‘It takes 50 years to get a wrong idea out of medicine, and 100 years a right one into medicine.’ (Hughlings Jackson)
While these time frames have probably contracted, this century-old quotation is still germane. More than 15 years ago, a prospective, controlled study suggested that anti-lymphocyte globulin [ALG or anti-thymocyte globulin (ATG)] was superior to allogeneic transplantation for severe aplastic anaemia (SAA) (Speck et al, 1986). Although initially embraced by many, the conclusions of the study have not stood the test of time, and the recently published guidelines for the management of SAA (Marsh et al, 2003) now list allogeneic transplantation as the preferred treatment for young patients with matched siblings.
These guidelines also stated the following: ‘The use of high dose cyclophosphamide without stem cell support cannot be recommended in either newly diagnosed patients or patients who have failed ATG and ciclosporin in view of its serious toxicity and high mortality’ (Marsh et al, 2003). This recommendation, one of only two with a grade A rating, is based on a single trial comparing high-dose cyclophosphamide + ciclosporin to ATG + ciclosporin (Tisdale et al, 2000). The investigators closed the trial early (after accruing only 31 of the planned 182 patients) because of excess toxicity in the cyclophosphamide + ciclosporin arm; however, this was without the recommendation of a Data and Safety Monitoring Board or the Institutional Review Board, as no stopping rules were met and there was no significant difference in overall mortality between the arms. Of note, at the actual enrolment rate of 10 patients/year, completion of the trial would have taken 18 years.
This trial had other serious shortcomings. It was performed by a single institution inexperienced with this intensive approach. Few would give credence to a study of allogeneic transplantation in SAA performed by investigators without allogeneic transplantation experience. In fact, the treatment differed substantially from the promising experience with high-dose cyclophosphamide in SAA (Brodsky et al, 2001): growth factor support was not routinely used, and ciclosporin was used concomitantly with, and for 6 months after, cyclophosphamide (Tisdale et al, 2000). When combined with high-dose cyclophosphamide, ciclosporin both increases toxicity (Deeg et al, 1986) and blocks a key mechanism of its action, induction of immunological tolerance (Mayumi et al, 1996). A follow-up report also showed that the two arms of the randomized trial were not balanced. A patient with clinical paroxysmal nocturnal haemoglobinuria (PNH), a trial exclusion, was compassionately assigned to cyclophosphamide + ciclosporin. This arm also included two other patients with 23% and 83% PNH cells, while no patient on ATG/ciclosporin had this high a percentage of PNH cells. These patients accounted for two of the three treatment-related deaths with cyclophosphamide + ciclosporin. As the guidelines point out, SAA patients with ‘small’ PNH clones should be treated exactly the same way as those lacking PNH clones. However, patients with large PNH clones are probably another matter and their inclusion could certainly skew the results of a small study.
Although major advances in SAA, allogeneic transplantation is limited by donor availability and graft versus host disease, especially in older patients, and ATG/ciclosporin by high rates of relapse and clonal evolution (Speck et al, 1986). In a study involving three centres, not one as the guidelines stated (Marsh et al, 2003), high-dose cyclophosphamide produced long-term control of SAA in more than 80% of patients whose median age was 47 years (Brodsky et al, 2001). The mortality is now <10%, with no relapses or clonal evolution. Not only are these the best published results in older SAA patients, but also several other groups have now confirmed effectiveness of this treatment in smaller numbers of patients (Jaime-Perez et al, 2001).
The results of well-designed observational studies are often as valid as randomized, controlled trials, while the results of a single clinical trial, even when randomized, are subject to limitations in study design and execution and thus should be interpreted cautiously (Concato et al, 2000). Moreover, the SAA guidelines (Marsh et al, 2003) proscribed the use of high-dose cyclophosphamide (Brodsky et al, 2001), although the single, small randomized trial on which the recommendation was based employed a significantly different regimen (Tisdale et al, 2000). A more balanced recommendation would be that SAA patients should only receive high-dose cyclophosphamide as part of a trial and only by those experienced with this approach.