Late Relapsing Childhood Lymphoblastic Leukemia

EVEN AFTER TREATMENT with the best current therapy, approximately one in four children with lymphoblastic leukemia (ALL) relapse.1 The risk of relapse is mainly during the period immediately after stopping treatment and diminishes with time so that patients in complete remission (CR) for 10 years are considered to be effectively cured of their disease.2Relapse after that time period has been reported,3 but until a few years ago, it was difficult to be certain whether these patients had a true recurrence of the original clone or a second or secondary leukemia.

Confirmation of a true relapse can now be obtained by documenting the presence of an identical clone-specific molecular signature in lymphoblasts at diagnosis and recurrence. Immunoglobulin (IgH) and T-cell receptor (TCR) gene loci consist of dispersed gene segments, which undergo somatic recombination early in B- and T-cell ontogeny to produce functional Ig or TCR molecules. Clonal IgH or TCR rearrangements are detectable in almost every case of ALL. Each rearranged gene is unique and therefore provides a good clonal marker to monitor minimal residual disease in patients in apparent remission.

Such molecular techniques have been used by several investigators, including ourselves, to confirm relapse in cases of late recurrence.4-6 These cases not only raise fundamental questions about the biology of childhood ALL, they also generate anxiety about the resilience of cure in patients whose leukemia has been in long-term remission. To answer the latter question in particular, we studied long-term follow-up data on a large cohort of children treated on the MRC national ALL trials. We tried to estimate the frequency of very late relapse, what type of leukemia gives rise to it, and what type of patient is at risk.

Since the late 1960s, an increasing proportion, now over 90%, of children diagnosed with ALL in the United Kingdom have been treated on MRC childhood leukemia studies. Long-term follow-up of these patients is undertaken by individual treatment centers and reported every 2 years to the Clinical Trial Service Unit (CTSU) in Oxford. Data are available for nearly all patients treated between 1970 and 1984 with less than 1% lost to follow-up at 10 years. Median follow-up of surviving patients is 18 years. Information on the number of patients surviving for 10 years in first remission and those who subsequently relapsed was extracted from the CTSU database. Where detailed clinical treatment or outcome information on patients with late relapse was unavailable in the database, individual centers were contacted to obtain this information. The actuarial risk of relapse after 10 years in first complete remission (CR1) was calculated using Kaplan-Meier life table analysis.

Central morphology review and archiving of marrow smears is undertaken on all cases entered into MRC childhood leukemia studies. Diagnosis and relapse marrow smears were retrieved from the central archive or were kindly provided by individual centers from their local archives where these were unavailable in the central archives.

DNA was extracted from marrow smears archived at diagnosis and relapse using a previously described technique.7 Briefly, material was scraped from the slide using a sterile blade and DNA was extracted by proteinase K digestion, phenol/chlorofom extraction, and ethanol precipitation. Polymerase chain reaction (PCR) amplification of the IgH gene (Frameworks 1, 2, and 3) or TCR gene (Vδ2-Dδ3) were performed on 1 μg of patient DNA. The three IgH PCR reactions used a 3′ consensus primer to the joining region of the IgH gene and a consensus 5′ primer to the appropriate framework region of the variable gene.8,9 The TCR δ region PCR used a 5′ Vδ2 primer and a 3′ Dδ3 primer.10 PCR products were visualized using polyacrylamide gel electrophoresis and ethidium bromide staining. Clonal products were sequenced with forward and reverse primers using an ABI 373 automated sequencer (Applied Biosystems, Foster City, CA).

A total of 2,746 children were treated on Medical Research Council (MRC) ALL trials between 1970 and 1984. Details of study treatment protocols and patient outcome have been published elsewhere.11-15 A total of 1,134 children survived for 10 years in CR1, of whom 12 subsequently relapsed giving an actuarial risk of 1% over the next 10 years.

There were equal numbers of males and females within the 12 relapse cases (Table 1). Based on their age and white blood cell count at presentation, all had standard risk disease according to current risk stratification criteria for childhood ALL.16 Although the diagnosis of ALL was based primarily on cytomorphology and cytochemistry, it was further confirmed in six patients by documenting CD10 expression (common ALL antigen) by the lymphoblasts. Treatment varied according to the then current study protocol, but most patients did not receive any postinduction intensive consolidation therapy,17 and all could be regarded as having received inadequate treatment by current standards.1 

Although relapse was diagnosed after 10 to 15 years in CR1 in 10 children, it did not occur until after 20 years in two patients (Table 2). All relapses were in the marrow except one, a patient who had a pelvic lymphoma with no morphological evidence of marrow disease. Blast cells were of precursor B lymphoid origin in all cases. Four of seven patients in whom a blast cell karyotype analysis was successful had a clonal abnormality. One of these was a Philadelphia translocation, which had, in retrospect, probably been present at diagnosis.

All patients were successfully reinduced into a second complete remission and the majority received intensive and continuing chemotherapy to consolidate and maintain the remission without a marrow transplant. Of the 12 patients, eight remain in a second complete remission at a median follow-up of 52 months and the actuarial disease-free survival 5 years after relapse is 56%. Of the four patients who relapsed a second time, two had received an autologous transplant and two chemotherapy alone.

Amplifiable DNA could be extracted from both diagnosis and relapse marrow smears in 8 of 12 cases studied. Polyacrylamide gel electrophoresis of the amplified DNA showed an identical size clonal TCR (1 case) or IgH (7 cases) gene rearrangement at diagnosis and relapse in these 8 cases (Tables 1 and 2, Fig 1, patient 1). Automated sequence analysis confirmed complete sequency homology of rearrangement at diagnosis and relapse in seven patients. The remaining patient has been reported previously5 and was shown to have an identical FR3 IgH rearrangement by dot hybridization using a clone-specific oligonucleotide probe at diagnosis, end of treatment, and relapse. In two cases, a clonal IgH rearrangement could be detected at relapse (Table 2, patients 11 and 12), but we failed to extract DNA from the diagnostic smear.

With the large number of children treated on MRC ALL studies between 1970 and 1984, almost all of whom were rigorously followed for more than 18 years, we can be confident that the 1% risk of late relapse uncovered in this study is representative. Therefore, physicians treating children with ALL can be reassured that the vast majority of long-term remitters are truly cured of their disease. Whether even this small risk is faced by patients treated on later protocols is uncertain, as all of the patients reported received less intensive treatment than is used in current protocols.1 17 

The evidence provided in this study that relapse can occur after 10 years in remission raises some fundamental questions about the biology of childhood ALL. Like the requirement for a prolonged “maintenance” treatment program, late relapse seems to be unique to this cancer. Together with recent evidence that PCR techniques, similar to those used in this study, can detect residual leukemic cells in a substantial proportion of patients in long-term remission,18 our findings suggest that cells from the original clone can remain dormant for long periods of time. Although the mechanisms responsible for this dormancy are not known, they are likely to involve either an ability of the clonal cells to remain out of cell cycle (in G₀) for long periods, or the ability of host immune surveillance to maintain minimal residual leukemia in limbo, or a combination of the two. Clinical recurrence would only arise when there was either a breach in immune surveillance or a return to cell cycle triggered by the acquisition of new mutations within the dormant cell genome allowing clonal escape. The absence of subclones with mutated IgH or TCR genes at late relapse (there was complete sequency homology between the clonal rearrangements found at diagnosis and relapse) in the patients reported suggests the former to be a more likely explanation. However, we do not have sufficient information to exclude immunophenotypic or karyotypic clonal evolution at relapse, so this hypothesis remains unsubstantiated.

The relative ease with which the leukemia could be reinduced into an easily maintained second remission suggests that the lymphoblasts retain chemosensitivity. If so, there is every reason to suppose that such patients can be salvaged simply with more intensive conventional chemotherapy than that which they received on the first occasion and without recourse to progenitor cell transplantation.

We are grateful to all previous and current members of the MRC Childhood Leukaemia Working Party and the following physicians, data and laboratory managers who contributed material for this study: Birmingham Children's Hospital, Professor J.R. Mann and P. Short; Queen Elisabeth Hospital, Birmingham, D. Boffey; Bristol Royal Hospital for Sick Children, Professor A. Oakhill; Addenbrooke's Hospital, Cambridge, Dr V. Broadbent, Dr R. Marcus, and D. Bloxham; Walsgrave Hospital, Coventry, Dr M.J. Strevens; Royal Hospital for Sick Children, Edinburgh, Dr A. Thomas and Dr P Shaw; Milton Keynes General Hospital. Dr S.S. Jolloh and Dr D.J. Moir; and North Staffordshire Hospital, Stoke-on-Trent, Dr R.N. Ibbotson.