definition used for restoration of normal pharmacokinetics is, therefore, a post-washout half-life of >7 h or a measureable FVIII trough level at 48 h in an individual receiving standard prophylaxis (20–50 iu/kg). The regimen supports the consideration of dose reduction, whilst aiming to minimize joint bleeds and preserving long-term joint status, by tailoring to a 24- or 48-h- trough level ≥1 iu/dl, once these have become measurable (Collins