Relapsed Lymphoblastic Leukaemia in Children: A Continuing Challenge

The treatment of childhood lymphoblastic leukaemia has been one of the success stories of haematology, but despite this triumph 25–30% of children still relapse. Relapsed lymphoblastic leukaemia (R-ALL) is thus as common as most paediatric solid tumours, and more common than acute myeloid leukaemia in childhood. The treatment of R-ALL is usually unsuccessful, causes heartache and stress for both families and staff, and consumes large amounts of time and resources. There have been few randomized therapeutic trials to inform the choice of treatment and intense medical and lay interest in high-dose therapy and all variants of stem cell rescue. The increased availability of bone marrow or stem cells donors raises the vision that, in theory, BMT could be available to all children with R-ALL at least in societies whose resources permit this option. Yet high-dose therapy carries at least a 10–20% risk of immediate death and a much greater risk of delayed complications such as growth failure, infertility, organ damage and neuropsychological impairment, and is itself associated with a high relapse rate. This article will attempt to review the therapeutic options for the child with R-ALL, to make recommendations based on the limited evidence available about choice of treatment, and to suggest priority areas for research. It must, however, be recognized that this whole topic is clouded by a lack of randomized trials, difficulties over selection bias, and publications involving small numbers of patients. References were retrieved from the Silver Platter CANCER-CD database from 1987 until mid-1997, using the terms lymphoblastic and lymphocytic leukaemia and relapse and recurrence; they were confined to English Language publications. These were supplemented by references from the author's own database, articles and abstracts cited in those found in the literature search, and by personal communication with groups working on this area. It is usual to classify relapses as early, intermediate and late as determined by the length of the first remission, or alternatively, in the case of relapses occurring off treatment, the time elapsing since treatment was electively stopped. There is no uniform definition of these terms and not all publications include a class of intermediate relapse, but the definition of early relapse always includes those occurring in the first 18 months from diagnosis and usually in the first 2 years. Most relapses in ALL occur during the first 5 years from diagnosis, but in a population-based study from the Nordic countries 6.9% of children who were still in first remission at 5 years from diagnosis had relapsed by 10 years ( Nygaard et al, 1989 ). Combined relapses, that is those with overt bone marrow and extramedullary relapse, tend to occur later than isolated marrow relapses ( Gaynon et al, 1998 ). Isolated central nervous system (CNS) relapse tends to occur within 3 years from diagnosis ( Gaynon et al, 1998 ) and testicular relapses after treatment is electively stopped. Other sites of recurrence such as the iris, skin, lymph nodes or ovaries are rare and may occur in apparent isolation or with obvious marrow involvement. The overall outlook for patients with R-ALL has always been poor ( Cornbleet & Chessells, 1978; Chessells et al, 1987 ) and remains so. Table I shows the event-free survival (EFS) in second remission figures from recent publications selected either because they are population based (Holland, Scandinavia) or they report the follow-up of all patients relapsing after large first-line trials of treatment (American Children's Cancer Group CCG and MRC UKALL protocols). These reports confirm the widespread clinical impression of the very poor prognosis for children with early marrow relapse, which has not improved in the last 20 years. In the era of modern treatment, fewer than 1 in 10 children with a marrow relapse during the first 1–3 years from diagnosis are cured. The outlook is somewhat better following later bone marrow or extramedullary relapses, but very-long-term follow-up is needed to assess the results of a second course of chemotherapy in such cases, since second recurrences may occur many years from diagnosis ( Chessells et al, 1994 ; Miniero et al, 1995 ). The most important factors influencing the chance of remaining in second remission are the length of the first remission (Fig 1) and the type of relapse (Fig 2) ( Wheeler et al, 1998 ; Gaynon et al, 1998 ). Despite the fact that there is frequently molecular evidence of disease in the bone marrow of children with apparent isolated extramedullary relapse ( Goulden et al, 1994 ; Neale et al, 1994 ), patients with isolated extramedullary relapse as defined by conventional bone marrow examination, in particular those involving the testicle, have a good prognosis ( Wheeler et al, 1998 ; Gaynon et al, 1998 ). It has been suggested that children with combined marrow relapses have a better prognosis than those with bone marrow disease only ( Buhrer et al, 1993 ), but this may be in part related to their relative lengths of first remission ( Gaynon et al, 1998 ). 997; no patient has been lost to follow-up. Fig 2. Survival by type of relapse in the same group of patients as in Fig 1. Older patients ( Wheeler et al, 1998 ) and those with T-ALL ( Henze et al, 1991 ; Wheeler et al, 1998 ) have a worse prognosis: no patient in the Berlin-Frankfurt-Munster (BFM) relapse study with a relapse within 18 months of diagnosis or with T-ALL has survived beyond 15 months ( Henze et al, 1991 ). The length of second remission, in contrast to that of first remission, is not influenced by presenting leucocyte count or by gender ( Henze et al, 1991 ; Wheeler et al, 1998 ). The primary treatment of ALL has become progressively more intensive over the last 10 years, and it might be expected that the intensity of initial treatment would influence the duration of second remission and survival. Both the recent MRC UKALL X and the American Children's Cancer Group (CCG) 100 series of protocols involved randomization to one or more blocks of intensive treatment, and in neither was there evidence that the intensity of previous treatment had a significant influence on survival ( Wheeler et al, 1998 ; Gaynon et al, 1998 ). The other recent change in management of ALL has been the avoidance of cranial irradiation and reliance on regular intrathecal chemotherapy for CNS-directed therapy. The CCG 105 protocol randomized a large number of children with intermediate-risk ALL to receive regular intrathecal methotrexate with or without additional cranial irradiation ( Tubergen et al, 1993 ). The results showed that intrathecal methotrexate, in combination with intensified systemic therapy, afforded equivalent CNS protection in the younger child to that provided by additional cranial irradiation. Patients with CNS relapse despite cranial irradiation had an increased risk of death in comparison with unirradiated patients, but this did not reach statistical significance ( Gaynon et al, 1998 ). A recent review of the results of treatment of R-ALL by the BFM study group ( Buhrer et al, 1996 ) showed that, among 260 children with isolated marrow relapse occurring over 6 months off therapy, the 38 without circulating blasts at the time of recurrence had a significantly better 10-year probability of EFS at 64% (SD 11) than the 171 with up to 10 × 109/l at 32% (SD 5) or the 51 with more than 10 × 109/l at 10% (SD 6). Only a length of first remission >48 months was an additional prognostic factor in this subset of patients. This finding, which relies on the differential leucocyte count, contrasts with previous reports where leucocyte count at the time of relapse had no prognostic significance ( Henze et al, 1991 ; Wheeler et al, 1998 ) and requires confirmation by further studies. Patients with relapsed ALL often exhibit morphological shifts from the small L1 blasts at diagnosis to the larger more pleomorphic L2 blasts, which in untreated ALL have a worse prognosis ( Abshire et al, 1992 ; Lilleyman et al, 1995 ) and may show shifts in immunophenotype, usually to a more undifferentiated type, with, for example, loss of HLA-DR or CD10 antigen ( Abshire et al, 1992 ; van Wering et al, 1995 ). Sequential study of cytogenetics disclosed major changes in karyotype in 10% of patients in one study from St Jude Children's Research Hospital, but these patients had received very intensive treatment and developed therapy-induced secondary leukaemia ( Raimondi et al, 1993 ). Less obvious evolution in karyotype has been reported in other groups of patients ( Heerema et al, 1992 ; Farah et al, 1997 ), and it has been claimed that change in karyotype is associated with a worse prognosis after relapse ( Vora et al, 1996 ), a report not confirmed by other studies ( Farah et al, 1997 ). Molecular genetics may also influence prognosis; it has recently been reported that children with R-ALL in association with the TEL-AML 1 fusion transcript had significantly longer second remissions than those without ( Seeger et al, 1997 ). Analysis of cellular drug resistance has shown that cells from children with relapsed ALL are more resistant to steroids, L-asparaginase, and than those with leukaemia ( et al, ). These the clinical that disease after relapse is more to therapy and the that further studies of the of relapsed ALL may to with and L-asparaginase, a remission in of children with lymphoblastic leukaemia ( et al, ), but in fewer of those who relapse. Table shows the results of more intensive of therapy in children who have relapsed on more recent first-line and it be that the of an the remission to in most of the in children who have been on modern intensive Most of these protocols remission after of these with additional therapy after this The BFM relapse protocol more with for patients that remission is later in shown in Table most of the protocols the of however, their may in be by have been during first-line therapy ( et al, 1992 ). There have been no recent randomized trials of therapy for that recently by the Cancer Group ( et al, 1996 ), the of which are shown in Table A of at more than at and the was to 10 ; however, event-free survival significantly at for children who received the of to 10% for and for the of The was not at 3 years and the overall despite intensive chemotherapy were poor Table protocols for relapsed ALL were by a high of CNS relapse, and the for CNS-directed therapy apparent in these cases, as in lymphoblastic leukaemia ( Chessells et al, ; Buhrer et al, 1994 ). Most protocols have regular intrathecal methotrexate, with or without and and in or high The BFM study group randomized patients with relapsed ALL to receive either methotrexate as a or 1 as a the randomization was after it apparent that the patients high-dose did not better systemic or CNS protection than methotrexate ( Henze et al, 1991 ). The group have cranial irradiation and intrathecal chemotherapy in children with late isolated relapse because of their about the risk of CNS relapse, but this has been and not in a randomized ( Buhrer et al, 1994 ). There evidence that children who relapse those at receive CNS-directed therapy, but the relative additional of high-dose systemic chemotherapy cranial irradiation in to intensive intrathecal chemotherapy The survival figures for patients with ALL have been shown in Table Table the results of recently intensive treatment protocols for such all additional to and methotrexate The number of patients and the of randomized studies is but it is apparent from as from population-based that few children with early marrow relapses to intensive treatment and that a of such children relapsing later may be cured. example, in the study by et ) in of the patients, whose initial remission was years, the EFS was The event-free survival for late relapses, defined in Table from to but the as have involved small of patients. It is more than 20 years since the that patients with leukaemia could be by high-dose and from a ( et al, ). BMT has been as a of treatment for ALL in second remission for many years ( et al, 1987 ), but the for BMT in ALL are still not The of publications on this topic from or they an selection because of the of second remissions ( Chessells et al, ) and the fact that most patients lack a has in recent years ( et al, 1996 ) and the to success in as in other of treatment, is relapse of The BMT recently the factors influencing of BMT in second remission of ALL from a length of previous remission, initial leucocyte count, and previous CNS as prognostic factors ( et al, 1994 the overall risk of relapse in patients was 11) with no relapses beyond years. The most for BMT remains the combination of and but a of other protocols have been as in Table as shown in Table most of these reports are studies involving small numbers of patients and there is a of randomized trials There is no evidence that one is to the the relapse has been reported after the of irradiation and ( et al, 1987 ). The late of irradiation are in the and include growth ( et al, ; et al, 1987 ), in children who have received previous cranial significant neuropsychological ( et al, ; et al, 1994 ). there has been of which in there have been few reports of their in The of patients of have a and this has to interest in BMT has a much than BMT but in contrast to chemotherapy it however, associated with a high risk of relapse. The as are and the event-free survival in a group of 51 children with first remissions of at least months was after high-dose irradiation and of however, the patients were a selected group since the most significant of survival was length of first remission and the duration of second remission ( et al, 1993 ). recently the of donors has been by the of ( et al, 1995 ) and other such as stem recent publications ( et al, 1996 ; et al, 1997 ), both from have reported event-free survival for of and In the of these studies the event-free survival for a group of children in second remission of ALL was to that of children with at 2 years et al, A further recent report that may remissions in children in remission ( et al, 1997 ). There are no randomized trials chemotherapy and high-dose therapy in the management of relapsed and all other of comparison are by selection for one small early study ( Chessells et al, ) there have been no on an for patients with R-ALL with and without have been to BMT and chemotherapy since the early ( et al, ; et al, ), and the first comparison from the suggested that BMT was to chemotherapy in children with a first remission of months ( et al, 1987 ). Table the studies reported in the last 10 years which have involved a comparison of chemotherapy and comparison on a but all have of attempt to for in terms of time to Most small numbers of patients. The study ( et al, 1991 ) and the study ( et al, ) both that BMT was to chemotherapy for patients with early The study involved a comparison ( et al, 1994 ) children with American Group protocols and 1991 and children a BMT in second remission of ALL during that time who were reported to the to associated with treatment failure, an attempt was to of children for these The survival at 5 years for the whole BMT group was and for the chemotherapy group a significant The results for are shown in Table and a significant for BMT was shown both for early and late relapse, and also in the of children with an initial relapse >48 months or the 18 with remissions Wheeler et ) have to the survival for prognostic factors and time to and have also showed for BMT of the length of first It be from Table that survival after BMT usually to this not reach conventional statistical significance in patients with late relapse, despite the by the of these is to be from which from these studies is the relative of children who received a of these was the the of families and the bone marrow and The in the is with this in patients relapsing from MRC UKALL from a and from In a of patients in and the as from results ( et al, 1996 ; Henze et al, 1991 ), are with BMT in second Most reports not a of results of BMT by length of first The of and the lack of have to interest in the of in relapsed Table shows the results of studies with of treatment of these studies was The early study by et ) patients to a and was not confined to patients in second was high for the BMT group and the relapse was for the The by et ) was confined to patients with late relapse and only a small number of The Research on childhood leukaemia to a randomized and chemotherapy in children with no but only 15 patients were for leucocyte count and length of first remission, as shown in and show that BMT is to chemotherapy and that there is no significant and by the BFM group ( et al, ) and a study by the group ( et al, ) showed no significant in and there is available A report from the in patients with leukaemia who received either BMT from an or involved a very group of patients and showed a for both ( et al, 1994 ). A more recent from involved a larger number of patients with the of were in second remission and as as survival was in the groups but was in the BMT group and relapse was ( et al, 1997 ). In in recent studies of BMT was in the of ( et al, 1997 ; et al, 1996 ). The outlook following a first of central nervous system leukaemia is by a report from the CCG with a of 10 years follow-up ( et al, 1987 ). children in a randomized of CNS-directed therapy, which one of intrathecal methotrexate developed CNS relapse as a first The survival after CNS relapse was only most patients developed a marrow relapse or CNS The time to initial CNS relapse was the most important factor for an confirmed by more recent reports where all children have received CNS-directed therapy in first remission ( Gaynon et al, 1998 ; Wheeler et al, 1998 ). with a first CNS relapse are at high risk of marrow relapse ( & Chessells, et al, ; et al, 1989 ), further of systemic therapy is as important as of CNS There are a number of in the treatment for CNS relapse. The of isolated CNS relapse as a first is trials of therapy small numbers of patients or widespread There has been a major in primary CNS-directed treatment, fewer children receive cranial irradiation. A further is the lack of an definition of CNS since a cell count with blasts has been for diagnosis in studies ( et al, ; et al, 1995 ) but not in ( Tubergen et al, 1994 ). Table shows the results of treatment for CNS relapse in recent the of which by is that by the Group where CNS treatment was uniform but patients were randomized to one of for systemic therapy. The event-free survival was with no statistical the treatment length of first remission was an important prognostic factor and the major of was bone marrow relapse. This showed that CNS be in most children without to irradiation but the for intensified systemic therapy ( et al, 1993 ). In of the overall poor prognosis of early CNS relapse, bone marrow could be a but there are few reports of results of BMT in this A comparison chemotherapy and BMT or from the BFM group showed no significant BMT and and results for however, there were only children in the group ( et al, ). The group have suggested that may be an treatment for isolated extramedullary relapse in first remission and the event-free survival for children in second remission was ( et al, 1996 ). This group have more recently reported the of in patients with early CNS relapse which with chemotherapy ( et al, 1998 ) ( Table The most common of relapse is the testicle, usually presenting within a after of therapy. A number of previous protocols for ALL involved of treatment testicular and publications may include of either or overt testicular results of treatment of testicular relapse are shown in Table all show event-free survival for patients with late isolated testicular relapse with further intensive systemic and intrathecal chemotherapy and irradiation. Patients relapsing have a worse These results that conventional therapy is for children with late relapse. A recent report from ( et al, 1997 ) in testicular irradiation was in of high-dose methotrexate, with apparent testicular in the ovaries in usually presenting with an most reported have within the first 3 years and patients have survived intensive ( et al, 1991 ). relapses frequently within the first off treatment ( et al, 1987 ) with and irradiation and intensive treatment may in patients, but there is a high risk of marrow relapse ( et al, 1994 ). to treatment in R-ALL have not been such as methotrexate and ( et al, 1991 ), and ( et al, 1997 ) and with and ( et al, 1997 ) have shown in or second relapse of ALL and there has also been interest in ( et al, 1993 ). no combination of has been shown to be of significant and the limited studies of the in have at least significant resistance the ( et al, ). of resistance and ( & have been to myeloid and a study in childhood ALL has not of resistance in ( et al, 1992 ). is also the major to and from donors it is that such donors may to There has been recent interest in the of to a despite the of of disease and This treatment has the most results in myeloid leukaemia where remissions have been but has been in ALL ( et al, 1996 ; et al, 1997 ). of children with ALL relapsing after BMT remission following at the of disease in ( et al, 1997 ). there are in the of from but this of treatment is in management and may relapse after is a second bone marrow after first remissions ( & but of more in acute myeloid leukaemia ( et al, 1997 ). with intensive have an chance of second the of relapse, all CNS-directed treatment with intensive intrathecal the of additional high-dose therapy cranial irradiation are not but these are in therapy is for children with overt CNS relapse, with or without marrow but be delayed further therapy since the risk to these children is a marrow relapse. There is evidence to the of cranial irradiation with intensive intrathecal chemotherapy than thus and the late of irradiation. A with testicular is for children with overt testicular relapse. be for 18 months to at least 2 years, and of treatment is therapy. This conventional to treatment will the of patients with late testicular relapses and many children with late CNS relapse, and will marrow remissions in children who have had first Most children who relapse a second time are but with second remissions or a second relapse which is extramedullary ( et al, 1997 ), may be by a in remission ( et al, 1997 ). There have been recent to for the of BMT in childhood leukaemia ( et al, 1996 ; et al, 1996 ; et al, 1997 ). In the of randomized trials the evidence from studies that following bone marrow relapse of leukaemia high-dose chemotherapy and of bone marrow from a a greater chance of survival than conventional This is in children with both early and late relapses, in the author's the of BMT for very early marrow relapses within 2 years from diagnosis is There is evidence for for the of a clinical The choice BMT and chemotherapy for children with late relapses is a about risk of disease and late of on the one and the risk of treatment on the The availability of to this choice and is in early marrow It may a greater than of a is no more than that from a as claimed by it be on an with BMT for later The to about treatment is a randomized and a case could be for a such a in children with late marrow relapses with a randomization to chemotherapy or BMT from a or with the of in remission for those who patients and not all have on the management of and the of such a in this of of relapsed ALL much heartache and for patients and the risk of and late of therapy, and consumes large amounts of resources. are both to relapse by first-line treatment and to a better of the factors influencing the chance of second There is a for trials of relapsed ALL and for of early, intermediate and late relapse. This will the and which are to the to review of results of treatment not suggest that one particular is to the There is for further of or to therapy. randomized trials of bone marrow may be to there be in uniform in randomized trials of and of and are children with lymphoblastic leukaemia are to be an second chance of The is by the Research and at the in for with the literature and and for their