Second malignancy after treatment of adult acute myeloid leukemia: cohort study on adult patients enrolled in the GIMEMA trials


The risk of developing secondary tumors after treatment of hematological malignancies is well known and is especially relevant for diseases where chemo- and/or radiotherapy often warrant a long survival.1 In acute myeloid leukemia (AML) this is a rare event; however, a recent study demonstrated that in children treated for AML, surviving at long term, there is a 10-fold increased risk of development of cancer, indicating that AML treatment may enhance the risk of second malignancies (SMs).2 On the other hand, the Copenhagen Hematology group, analyzing 174 AML adult patients enrolled in anthracyclines-based trials, demonstrated that standardized incidence ratio (SIR) was not significantly influenced by intensive chemotherapy.3

The higher number of long-term adult AML survivors, due to treatment improvement, underlines the importance of the survey of SMs and the eventual correlation to specific agents, which are the objects of the present study.

The study population includes 3484 patients, aged 15–76 years, with newly diagnosed AML and acute promyelocytic leukemia (APL), registered between February 1982 and July 1999 in 13 consecutive trials for previously untreated AML or APL. Patients were treated in 62 Hematology Divisions in tertiary care or University Hospitals, members of the GIMEMA cooperative group.

Patients with SM were identified from the GIMEMA registry of patients entered in the following protocols:

  • AML trials: LANL 8201, LANL 8202, LANL 0388, EORTC-GIMEMA AML 8A and 8B, LANL 0491, LANL 93 and AML10p (pilot studies for EORTC-GIMEMA AML 10), EORTC-GIMEMA AML 10 and LAM 0594.

  • APL trials: LAP 0383, 0389 and 0493.

All GIMEMA centers were asked to report the outcome of patients surviving more than 3 months from the time of diagnosis. The diagnosis of an SM was ascertained by the clinicians who attended the patients during all treatment phases and in the following follow-up and reported on a standard questionnaire. The form included statistical demographic data, type and date of onset of SM, latency between AML and SM, treatment and outcome of SM. Furthermore, data on laboratory characteristics at the time of AML diagnosis were available. Data on response to induction treatment and complete remission (CR) duration were obtained from the GIMEMA database. Patients resistant or dying during induction treatment were excluded from the study. Patients dying in CR for non-neoplastic causes were censored from the analysis at the time of death. Last follow-up day is December 2001.

Patients were divided in: AML patients younger that 60 years (group 1); AML patients older than 60 years (group 2); all APL patients (group 3).

Survival and cumulative incidence of SM were studied using the Kaplan–Meier method. Differences among the three groups were evaluated using the log-rank test. The risk of SM in comparison to the general population is expressed as SIR. The SIR was calculated as the ratio between observed and expected number of cases. The expected number of cases is based on age- and sex-specific incidence rates provided by Italian Cancer Registries, for the same period on study.4 SIR and 95% confidence intervals (CIs) were calculated assuming a Poisson distribution of cases. Recurrence of AML or APL was not considered as an SM.

The study population included 3484 patients: 2603 patients achieved CR following induction chemotherapy (75%), while 575 were resistant to treatment (16%) and 306 died during induction (9%).

When grouping AML according to FAB classification, excluding APL enrolled in different trials, 3% of patients had an M0 AML; 16% M1; 34% M2; 27% M4 (2% M4 eo); 16% M5; 3% M6 and 1% M7.

Of 2603 patients achieving CR and at risk of an SM, 1618 had AML (783 aged <60 years old and 835 aged >60 years) and 985 had APL. The median age for AML patients was 50 years (range 14–76 years), with 783 patients younger than 60 years (median age 38 years, range 15–59 years) and 835 older than 60 years (median age 64 years, range 60–76 years). The median age of APL patients was 42 years (range 14–76 years). Males and females were equally distributed in each group (48 vs 52%, age <60 years: 48 vs 52%, age >60 years 48 vs 52%; APL 48 vs 52%). In all, 19 patients (0.7%) developed an SM. Table 1 reports patients' characteristics and type of SM.

Table 1 Characteristics of 15 AML and four APL patients enrolled in GIMEMA trials who develop an SM

All patients, except for three patients enrolled in the GIMEMA 8202 trial, received induction treatment including anthracyclines (idarubicin or daunorubicin or mitoxantrone) at different doses, while nine patients, enrolled in the EORTC-GIMEMA AML 8A and 8B and EORTC-GIMEMA AML 10 trials, additionally received tenoposide.

The median latency between acute leukemia and SM was 31.8 months (range 2.8–100 months) with significant differences within the three groups (P<0.022). The median latency for AML patients under 60 and over 60 years was 50.7 months (range 2.8–87.8 months) and 31.8 months (12.3–70.1 months), respectively. The median latency for APL patients was 6.6 months (3.8–7.7 months). The most frequent SM were lung and breast cancer (four cases each), followed by cancer of the bowel (three cases), bladder, stomach and melanoma (two each), kidney and ovary (one each).

The incidence of SM in the study population was 2.9 per 10 000 person-year. The 19 SM observed in our AML patients (SIR=0.69, 95% CI: 0.41–1.07) translates into a lower incidence than that estimated by the combined Italian Cancer Registries in the general AML population (27.73 patients).

When stratifying patients according to age and analyzing patients with APL separately, the results were quite different (Table 2). The incidence was lower than expected for patients older than 60 years and for APL patients, while in patients younger than 60 years the number of SM was three-fold higher than expected was observed (SIR=3.63, 95% CI: 1.56–7.15). These results were also confirmed when considering patients in continuous CR at 3 and 5 years (SIR 5.62 and 9.38, at 3 and 5 years, respectively). No differences were found in APL patients grouped according to age.

Table 2 Cases of SM observed in GIMEMA acute leukemia patients stratified for age. Comparison with SM observed in normal population

The estimated cumulative incidence (ECI) of an SM over 5 years was similar for SM following AML, independent of age (between 3.8 and 4.6%), while it was lower than expected for patients with a previous APL (0.44%). This incidence remarkably increased with longer follow-up in younger AML patients, reaching 13% at 7 years. In older AML patients, it was not possible to calculate the incidence at 5 years, due to the low number of long-term survivors in this group. The ECI did not change in SM patients with previous APL.

When considering hematological parameters at diagnosis of AML, patients developing an SM were not different from those who did not develop an SM. Interestingly, a statistically significant higher number of patients with acute myelomonocytic leukemia (FAB morphology M4) developed an SM (7/440 vs 12/2163; OR 2.90, CI: 95% 1.03–7.95; P=0.03). This translated into a 1.44-fold increased risk in M4 patients (stratified for age), compared to the normal population (seven cases observed vs 4.87 expected; SIR 1.44, CI: 0.58–2.96). We found no differences in the incidence of SM between patients who did or did not receive tenoposide as part of an induction treatment. No differences in treatment schedules between patients who did and did not develop an SM were observed. Nine patients died due to progression of the SM, and the mortality rate for SM was 47%; four patients died of leukemia relapse (21%). Six patients (32%) are alive in CR of both malignancies.

In this prospective study, we confirm the absence of an increased risk of SM in adult long-term AML survivors. However, when limiting the analysis to patients aged less than 60 years, a significantly higher number of SM was observed. This could be explained by the aggressive and prolonged treatments received by these patients and by the higher cure rate compared to older patients, which increases the number of patients at risk.

The risk of SM after APL treatment was not increased, and the latency between acute leukemia and SM was 6 months. The brief latency between the two malignancies may indicate that the SM is not related to the drugs employed in the APL treatment, but probably to a chance association. The different features of SM after APL compared to other AML types, reflects, as already suggested by others, the different biologic etiology of APL.5

Although patients received during AML treatment a combination of potentially tumorigenic drugs (ie topoisomerase II inhibitors and/or alkylating agents), the risk of an SM is not generally increased, with the exception of patients with AML-M4 and of those aged less than 60 years. We did not find any association between specific chemotherapy agents or class of agents and SM, perhaps due to the small number of patients with an SM and the contemporary use of multiple drugs during induction treatment.

The onset of a secondary acute leukemia due to the stem cell damage induced by the association of multiple potentially leukemogenic drugs used for the treatment of the primary tumor cannot be advocated as cause of an SM after acute leukemia. In fact, in our series the majority of patients received either anthracyclines or tenoposide or both. One possible explanation is that, in some genetically predisposed patients, immunosuppressive therapies for acute leukemia could induce a failure of immune surveillance or of T-cell regulation and may lead to the overgrowth of cancer cells, as also observed during immunosuppressive treatment after bone marrow transplantation.6 The increase of SM in patients with myelomonocytic leukemia is difficult to explain. It could be due to the reduction of effective monocytes involved in immune surveillance, due to leukemic infiltration or to chemotherapy. However, the causality, in our opinion, cannot be excluded.

In conclusion, the cumulative risk of SM in adult patients cured for AML or APL seems low, and of little clinical relevance. However, young patients and particularly those with an acute myelomonocytic leukemia must be followed for several years following AML treatment, in consideration of their higher SM risk.


  1. 1

    Leone G, Mele L, Pulsoni A, Equitani F, Pagano L . The incidence of secondary leukemias. Haematologica 1999; 84: 937–945.

  2. 2

    Leung W, Ribeiro RC, Hudson M Hudson M, Tong X, Srivastava DK, Rubnitz JE et al. Second malignancy after treatment of childhood acute myeloid leukemia. Leukemia 2001; 15: 41–45.

  3. 3

    de Nully Brown P, Hoffmann T, Hansen OP Hansen OP, Boesen AM, Gronbaek K, Hippe E et al. Long-term survival and development of secondary malignancies in patients with acute myeloid leukemia treated with aclarubicin or daunorubicin plus cytosine arabinoside followed by intensive consolidation chemotherapy in a Danish national phase III trial. Danish Society of Haematology Study Group on AML. Leukemia 1997; 11: 37–41.

  4. 4

    Zanetti R, Crosignani P, Rosso S In: Cancer in Italy: Incidence Data from Cancer Registries, Lega Italiana per la Lotta contro i Tumori. Associazione Italiana dei Registri Tumori, Vol II. Torino: Il Pensiero Scientifico eds, 1997.

  5. 5

    Mann KK, Shao W, Miller Jr WH . The biology of acute promyelocytic leukemia. Curr Oncol Rep 2001; 3: 209–216.

  6. 6

    Witherspoon RP, Fisher LD, Schoch G, Martin P, Sullivan KM, Sanders J et al. Secondary cancers after bone marrow transplantation for leukemia or aplastic anemia. N Engl J Med 1989; 321: 784–789.

Download references


This work was supported in part by a grant from Ministry of University and Technological Research (MURST) of Italy.

Author information

Correspondence to L Pagano.

Rights and permissions

Reprints and Permissions

About this article