Therapy-related acute lymphoblastic leukemia is more frequent than previously recognized and has a poor prognosis




Acute lymphoblastic leukemia (ALL) occurring in patients with a history of prior chemotherapy/radiotherapy exposure has been previously reported to be rare, accounting for <2.5% of ALL cases.


All cases of adult ALL with a history of prior cytotoxic or radiation therapy at a leukemia referral center over a 13-year period were analyzed.


Twenty-three cases, representing 6.9% of all ALL cases, were identified. Of these, 17 (74%) had at least 1 high-risk feature; 8 (35%) had MLL rearrangements, and 4 were BCR-ABL+. MLL rearrangements were correlated with CD15 expression and absence of CD10, and also tended to have a shorter mean latency period and more prior topoisomerase II exposure. Twenty-one patients received induction therapy, and 18 (86%) achieved a complete response, 17 with 1 induction. Six patients have relapsed and died, and 4 others died of other complications, 2 of these postallogeneic stem cell transplantation. Median disease-free survival (DFS) and overall survival (OS) were 27 and 13.6 months, respectively, and 3-year DFS and OS were 37.1% and 37.6%, respectively.


The frequency of therapy-related ALL is higher than previously reported and has a poor prognosis, probably related to the high frequency of adverse risk features. Cancer 2012. © 2011 American Cancer Society.


Secondary leukemias are a well-known complication of chemotherapy (CT) and/or radiotherapy (RT) for previous malignancies,1, 2 including breast cancer,3 ovarian cancer,4 and Hodgkin and non-Hodgkin lymphoma.2, 5 Most secondary leukemias are myeloid, which represent 5% to 10% of all acute myeloid leukemia (AML).1, 2 In contrast, the occurrence of acute lymphoblastic leukemia (ALL) after CT/RT is rare; previous reports indicate that these represent 1.2% to 2.5% of adult ALL.6, 7

Therapy-related acute leukemia is divided into 2 major types. The first, alkylating agent/RT-related leukemia, has a mean latency period of 5 to 7 years, is often preceded by myelodysplasia, and is frequently associated with complete or partial deletion of chromosome 5 or 7, and complex karyotypes.1, 2 The second, occurring in patients treated with DNA topoisomerase II inhibitors, typically has a relatively short latency period, between 1 to 3 years, and is not preceded by a myelodysplastic phase. It is usually associated with chromosomal translocation involving 11q23, the MLL gene locus; however, cases without 11q23 abnormality have been rarely described.8 Because we have noted several recent cases of therapy-related ALL at our center, we analyzed our cases in greater detail, to ascertain the incidence, presenting features, and response to therapy.


A total of 335 new cases of ALL were diagnosed at Princess Margaret Hospital between 1998 and April 2011. Of these, 23 (6.9%) were identified from the leukemia database as having received prior CT or RT for another malignancy or other disorder. Ethics approval for this retrospective analysis was obtained from the local institutional research board.

The diagnosis of ALL was based on standard morphology and immunophenotyping according to the current World Health Organization criteria. Immunophenotyping of bone marrow or circulating blasts was performed by flow cytometry as previously described,9 using a panel of B, T, and myeloid markers. Antigens expressed by >20% of the blasts were considered positive.

BCR-ABL testing was performed by quantitative reverse transcriptase polymerase chain reaction as previously described10, 11; BCR-ABL–negative cases then underwent full karyotype analysis at a 400-band level of resolution. Fluorescent in situ hybridization (FISH) testing for MLL rearrangement was also performed in all except 1 of the BCR-ABL–negative cases. Briefly, fixed cells were analyzed using the MLL dual-color break-apart rearrangement FISH probe set (Abbott Molecular, Des Plaines, Ill). Signals were enumerated in at least 200 nuclei per sample by at least 2 observers, using an epifluorescence microscope with single interference as well as dual and triple band-pass filters. In this assay, the native state appears as fused or adjacent red/green (yellow) signals, whereas a rearrangement with a breakpoint at the 11q23 MLL region produces single red and single green signals. The occurrence of an MLL rearrangement (MLL FISH positive) was scored as positive if the percentage of cells showing split red and green signals and/or single red or green signals was above the 99% confidence interval as calculated using the B-inverse function in Excel (Microsoft, Redmond, Wash) on the scores from a set of 30 normal control blood/bone marrow samples, for each signal configuration. The upper limit of normal was 4% for the 1R1G1F signal configuration. If all break-apart signal configurations were within normal limits, the sample was classified as MLL FISH negative.

Statistical Analysis

Complete response (CR) was defined as a normocellular marrow with <5% blasts, absolute neutrophil count >1 × 109/L, and platelet count >100 × 109/L. The overall survival (OS) was determined from the date of diagnosis of ALL to the date of death or last follow-up. The disease-free survival (DFS) of the CR patients was defined as the time from documentation of CR to the date of death, relapse, or last follow-up. OS and DFS probabilities were calculated using the Kaplan-Meier method, and 95% confidence intervals were calculated using the log-transformation. Fisher exact tests were used to determine whether MLL rearrangements were associated with topoisomerase II exposure, or CD10 or CD15 expression. To evaluate whether the median latency period differed between patients, a Mann-Whitney-Wilcoxon test was used. A P value of .05 was used to assess significance for all statistical tests.


Baseline Data

The baseline data for the 23 patients are summarized in Table 1. The individual patient details are shown in Table 2. The most common prior malignancy was breast cancer, with 9 cases (39%). Therapy of the primary disease consisted of CT alone in 6 (26%), RT alone in 8 (35%), and both in 9 (39%). CT regimens consisted of a combination of a topoisomerase II inhibitor and alkylator in 11 cases, topoisomerase II inhibitor with no alkylator or RT in 3 cases, and alkylator alone in 1 case (busulfan and cyclophosphamide conditioning with allogeneic stem cell transplantation [alloSCT]).

Table 1. Baseline Patient Data (n=23)
CharacteristicValue (Range)
  • Abbreviations: F, female; M, male; Neg, negative; Pos, positive; WBC, white blood cell count.

  • a

    Duration from end of treatment of prior malignancy to diagnosis of acute lymphoblastic leukemia.

  • b

    Based on at least 1 of the following: WBC >30 (B) or >100 (T) × 109/L, BCR-ABL+, MLL+, or adverse risk karyotype.

Age, median y51 (17-75)
Sex, M:F8:15
Median latency period, moa48 (5-360)
High riskb17
WBC, median, ×109/L18.2 (1-503)
 No. >30 × 109/L9
 Not done/unsuccessful5
Table 2. Patient Details
Patient #/Age, yPrimary TumorPrior TherapyLatency, moCytogeneticsMolecularCD10/CD15Outcome (Months in CCR)
  1. Abbreviations: ABVD, doxorubicin, bleomycin, vinblastine, dacarbazine; AML, acute myeloid leukemia; ASCT, autologous stem cell transplantation; BuCy, busulfan, cyclophosphamide; CA, cyclophosphamide, doxorubicin; CCR, continuous complete remission; CEF, cyclophosphamide, epirubicin, fluorouracil; CR, complete response; CVP, cyclophosphamide, vincristine, prednisone; DNR, daunorubicin; EG, eosinophilic granuloma; HL, hypereosinophilic leukemia; alloSCT, allogeneic stem cell transplantation; ND, not determined; NR, not reported; RCHOP, rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone; RT, radiotherapy.

1/52EGRT360Failed−/−−/−CCR (43+)
2/69BreastRT120NDBCR-ABL+−/+CCR (43+)
3/59BreastRT1801(1p)−/−+/−Indeterminate death
4/71ProstateRT60Near triploidy−/1+/−CR, relapse
5/42ThyroidRT156NDBCR-ABL+−/+CR, relapse
6/56BreastRT48NDBCR-ABL+−/+CR, relapse
7/69ThyroidRT48NDBCR-ABL+−/+CR, relapse
8/55OvaryRT144NDMLL+−/+No induction
9/18HLBuCy + alloSCT2446,X,Add(Y),−2, +3 (q21),−4,−7, +7(q22), +2markers−/−+/−NR
10/44ThymomaVP16/cisplatin5Failed−/−−/−CR, relapse
11/24AML-M4DNR/Ara-C746,XY,del(6)(q21q23),t(11;13)(p11.2;q12)−/−+/+CR, relapse
12/75AMLDNR/Ara-C/mitoxantrone/VP162846,XY,t(17;18) (q23;q21)−/−+/−CCR (19+)
13/51HodgkinABVD1846,XX,t (1;11)MLL+−/−CCR (60+)
15/65LymphomaRCHOP/RT7046,XY−/−+/−No induction
16/45BreastCA/RT180Near triploidy−/−+/−CR, relapse
17/17NeuroblastomaCA/ifosfamide/VP-16/ carboplatin/RT82t(9;22) (q34;q11.2)BCR-ABL+−/+CCR(33+)
18/41BreastCEF/RT2045,XX,der(13;14)(q10;q10),t(1; 11)(p32;q23)MLL+−/+CCR (132+)
19/25BreastCEF/RT9NDMLL+−/+CCR (6+)
20/33LymphomaCVP/RT, RCHOP/ASCT7246,XY,inv(11)MLL+−/+CR, died post-HSCT
21/39BreastCEF/RT1646,XX−/−−/+CR, died recurrent breast CA
22/42BreastCEF/RT1246,XX,t(4;11)(q21;q23)MLL+−/+CR, died sepsis
23/67BreastCA/RT3646,XX,t(4;11)(q21;q23)MLL+−/+Indeterminate death

Of the 8 patients with MLL rearrangements, cytogenetics were available in 6 cases; 3 had t(4;11), 2 had t(1;11), and 1 had inv(11). Of the 10 patients who had neither BCR-ABL nor MLL rearrangements, 3 had normal karyotypes, 2 had near triploidy, 1 had 2 copies of a structurally abnormal 1p, 1 had t(17;18), and 1 had both del(6)(q21q23) and t(11;13)(p11.2;q12); the remaining 2 cases failed to produce dividing cells.

As shown in Table 3, patients with MLL rearrangements had a trend for a shorter median latency period compared with the other patients, although this difference was not statistically significant. MLL rearrangements were more likely to have had prior topoisomerase II exposure (although this trend did not reach statistical significance), and were also significantly correlated with the absence of CD10 and presence of CD15 expression (Table 3). There was no correlation between MLL rearrangements and CD34 expression or initial white blood cell count (WBC; data not shown).

Table 3. Correlation Between MLL Rearrangement (MLL+) and Baseline Features
FeatureMLL+, n=8MLL, n=15P
Topoisomerase II exposure77.086
No topoisomerase II exposure18 
Median latency period, mo (range)19 (8-144)60 (5-360).146

Of the 8 patients who had received RT alone, 4 patients were BCR-ABL positive (Table 2).

Two of 6 patient who had received CT alone for their primary malignancy had MLL rearrangements, as compared with 1 of 8 who had received RT alone (P = not significant). The patients who received CT alone had a shorter median latency period (13 months; range, 5-28 months) as compared with those who received RT alone (132 months; range, 48-360 months; P = .0024).

Response to ALL Therapy

Two patients died before induction therapy could be instituted—1 presented with progressive pneumonia and acute renal failure, and rapidly deteriorated; the second, presenting with WBC >500 × 10/L, died of an intracranial hemorrhage. The remaining 21 patients all received ALL induction therapy and are evaluable for response. Of these, 11 patients received the pediatric-based Dana-Farber Cancer Institute (DFCI) protocol,12 5 received a modified version of the DFCI protocol for patients older than 60 years, and the remaining 5 received a variety of other ALL protocols (1 each). The 4 BCR-ABL–positive patients received imatinib in combination with the DFCI protocol. Of these 21, 17 (81%) achieved a CR with 1 induction; 1 additional patient, treated with the DFCI protocol, had persistent disease but achieved CR with a second induction using hyperCVAD.13 Of the remaining 3, 2 died of septic complications during induction, and 1 was not able to achieve CR despite 3 induction attempts and subsequently died.

Of the 18 patients who achieved CR, 6 have relapsed and died of progressive leukemia. Two other patients died of complications after alloSCT (see below), 1 died of Candida septicemia on CT, and 1 died of recurrent metastatic breast cancer (her original malignancy). The remaining 8 patients remain alive and in continuous CR at a median follow-up of 37.3 months (range, 1-131 months). These include 1 BCR-ABL+ patient post-alloSCT and 3 with MLL rearrangements (1 post-alloSCT). The OS of the patients is shown in Figure 1. The median OS was 13.6 months, and 3-year OS was 37.6% (95% confidence interval [CI], 21.4-66.0). The median DFS of the CR patients was 27 months, and 3-year DFS was 37.1% (95% CI, 19.3-71.3).

Figure 1.

Overall survival for all patients from acute lymphoblastic leukemia (ALL) diagnosis, and disease-free survival for patients who achieved a complete response (CR) with ALL induction therapy (shown from time of CR) are shown.

A total of 5 patients have undergone myeloablative alloSCT in CR-1, 4 related and 1 unrelated. Two of these were BCR-ABL+ and 2 MLL+. Of these 5, 1 died of complications related to graft-versus-host disease (GVHD). A second relapsed, then was salvaged with a second transplant, but died 8 years later because of complications of chronic pulmonary GVHD, and a third relapsed and died. The remaining 2 patients are in continuous CR at 2 and 10 years post-SCT, respectively.


The frequency of therapy-related ALL in this retrospective series (6.9% of all ALL cases) is higher than has previously been reported. Pagano et al, in a retrospective analysis of 901 cases of ALL, identified 21 patients with a prior malignancy; of these, only 11 had received CT/RT, for an overall incidence of 1.2%.6 Ishizawa et al reported 6 of 152 ALL cases with a prior malignancy, of whom 4 had received CT/RT, for an overall frequency of 2.6%.7 This suggests that therapy-related ALL may be occurring more frequently in recent years than previously seen or recognized. This may be related to the increasing use of topoisomerase II-containing CT regimens for other malignancies, particularly breast cancer; nearly 40% of our cases had treatment for prior breast cancer, most with CT or combined CT and RT, and 39% of patients had MLL rearrangements. Two recent reviews of the literature suggests that ALL with 11q23 abnormalities are more frequently seen in patients treated for prior breast cancer than for other malignancies.7, 14

MLL rearrangements involving 11q23 have been described in both AML and ALL, representing approximately 5% of therapy-related acute leukemias. However, these are relatively more frequent in ALL; nearly 1/2 of therapy-related acute leukemia with 11q23 rearrangements were ALL in 1 series.7 The high frequency of MLL rearrangements in our series is consistent with this. These rearrangements were associated with a shorter mean latency period (Table 3), as previously described.15 Only 1 MLL case had a latency period of >6 years (144 months); this was the only MLL case that had not previously been exposed to topoisomerase II inhibitor therapy, and therefore may have had a different pathogenesis.

The presence of an MLL rearrangement was more frequently associated with the absence of CD10 expression, as previously described.6, 7 Therefore, CD10 negativity, particularly in the setting of prior exposure to a topoisomerase II inhibitor, is strongly suggestive of the presence of an MLL rearrangement. Conversely, this abnormality is very unlikely in the presence of CD10 expression; there were no such cases in our series. The association between MLL and CD15 expression has also been previously described.7

Eight patients on our cohort had received RT but no CT as treatment for their prior malignancy. The occurrence of ALL after RT has been previously described; Pagano et al found 6 such cases,6 3 of which had the Philadelphia (Ph) chromosome. Our cases were associated with a variety of cytogenetic and molecular abnormalities; only 1 had a del(7) typically described for such secondary leukemias, 1 had an 11q23 abnormality, and 4 (50%) had BCR-ABL rearrangements. These 2 series suggest the possibility that Ph+ ALL may be a rare consequence of RT, although we cannot exclude the possibility of a chance association or a genetic predisposition in these patients. B-ALL has been well described in patients with previous malignancies who have only been treated surgically,6 bringing up the possibility of any underlying predisposition to cancers in some cases.

The marked difference in latency period between patients who received RT alone versus CT alone has not, to our knowledge, been previously reported. It did not correlate with the presence of MLL rearrangements; only 2 of 6 CT-treated patients had this. However, 5 of 6 of the CT-treated patients had received topoisomerase II-containing regimens, suggesting that exposure to these agents may predispose to the early development of ALL even in the absence of MLL rearrangements.

The prognosis of therapy-related ALL appears to be poor, based on our series and those previously reported. Our CR rates are higher than previously reported for secondary ALL,14 but similar to what has been described in age-equivalent series of adult patients with ALL.7, 8, 12, 13 Despite this, relapse rates are relatively high and corresponding survival is low. This is likely related to the high frequency of adverse cytogenetic and/or molecular features, including BCR-ABL positivity, MLL rearrangements, near triploidy, and high initial WBC. In our series, based on these criteria, 17 cases (74%) would fall into a high-risk category. Therefore, the poor OS is not surprising. Because of the small number of patients who did not fall into a high-risk category, it is unclear if prior malignancy itself represents an independent prognostic factor. Furthermore, because of the heterogeneity of the prognostic features in our poor prognosis patients, we were unable to properly compare this with an equivalent group of poor prognosis de novo ALL patients to ascertain whether therapy-related high-risk patients were doing worse.

The poor prognosis suggests that patients with therapy-related ALL should be considered for alloSCT in first CR. Unfortunately, the small number of patients who underwent this procedure in our series precludes evaluation of the efficacy of this treatment modality. However, there are data suggesting that alloSCT may be beneficial for therapy-related acute leukemias3 as well as for ALL patients with MLL rearrangements.16


No specific funding was disclosed.


The authors made no disclosures.