- Top of page
- Patients and methods
The potential effect of age at the start of replacement therapy on the development of factor VIII (FVIII) inhibitors was assessed in 62 severe (FVIII < 2 IU/dl) haemophilia A patients who started FVIII therapy at one of two haemophilia centres. Inhibitors were tested on an annual basis. Persistent or high-titre inhibitors were detected in 15 patients (24%). Kaplan–Meier cumulative incidence at 3 years from first FVIII exposure was 41% (95% CI 22–67%) in patients starting therapy before the age of 6 months, 29% (95% CI 13–57%) in patients starting therapy between 6 and 12 months of age, and 12% (95% CI 4–34%) in those starting therapy beyond 1 year of age (P = 0·03). By multivariate analysis, the influence of age was shown to be independent of other variables, including calendar year at the onset of therapy and baseline FVIII plasma levels. In conclusion, patient age at initial treatment appears to influence inhibitor formation. If confirmed, this finding would have a major impact on the management of haemophilia.
A number of haemophilia A patients develop factor VIII (FVIII) inhibitors after exposure to FVIII concentrates, which makes them unresponsive to replacement therapy. Overall, FVIII inhibitors appear in 20–25% of severe haemophiliacs (Scharrer et al, 1999), although higher figures have been reported (Ehrenforth et al, 1992). It is widely accepted that inhibitor incidence is influenced by haemophilia severity (Ehrenforth et al, 1992; Addiego et al, 1993; Lusher et al, 1993; Hoyer, 1995) and mutation type (Schwaab et al, 1995; Scharrer et al, 1999), but no other risk factors have been clearly identified (Scharrer et al, 1999). Probably because most inhibitors appear during childhood, age is often considered to influence inhibitor development (Penner, 1999; Scharrer et al, 1999). However, there is no evidence that age itself (age at initial treatment) is a risk factor for inhibitor development, and the predominance of inhibitors during childhood is likely to be associated with the onset of FVIII exposure.
The aim of this study was to analyse the potential influence of age at initial therapy on the incidence of FVIII inhibitors in severe haemophilia A.
Patients and methods
- Top of page
- Patients and methods
The study included 62 previously untreated severe haemophilia A patients (FVIII < 2 IU/dl) from two Spanish haemophilia centres. The first group comprised 47 consecutive patients born after January 1975 and first treated with FVIII concentrates before December 1996 at the Unit of Congenital Coagulation Disorders in Valencia. High-purity FVIII concentrates have been introduced in this unit progressively since 1988 in place of low- or intermediate-purity concentrates. Recombinant FVIII has also been used increasingly since 1995. Fourteen of these patients became HIV infected during the study period before 1986. The second group consisted of 15 HIV-negative patients born in 1990 or later and first treated before 1997 at the Haemophilia Unit of Vall d'Hebrón in Barcelona. Monoclonal or recombinant (since 1995) FVIII concentrates were used in the latter centre.
In both centres, follow-up visits were made three times a year. Inhibitor screening was carried out at least once a year, as well as before surgical procedures and whenever an inhibitor was clinically suspected. Once an inhibitor was detected, it was assayed at each visit. An ‘exposure day’ (ED) was defined as a day on which at least one FVIII infusion was given. The number of exposure days was counted up to a maximum of 75 d.
Factor VIII activity was measured by a one-stage clotting assay. The Barcelona cohort was screened for inhibitors by the Bethesda assay (Kasper et al, 1975). In the Valencia cohort, patients were screened for inhibitors by one of the following three methods while they remained inhibitor negative: Bethesda assay (until 1985); a modified agarose gel diffusion method that had shown high sensitivity and an excellent correlation with the Bethesda assay (Jorquera et al, 1985) (between 1985 and 1993); or a previously described sensitive mixing clotting test (Ewing & Kasper, 1982) (since 1993). All inhibitor-positive patients and all samples with a positive screening test were systematically tested by the Bethesda assay as the reference method. Results over 0·6 Bethesda units (BU)/ml were considered to be positive, but only clinically relevant inhibitors were computed as inhibitor cases. An inhibitor was defined as clinically relevant if its peak titre reached 10 BU/ml or if it persisted > 1 year, regardless of the titre. Inversions in intron 22 of the FVIII gene were studied in 40 patients; other mutations are being investigated, but results are not yet available.
The cumulative incidence of inhibitors was estimated by Kaplan–Meier analysis. Observations were censored on the date when the patients were last known to have no inhibitor, and subgroups were compared by Breslow's test (with test for trend when appropriate). Multivariate analysis was carried out by Cox's proportional hazards regression, with a stepwise selection procedure based on the maximized log-likelihood. P-values below 0·05 were considered to be statistically significant.
- Top of page
- Patients and methods
Descriptive patient data are shown in Table I. FVIII level was < 1 IU/dl in 56 of the 62 patients. Four patients (6·5%) had < 10 ED, 17 (27%) had between 10 and 50 ED and 41 (66%) had > 50 ED. No significant differences were observed among the three age groups (age at initial therapy ≤ 6, 7–12 and > 12 months) regarding ED or baseline FVIII level. An inversion in intron 22 was found in 19 of the 40 patients studied (47%). The mean baseline FVIII level and the mean age at first FVIII infusion were similar in patients with and without inversions.
Table I. Descriptive features of the patients, grouped by age at first FVIII exposure.
| ||Age at first FVIII exposure|| |
| 6 months (n = 17)||7–12 months (n = 17)||> 12 months (n = 28)||All patients (n = 62)|
|Age at first infusion (months)|
| Median (range)||2·6 (0–6)||9·5 (7·4–12·9)||16·6 (13·2–65·7)||10·5 (0–65·7)|
| Mean (95% CI)||2·2 (1·2–3·2)||9·5 (8·7–10·2)||21·9 (16·8–27)||13·2 (9·4–17·1)|
| Median (range)||> 75 (7–> 75)||> 75 (7–> 75)||> 75 (6–> 75)||> 75 (6 –> 75)|
| Mean (95% CI)||52 (37–67)||61 (48–73)||57 (46–67)||57 (50–63)|
| Median (range)||56 (0·5–230)||84 (9–209)||65 (2–247)||65 (0·5–247)|
| Mean (95% CI)||87 (46–127)||83 (50–116)||88 (61–114)||86 (69–104)|
|Baseline FVIII level (IU/dl)|
| Median (range)||0·5 (0·1–1·4)||0·5 (0·1–1·9)||0·6 (0·1–1·9)||0·5 (0·1–1·9)|
| Mean (95% CI)||0·6 (0·4–0·7)||0·7 (0·4–0·9)||0·7 (0·5–0·8)||0·6 (0·5–0·7)|
Fifteen patients developed clinically significant inhibitors, as defined above, between 0·5 and 35·5 months after first FVIII exposure (median 13 months) and after 6 to > 75 ED (median 21 ED). The estimated probability of inhibitor development was 25% (95% CI 16–39%) at 3 years from first FVIII exposure (Fig 1). No inhibitors developed beyond the third year after first FVIII infusion. The 3-year cumulative incidence of inhibitors was 41% (95% CI 22–67%) in patients who had received the first FVIII infusion before the age of 6 months, 29% (95% CI 13–57%) in patients who had received the first infusion between 6 and 12 months of age and 12% (95% CI 4–34%) in patients who had started treatment after 12 months of age (P = 0·03). No significant differences in inhibitor incidence were observed according to calendar year at first FVIII exposure (before or after 1985), baseline FVIII level (under or over 1 IU/dl) or treatment centre (3-year incidence was 33% in the Barcelona cohort and 23% in the Valencia cohort).
Multivariate analysis of age at first FVIII exposure, calendar year at first exposure (before or after 1985) and baseline FVIII level (greater or less than 1 IU/dl) showed age to be the only significant covariate, being inversely related to the probability of inhibitor development (P = 0·03). The relative risk was 0·51 (95% CI 0·27–0·94) for each category increase in the age covariate. Age at start of therapy remained as a significant predictor of inhibitor risk after adjustment for calendar year at first FVIII exposure, baseline FVIII level and treatment centre.
By computing the observation period in terms of ED, the cumulative incidence of inhibitors at 50 ED was 24%, and the corresponding results according to the age at first FVIII exposure remained practically the same as those obtained with the observation period computed as time from first exposure.
The frequency of inhibitors in patients with intron 22 inversion was 26% (5/19) and in those without inversions 29% (6/21).
- Top of page
- Patients and methods
Patient age is generally accepted to be an important risk factor for inhibitor development (Ehrenforth et al, 1992; Kavakli et al, 1998; Scharrer et al, 1998). However, this assumption might derive from a misinterpretation of the fact that most inhibitors appear during infancy or childhood, when exposure to FVIII usually starts. Probably, it is the onset of therapy early in life that makes age appear to be a risk factor. On the other hand, if age is to be considered a true risk factor, the incidence of inhibitor development should depend on the age when therapy is started. Although age at onset of therapy has been mentioned as a potential risk factor for inhibitors (Scharrer et al, 1999), and some studies may have analysed this question marginally (Kavakli et al, 1998), no studies have focused specifically on the question of the age at the start of therapy as a potential risk factor for inhibitor development in haemophilia.
Our study shows that haemophilic patients starting therapy very early in life have a higher incidence of inhibitors. However, we cannot determine whether delaying therapy beyond 3 years of age would cause additional reduction in the risk of inhibitor development, as the majority of patients started therapy during the first three years of life.
Some data suggesting that age at start of therapy is a risk factor for inhibitor formation can be found in the literature. A detailed analysis of the individual data provided in two recent co-operative French studies, including over 50 severe (< 1 IU/dl) haemophilia A patients each (Guérois et al, 1995; Rothschild et al, 1998), additionally supports our results. In those two studies, the proportions of inhibitors in patients under the age of 1 year at first FVIII infusion were 11/32 and 5/31, respectively, compared with 3/18 and 0/25 in patients over the age of 1 year at first infusion respectively. The application of the Kaplan–Meier method to these data shows a significant influence of age on inhibitor risk (data not shown).
The potential causes of a higher inhibitor risk when therapy is started within the first 6–12 months of life are not known. Rather, during the neonatal period, the immune system is thought to be immature, and immune tolerance to foreign antigens would be more likely to occur. A possible explanation for the higher inhibitor risk in patients treated early in life would be that the time elapsed until the first FVIII infusion is required might depend on the existence of minimal differences in plasma FVIII antigen concentrations, even in severe haemophiliacs. Thus, patients with extremely low or completely absent plasma FVIII might be at a higher risk of inhibitor development than other severely affected haemophiliacs in whom plasma FVIII is present at slightly higher concentrations. This concurs with the generally accepted idea that a very small concentration of the endogenous FVIII molecule in plasma may prevent inhibitor formation and could also be related to the mutation type. Thus, age at start of therapy could also be a surrogate marker for mutations that cause complete absence of FVIII molecule in plasma and predispose to inhibitor formation (e.g. large deletions). The intron 22 inversions of the FVIII gene were not included in the analysis of cumulative incidence of inhibitors because they were not available for a third of the patients. However, the proportion of inhibitor patients did not differ significantly according to the presence or absence of intron 22 inversions.
In summary, although further studies are required to confirm the importance of age at initial therapy for inhibitor development in haemophilia, the results of this work suggest that starting haemophilia replacement therapy very early in life might increase the risk of inhibitor formation. If so, exposure to FVIII concentrates should perhaps be delayed, and bleeding during the first 6–12 months of life might be treated with non-FVIII-containing products, such as recombinant activated factor VII. Moreover, future studies on inhibitor incidence should take into account the age at start of therapy, particularly when examining other potential risk factors, such as haemophilia severity, mutations or the type of factor VIII concentrate.