New recurring cytogenetic abnormalities and association of blast cell karyotypes with prognosis in childhood T-cell acute lymphoblastic leukemia: a Pediatric Oncology Group report of 343 cases

The clinical utility of cytogenetic evaluation of leukemic blasts is nowhere better exemplified than in childhood acute lymphoblastic leukemia (ALL). The independent prognostic importance of blast cell karyotypes (eg, the good prognosis associated with hyperdiploidy in contrast to the poor prognosis associated with several specific translocations) is so well established that the karyotype has been used during the past decade for the accurate assignment of risk and selection of risk-directed therapy for patients with ALL.1-5 The data, however, that established the blast cell chromosome number as an independent prognostic indicator in childhood ALL did not distinguish between B-cell and T-cell lineage ALL (B-lineage ALL and T-ALL, respectively). The most frequently occurring clinically important translocations in ALL, such as t(9;22), t(1;19), and t(4;11), are associated primarily with B-lineage ALL.

T-ALL comprises 12%-15% of childhood ALL. T-ALL has clinical, biochemical, immunologic, and chromosomal features that are distinct from those of B-lineage ALL.6-10 Because most children with ALL have been stratified and treated without regard to immunophenotype, evaluation of the prognostic importance of clinical or laboratory variables in T-ALL has been difficult. To address this problem, the Pediatric Oncology Group (POG) adopted the strategy of treating children with T-ALL with a common, aggressive protocol, as described elsewhere,11 that included cytogenetic evaluation of blast cells at diagnosis. The blast cell karyotypes of this large series of childhood T-ALL patients were analyzed to further define the cytogenetic differences between B-lineage ALL and T-ALL and to determine the prognostic value of cytogenetics in childhood T-ALL.

From January 1986 to January 1992, 439 eligible children (ages, at least 1 year to less than 22 years) with newly diagnosed T-ALL were entered in POG study 8691 (the pilot study for study 8704) or 8704. Diagnosis of T-ALL was made according to the morphological and cytochemical criteria of the French-American-British classification and by immunophenotyping. All patients were treated with an intensive 7-drug protocol. The total duration of therapy was 2 years. The patients were randomly assigned to receive weekly doses of L-asparaginase during initial maintenance therapy. The cut-off date for analysis was October 21, 1998.

Bone marrow samples for cytogenetic evaluation were obtained before initiation of therapy and shipped overnight to a single laboratory at the University of Alabama at Birmingham. Routine methods were employed for overnight culture, harvest, slide preparation, and G-banding with trypsin and Wright's stain. Whenever possible, a minimum of 20 metaphase cells were analyzed for each sample. A minimum of 10 metaphase cells were analyzed before reporting a normal karyotype. Chromosome abnormalities were described according to the International System for Human Cytogenetic Nomenclature.12 

This study is mainly descriptive. Event-free survival is defined as the time from registration to first event among failures (failure to achieve remission, relapse, second cancer, or death) or time from registration to last clinical contact (nonfailures). Comparisons of subgroups were done by both univariate methods (log-rank test13) and multivariate methods (Cox proportional hazards model14). Four-year estimates of event-free survival were constructed by the method of Kaplan-Meier,15 using SEs of Peto and coworkers.13 Statistical analyses were not done for clinical correlates because the standards for clinical significance are much greater than for statistical significance in this area.

The age, race, sex ratio, white blood cell (WBC) count, and survival characteristics of groups of patients with successful cytogenetic evaluations are shown in Table1. The median age, racial distribution, and median WBC count at diagnosis for this series of patients are typical for childhood T-ALL and in general do not differ substantially by karyotype.

There were 70 patients with unsuccessful cytogenetic studies who did not differ substantially from those with successful cytogenetic evaluations (data not shown) except that their median WBC count was 145×10⁹/L (145 000/μL) (quartiles, 48 000-435 000). This was more than twice the WBC count of those patients with successful cytogenetic evaluations whose median WBC count was 70×10⁹/L (70 000/μL) (quartiles, 19 000-189 000). The reason for this difference is not apparent.

Table 1 suggests that children whose blast cell karyotype contains any kind of derivative chromosome (ie, an unbalanced translocation or complex rearrangement) may be somewhat older and that t(11;14) may be associated with a younger age at diagnosis. The data also show that the proportion of girls with hyperdiploid karyotypes or with an extra chromosome 8 (+8) is higher than in other karyotype groups, and that the proportion of boys with t(10;14) is higher than in other karyotype groups. A higher median WBC count is associated with t(11;14), and a lower median WBC count is associated with hyperdiploidy of more than 50 chromosomes and a loss of chromosome 5 long arm (5q) material. The number of patients in several of the karyotype categories is relatively small, however, and hence, any associations or lack thereof need to be confirmed by study of additional series of patients.

Of 439 eligible patients with childhood T-ALL registered in 1 of the 2 POG studies, only 15 patients (3.4%) did not have a sample sent for cytogenetic evaluation. Of the 424 patients who had a sample sent, 70 patients (16.5%) had an unsuccessful cytogenetic evaluation. Of the 354 successfully karyotyped patients, only normal karyotypes were found in 153 patients (43.2%). Of these 153 patients, 31 (20%) had less than 20 metaphases (10-19) available for analysis. Of the 201 patients (56.8%) with an abnormal karyotype, 180 patients (89.6%) had at least one structural chromosome abnormality. All of the recurring numerical and structural chromosome abnormalities among the 201 abnormal karyotypes in this series of childhood T-ALL patients are shown in Table 2. The frequencies of some of the cytogenetic abnormalities in this series are compared in Table3 with the frequencies reported in smaller series of karyotyped T-ALL patients.10,16-24 A list of each abnormal karyotype in this series is available on the POG website (http://www.pog.ufl.edu/publ/news/T-ALL_ka.pdf).

Of the 201 patients with abnormal karyotypes, 16 patients (8%) were hypodiploid, 125 patients (62.2%) were pseudodiploid, 46 patients (22.9%) were hyperdiploid with 47-50 chromosomes, and 14 patients (7%) were hyperdiploid with more than 50 chromosomes. All of the hypodiploid cases had 43-45 chromosomes; none had a near-haploid karyotype. Of the 14 cases with more than 50 chromosomes, 10 had near-tetraploid karyotypes. The frequencies of ploidy groups in this series are compared in Table 4 with the frequencies reported in pediatric non–T-ALL and in other series of karyotyped T-ALL patients.

Twenty-eight structural chromosome aberrations occurred more than once in this series. Genes known or likely to be affected by these abnormalities are listed in Table 2. In this series, 60 patients (30% of those with abnormal karyotypes) contained a rearrangement at one or more of the chromosome bands (7p15, 7q32-36, and/or 14q11-13) that are the locations of T-cell receptor chain genes. Rearrangements in the 14q11-13 region, in which the T-cell receptor α/δ chain genes are located, were present in 21% of the karyotypically abnormal cases in this series; these aberrations included not only the relatively common and T-cell–specific translocations t(11;14)(p13;q11) and t(10;14)(q24;q11) but also a variety of other balanced and unbalanced translocations, deletions, and inversions.

Five new recurring abnormalities in this series that have not, to our knowledge, been reported as recurring aberrations in T-ALL are presented in Table 5. Chromosome abnormalities that were observed only once in this series but that have been reported once in other cases of T-ALL and thus can be designated new recurring aberrations in T-ALL are shown in Table6.

The series follow-up was excellent. In the first 5 years, only 14 patients were lost to follow-up (4 patients with an abnormal karyotype, 9 with a normal karyotype, and 1 with an unsatisfactory cytogenetic study). The actuarial 5-year event-free survivals by karyotype group were: 51% of those with an abnormal karyotype (4% SE); 62% of those with a normal karyotype (4% SE); and 60% of those who had unsatisfactory results (6% SE). The Cox analysis, which questioned if, taken together, any of the 11 variables listed in Table 1 were associated with the outcome, resulted in P = .16. Given the inconclusive nature of this overall result, those analyses that are univariately significant in Table 1 should be viewed as promising, but they require confirmation from another series study prior to consideration of any of those associations as definitive.

The present study is not only the largest reported series of karyotyped cases of pediatric T-ALL, but it is also the largest series of childhood T-ALL cases karyotyped in a single laboratory and treated with a single protocol. While this series failed to produce definitive associations between karyotypic features and event-free survival, it uncovered several karyotypic abnormalities needing further investigation concerning such an association. In contrast to pediatric B-lineage ALL, we were unable to demonstrate in childhood T-ALL a significant association between a hyperdiploid karyotype and a favorable outcome, a finding in agreement with that of Heerema et al.10 

Although this series contains a higher percentage of cases with only normal karyotypes than any other series of T-ALL cases except that of Uckun et al,23 inspection of Table 3 shows that the frequency and distribution of the abnormalities found are fairly similar to those in other series of childhood T-ALL cases. This similarity suggests that there were no karyotypic subsets of T-ALL that disproportionately failed to produce abnormal metaphases in the present series. The reason for the higher percentage of only normal karyotypes in this series is unclear. The great majority of samples in this series were sent by overnight express to a central cytogenetics laboratory; the same was true, however, for the majority of samples evaluated by Kaneko et al,20 who reported only 27% of cases with normal karyotypes. Therefore, a delay in specimen processing is unlikely to be the reason. If all 31 patients in this series with a normal karyotype based on fewer than 20 metaphases were assumed to have an undetected cytogenetically abnormal clone, then the proportion of patients in this series with a normal karyotype would be 34.5%, a figure closer to that reported in other studies.

The results of this series confirm that t(11;14)(p13;q11) is the most frequently occurring translocation in childhood T-ALL and are consistent with results of most other studies regarding the frequency of other T-cell receptor chain gene-related aberrations such as t(10;14)(q24;q11) and rearrangements involving 7q32-34 or 7p15. One large study of adult ALL patients16 found that t(10;14), not t(11;14), was the most frequent translocation in adult T-ALL, and that the frequency of 14q11 breakpoints was higher than in childhood ALL. In the series reported by Heerema et al,10 patients with 14q11 breakpoints had a higher WBC count than patients in other karyotype groups; in our series, however, only the t(11;14) patients had a substantially higher median WBC count than other karyotype groups.

The single most common chromosome abnormality in this series, present in nearly 1 case in 5, was del(6q). This abnormality was found in only 10% of a large series of cases of childhood ALL of all types by Hayashi et al.25 The present series, as well as others in Table 3, suggest that del(6q) is more common in T-ALL than in childhood ALL in general, a finding reported by the Groupe Français de Cytogenetique Hematologique (GFCH) in their large series of karyotyped childhood ALL cases, which included 56 T-ALL patients.18In this series, similar to the findings of Heerema et al,10 del(6q) was the sole aberration in 26% of the patients with a del(6q), and no common deleted region was apparent.

The frequency of cases with +8 is virtually the same in this series and that of Heerema et al,10 although the frequency of +8 reported in other series is lower. The results of this series confirm that +8 is more common in T-ALL than in B-lineage ALL.26,27 Of 21 patients with +8 in this series, +8 was the sole chromosome abnormality in 6 patients (4 of whom were reported previously by Pettenati et al26), thereby comprising 3% of all cases with an abnormal karyotype. In 2 cases in this series, +8 was associated with t(11;19)(q23;p13.3) as the only other chromosome abnormality. Trisomy 8 and t(11;19)(q23;p13) also occur together in acute myeloid leukemia, suggesting that these 2 chromosome abnormalities tend to arise in a multipotent cell.28 

A difference between the findings of this study and those of the 4 next-largest studies is that our series has a lesser frequency of cases with del(9p) or i(9q). When cases with a loss of a whole chromosome 9 or with an unbalanced rearrangement resulting in a loss of 9p material are added to cases with del(9p) and i(9q), the proportion of cases with loss of all or part of 9p in this series is 15.4%, which is comparable to the proportion of cases with a loss of 9p that was reported by Heerema et al.10 Murphy et al29found deletions of 9p significantly more often in childhood T-ALL than in non–T-ALL (15.7% vs 6.2%, respectively). The GFCH18found that 9p abnormalities were associated more often with T-ALL than with non–T-ALL, but the difference was not statistically significant. In this series, like that of Heerema et al,10 bands 9p22-24 appeared to be missing in all cases with a loss of the 9p material. The p15 and p16 genes, which are often deleted or inactivated in pediatric T-ALL,30 are on 9p.

The percentage of patients (5.5%) with a del(5q) is comparable to frequencies that have been reported in other series of T-ALL (Table 3). Although del(5q) is common in myeloid neoplasms, it is distinctly uncommon in ALL. In 1992, Theodossiou et al31 reported 3 ALL patients with del(5q) and found only 23 such patients reported in the literature; interestingly, 10 of the 19 phenotyped patients had T-ALL. The following year, the GFCH18 reported aberrations in the 5q31-35 region in 4 of 56 pediatric T-ALL patients, a proportion comparable to the findings in this and other studies of T-ALL patients.

Recurring and/or well characterized chromosome abnormalities in T-ALL that were not seen in this series include t(5;14)(q33-34;q11)32; t(7;9)(q34;q32); t(7;9)(q34;q34.3); t(7;19)(q34;p13)33; t(7;11)(q35;p13)10,34; and inv(14)(q11q32).10,35,36 Some structural aberrations that were observed only once in this series and are known to be rare recurring abnormalities in T-ALL include t(1;7)(p34;q32)10,17,23,33; t(1;11)(p34-36;p11-13)16,17; t(11;14)(p15;q11)10,37,38; inv(5)(p11q13)10,17; and del(17)(p11).10,22 

Five recurring chromosome abnormalities in this series have not been previously reported as recurring aberrations in T-ALL (Table 5), although all 5 have been reported in unphenotyped cases or in other types of ALL, and 3 have been reported in single cases of T-ALL. These abnormalities, which are described further below, are del(1)(p22); dup(2)(q21-32q31-37); t(8;12)(q13;p13); +9; and del(13)(q).

Two patients with ALL have been reported with del¹(p22); one patient had common ALL,1 and the other had an unclear immune phenotype.39 

The report by Murphy et al29 that includes some T-ALL cases describes a case with dup(2q), but the immune phenotype of that case was not reported. Two other cases with dup(2q) have been reported,40,41 but they were without immune phenotypes. One case of T-ALL with dup(2)(q11.2q31) has been reported.10 

Only 1 reported case of t(8;12)(q13;p12-13), a case of common ALL, was found.40 

Several cases of common ALL,23,42-44 mixed-lineage ALL,45 and unphenotyped ALL46,47 with +9 in a karyotype with no more than 51 chromosomes have been reported, but only one of these cases had T-ALL.10 

Cases of common ALL40,48,49 or unphenotyped ALL29,50 with del(13q) have been reported, but only one case was reported with T-ALL10; a molecular deletion in 13q was reported in one T-ALL cell line.49 Del(13q) is common in chronic lymphoid leukemia (CLL),51 and it may be an important step in the development of lymphoid leukemias of various types. That it may be an early or primary cytogenetic event is confirmed by 2 cases in this series with del(13q) as a sole chromosomal abnormality.

Out of 5 new recurring chromosome abnormalities in T-ALL in this series, 4 occurred as a sole abnormality in at least one case, which suggests that the abnormalities may be early or primary events in leukemogenesis. The fact that each has been observed in non–T-ALL indicates that these abnormalities are not specific for T-cell leukemogenesis.

Seven of the single observations in this series have been reported only once previously in T-ALL to our knowledge. Together with those single previous reports, our observation of another case in this series defines those aberrations as additional rare recurring chromosome abnormalities in T-ALL (Table 6).

In all, this study has identified 12 new recurring cytogenetic aberrations in T-ALL. This series confirms that the karyotypic characteristics of childhood T-ALL are clearly different from those of B-lineage childhood ALL, supporting the notion that B-lineage ALL and T-ALL are biologically distinct neoplasms. Not only are the types and frequencies of chromosomal abnormalities very different in B-lineage ALL and T-ALL, but the contrast between the prognostic importance of blast karyotype in childhood B-lineage ALL versus its relative lack of importance in T-ALL is striking. Hyperdiploidy of 51-60 chromosomes, present in the leukemic blasts of about 25% of B-lineage pediatric ALL patients and associated with a good to excellent prognosis, is very uncommon in T-ALL and, when present, is not associated with any survival advantage. The recurring translocations of B-lineage ALL are very rarely seen in childhood T-ALL, whereas the 2 most common translocations in T-ALL, t(11;14) and t(10;14), are virtually never seen in B-lineage ALL. Most recurring translocations in B-lineage ALL are or have been associated with a poor prognosis, whereas t(10;14) in T-ALL may be associated with a higher rate of survival. The longer survival associated with t(10;14) was first reported in adult T-ALL16 and is confirmed (by univariate analysis only) in this large series of pediatric T-ALL.