Clinical and biological features of acute myeloid leukaemia occurring as second malignancy: GIMEMA archive of adult acute leukaemia
Dr Livio Pagano, Department of Haematology, Catholic University, Largo Francesco Vito 1, I-00168 Roma, Italia. E-mail: email@example.com
Between July 1992 and June 1996, 3934 new cases of acute leukaemia were registered in the Gruppo Italiano Malattie Ematologiche Maligne dell'Adulto (GIMEMA) Archive of Adult Acute Leukaemia. Two hundred cases (5·1%) presented with a history of primary malignancy (PM), 179 of which were acute myeloid leukaemia (AML). The median age of these cases was significantly higher than that of other primitive AML (63 years vs. 57 years; P < 0·001). The number of men was significantly lower than the number of women [74/1544 (4·8%) vs. 105/1420 (7·4%); odds ratio (OR) 0·63, 95% confidence interval (CI) 0·46–0·87; P < 0·002], as was the number of patients aged < 65 years [104/1963 (5·3%) vs. 75/1001 (7·5%); OR 0·69, 95% CI 0·50–0·95; P < 0·01]. An increased incidence of cancer was observed among first-degree relatives of patients with AML occurring after a PM (secondary AML; sAML) [66/179 (36·9%) sAML vs. 757/2785 (27·2%) de novo AML, age adjusted; OR 2·62, 95% CI 1·07–6·42; P < 0·005]. Prevalent types of PM were breast cancer, lymphoma and Hodgkin's disease. sAML occurred after a median latency of 52 months (range 2–379). Of the 122 patients who received chemotherapy for sAML, 67 patients (55%) achieved a complete remission (CR), three a partial remission, 15 (12%) died in induction and 37 (30%) were unresponsive. The median duration of CR was 30 weeks (range 4–250). The median overall survival was 7 months (range 1–196). Comparing acute promyelocytic leukaemia with all other French–American–British (FAB) groups, a significant increase in CR achievement was observed [14/18 (77·7%) vs. 53/101 (52·4%), P < 0·046] as well as in median CR duration (55 vs. 24 months, P < 0·02). The analysis of our data suggests that not only previous chemotherapy but also genetic predisposition could play a role in the pathogenesis of sAML.
Secondary acute myeloid leukaemia (sAML) is usually defined as a disease arising in cancer survivors as a result of a previous chemotherapy and/or radiotherapy, and represents approximately 5–10% of all AML cases (Koeffler & Rowley, 1985; Levine & Bloomfield, 1992; Smith et al, 1996; Thirman & Larson, 1996). The risk of developing a sAML is well documented in different groups of patients, such as children cured of acute lymphoblastic leukaemia (Neglia et al, 1991; Pui et al, 1991; Winick et al, 1993), patients treated for Hodgkin's disease (HD) (Pedersen-Bjergaard & Larsen, 1982; Andrieu et al, 1990; Kaldor et al, 1990a; Van Leuween et al, 1994; Brusamolino et al, 1998) and, particularly, patients with breast cancer and ovarian cancer (Coleman et al, 1987; Kaldor et al, 1990b; Valagussa et al, 1994; Tallman et al, 1995).
A major cause of this event is the leukaemogenic activity of several drugs utilized in the treatment of primary cancer. Radiation therapy is also considered to be potentially mutagenic. In particular, epidemiological studies reported an increased leukaemia risk after treatment with combined chemotherapy, including alkylating agents or epipodophyllotoxins, and after radiotherapy (Vaughan et al, 1983; Brusamolino et al, 1986; Greene et al, 1986; Boivin, 1990; Hawkins et al, 1992).
Moreover, sAML can develop from myelodysplastic syndrome (MDS) or a chronic myeloproliferative disorder, i.e. polycythaemia vera or blastic crisis of chronic myeloid leukaemia (Kantarjian et al, 1986; Rosenblood et al, 1992; Najean & Rain, 1997).
However, to date, no studies have been carried out to determine the incidence of AML occurring after a previous primary malignancy (PM), independently of whether or not the patients have been previously treated with chemo- and/or radiotherapy.
The aim of the present study was to evaluate the proportion of AML occurring after a PM on the basis of data collected in the Gruppo Italiano Malattie Ematologiche Maligne dell'Adulto (GIMEMA) Archive of Adult Acute Leukaemia during a 4-year period and to analyse the epidemiological, clinical and laboratory characteristics of this form of AML.
Materials and methods
The study population comprised 3934 patients, aged 15–94 years, with newly diagnosed acute leukaemia who were registered in the GIMEMA Archive of Adult Acute Leukaemia and observed in 62 haematology departments in tertiary care or university hospitals during the period July 1992 to June 1996. A total of 2964 patients were affected by AML, 901 by acute lymphoblastic leukaemia (ALL) and 69 by acute leukaemia expressing both myeloid and lymphoid surface markers.
Trained haematologists interviewed the patients according to a standardized questionnaire. The form concerned statistical demographic data (race, age, gender), place and date of birth, residence at time of diagnosis, education, previous diseases, working activity at the time of diagnosis and previous activities of the patient. Family composition and history of haematological and non-haematological neoplastic diseases were investigated in patients and in first-degree relatives (parents, brothers and sisters). These data were available for all patients enrolled in the Archive. Moreover, for 127 patients out of 179 with AML diagnosed after a previous history of malignancy, further information could be obtained: type and date of onset of PM, treatment (chemotherapy, radiotherapy, surgery) and outcome of PM, latency between the PM and AML, morphological, immunophenotypic, cytogenetic and molecular biology studies performed at onset of AML, white blood cell (WBC) count, haemoglobin (Hb) and platelet count at diagnosis of AML, induction treatment, complete remission (CR) achievement and duration, and overall survival from AML diagnosis.
Patients with a previous myeloproliferative disorder (13 patients) or myelodysplastic syndrome (151 patients) not secondary to PM were specifically excluded from this study, as were patients previously treated with chemotherapeutic agents for non-neoplastic diseases (eight patients). In our group of 179 AML cases occurring after a PM, 47 patients (26%) had a myelodysplastic phase that preceded the onset of AML that, however, did not fulfil a diagnosis of overt myelodysplastic syndrome.
The end of follow-up was December 1997.
Statistical analysis The major clinical characteristics of patients studied were compared with those of the whole acute leukaemia population recorded in the Archive. The non-parametric Mann–Whitney test was used to analyse statistical differences on distribution of continuous variables. Stratified analysis and adjusted odds ratios (ORs) were computed by the Mantel–Haenszel method. The Kruskall–Wallis test was used for differences on frequencies among patients with sAML treated with or without chemoradiotherapy and secondary acute promyelocytic leukaemia (sAPL).
Among 3934 adult patients with a new diagnosis of acute leukaemia observed during the study period, 200 (5·1%) had a history of a previous PM. Of these, 179 had sAML and 21 secondary ALL (sALL). The characteristics of patients with sALL have already been described (Pagano et al, 1999).
When comparing AML occurring after a PM with the whole AML population recorded in the study period in the GIMEMA Archive (179 sAML vs. 2964 AML), the median age of patients with AML occurring after a PM was significantly higher than other primitive AMLs (63 years, range 26–88 vs. 57 years, range 15–94; P < 0·00001). The number of men was significantly lower than the number of women [74/1544 (4·8%) vs. 105/1420 (7·4%); OR 0·63, 95% confidence interval (CI) 0·46–0·87; P < 0·002]; furthermore, the number of patients aged < 65 years was lower than patients aged > 65 years [104/1963 (5·3%) vs. 75/1001 (7·5%); OR 0·69, 95% CI 0·50–0·95; P < 0·01]. Lifetime prevalence of malignant neoplastic diseases among first-degree relatives (parents, brothers and sisters), adjusted for age, was compared in patients with AML occurring after a PM with de novo AML; a significant increase of neoplastic diseases was observed among first-degree relatives in cases of AML occurring after a PM [66/179 (36·9%) vs. 757/2785 (27·2%) de novo AML, age adjusted; OR 2·62, 95% CI 1·07–6·42; P < 0·005]. Table I shows the types of malignancy in relatives of patients with AML occurring after a PM and de novo AML patients.
Table I. Type of neoplasia among relatives of patients with AML occurring after a PM and de novo AML patients.
|Central nervous system||3||28|
The types of PM are reported in Table II.
Table II. AML following a primary malignancy (127 evaluable cases).
|Hodgkin's disease||10 (7·9)|
|Multiple myeloma||3 (2·4)|
|Chronic lymphocytic leukaemia||2 (1·6)|
|Central nervous system||2 (1·6)|
Complete epidemiological, clinical and laboratory data were available on 127/179 patients (71%) with AML occurring after a PM and are reported in Table III.
Table III. Characteristics of AML occurring after a PM (n = 127).
|Median age (range)||58 (21–87)|
|Median age at the onset of PM (range)||52 (7–86)|
|Median latency between the two malignancy (months, range)||52 (2–379)|
|Previous treatment for PM|
| Only chemotherapy||34|
| Only radiotherapy||20|
| Surgery alone||41|
|Cytogenetic study (performed)||96|
|Median WBC count at AML (range)||6·7 × 109/l (0·6–153)|
|Median Hb level at AML (range)||8 g/dl (4–14)|
|Median platelet count at AML (range)||37·5 × 109/l (3–419)|
|No. of patients with AML who achieved CR (%)||67 (55)|
|Median CR duration of AML (weeks)||30 (4–250)|
|Median overall survival from diagnosis of AML (months)||7 (1–196)|
|Malignancies in first-degree relatives (%)||66* (37)|
AML occurred after a median latency of 52 months (range 2–379) from PM.
All patients were treated for PM: 41 patients underwent surgical resection only, whereas 86 patients underwent chemotherapy and/or radiotherapy (Table III). Sixty-six patients received chemotherapy with/without radiotherapy; among 34 patients treated with chemotherapy alone, 26 patients received alkylating agents, combined in nine of them with epipodophyllotoxins, and eight patients were treated with different chemotherapy protocols. Among 32 patients treated with chemotherapy plus radiotherapy, 26 received alkylating agents, associated in three of them with epipodophyllotoxins, one received epipodophyllotoxins without alkylating agents and five received other chemotherapeutic agents. Therefore, a total of 53 patients received alkylating agents and/or epipodophyllotoxins.
The remaining 20 patients were treated with radiotherapy without chemotherapy, with a median dose of 44 cGy (range 5–120).
In particular, in the sAPL group, seven patients received chemotherapy (in five patients in combination with radiotherapy). All received alkylating agents, associated with topoisomerase II inhibitors in five patients. The other five sAPL cases were treated with radiotherapy alone.
Cytogenetic analysis was performed in 96 patients, and 15 of them also underwent molecular biology studies. Results of the cytogenetic analysis are reported in Table IV. Abnormalities of chromosomes 5 or 7 were present in patients treated with chemo- and/or radiotherapy. Among 19 patients affected by acute promyelocytic leukaemia (APL), t(15;17) was observed in 9 out of 11 patients who underwent cytogenetic analysis; in two cases, metaphases were not sufficient for analysis. In the remaining eight patients, the hybrid gene PML-RARα was detected by the reverse transcription polymerase chain reaction (RT-PCR) technique.
Table IV. Cytogenetic studies in AML occurring after a PM.
|Chromosome 5 or 7 abnormality||2||14||16|
| −5 and −7||–||3||3|
|Trisomy|| || ||6|
| Trisomy 8||1||3||4|
| Trisomy 2||–||1||1|
|Chromosome 16 abnormality|| || ||4|
| Inv 16(p13;q22)||–||1||1|
|Partial deletion|| || ||2|
(+5, +6, +6, +7, +8, +8, +19, +20, +21, +mar)
|Missing or insufficient metaphases||5||7||12|
One hundred and twenty-two patients received chemotherapy for AML, five patients died before starting therapy. Sixty-seven patients (55%) were responsive to chemotherapy and achieved a CR, three additional patients achieved a PR and 37 patients (30%) were resistant; 15 patients (12%) died during induction. Results according to treatment received are reported in Table V. The median duration of CR was 30 weeks (range 4–250). The median overall survival was 7 months (1–196) or 13 months (4–196) for the entire population or for those patients who achieved CR respectively.
Table V. Outcome of AML occurring after a PM in relation to chemotherapy performed.
(hydroxyurea or LdAra-C)
|AIDA protocol or ATRA alone||17||13||–||4|
|3 + 7||18||6||1||11|
|Total*||119||67 (55%)||15 (12%)||37 (30%)|
The analysis of the outcome according to the French–American–British (FAB) classification demonstrated the absence of any statistical difference among each subgroup; however, comparing secondary sAPLs with all the other FAB subtypes, a significant increase in patients achieving CR was observed [14/18 (77·7%) vs. 53/101 (52·5%); P < 0·046] as well as in median CR duration (55 weeks, range 10–250 vs. 24 weeks, range 1–205; P < 0·02). In contrast, no difference was found in survival duration (P < 0·08), even though this result was certainly influenced by the short follow-up period.
As for the influence on CR rate of the chemotherapy utilized for treating sAML, after excluding sAPL patients who were treated with an ATRA-based protocol, the ICE combination (idarubicin, cytarabine and etoposide) gave a higher CR rate than other protocols [21/26 (80·8%) vs. 33/76 (43·4%); P < 0·002]. However, this result is certainly influenced by the significantly lower median age of patients treated with ICE (52 years, range 21–79 vs. 61 years, range 26–87) than those treated with different protocols (P < 0·006).
The CR rate of AML by PM type revealed that patients with previous breast cancer and non-Hodgkin's lymphoma (NHL) had a high CR rate (68% and 71% respectively), whereas patients with previous Hodgkin's disease (HD) showed a lower CR rate (30%) (Table VI), and the probability of achieving a CR when comparing the first two groups with HD was significantly different (P < 0·03).
Table VI. AML outcome by type of PM.
|Chronic lymphocytic leukaemia||2||1||–||1|
|Central nervous system||2||2||–||–|
|Total||119||67 (55%)||15 (12%)||37 (30%)|
Considering the treatments received for previous PM in the largest representative groups (breast cancer, NHL and HD), among patients with breast cancer, 10/37 (27%) were treated with surgical resection only, 7/37 (19%) with chemotherapy, 14/37 patients (38%) with chemoradiotherapy and 6/37 patients (16%) with radiotherapy alone. The majority of patients with NHL (16/21; 76%) were treated with chemotherapy, whereas the remaining five (24%) also received radiotherapy. Eight out of 10 HD patients received chemoradiotherapy and two received only radiotherapy.
Comparing patients treated by surgery alone with patients who received chemo- and/or radiotherapy and patients with APL who received chemo- and/or radiotherapy, significant differences were found: a higher median age was observed in patients treated with surgery alone (P < 0·0001); WBC count and platelet count were lower in sAPL patients (P < 0·001 and P < 0·005 respectively); a significantly higher percentage of CR was observed in the sAPL group (P < 0·03) as well as a longer survival (P < 0·03). No significant differences were found for sex, latency between two malignancies (PM and AML), Hb level, type of chemotherapy for AML (aggressive or palliative) or FAB subtype (Table VII).
Table VII. Comparison between AML previously treated with ablative surgical debridement of AML following chemoradiotherapy and APL following chemoradiotherapy.
|Median age (range)||66 (40–87)||53 (21–85)||58 (27–68)||0·0001|
|Median latency (months)||50 (2–379)||60 (5–336)||36 (19–142)||ns|
|WBC ( × 109/l)||6·9 (0,6–137)||9·5 (780–153)||1·8 (0,6–213)||0·001|
|Hb (g/dl)||8 (6–14)||10 (6–14)||8 (4–14)||ns|
|Platelet (× 109/l)||42·5 (4–41,9)||45·5 (3–30,5)||18 (8–71)||0·005|
|Chemotherapy for AML (aggressive/palliative)||30/9||60/10||12/0||ns|
|CR achievement||20 (52%)||36 (52%)||11 (90%)||0·03|
|Median CR duration (weeks)||24 (2–205)||28 (1–128)||68 (10–250)||ns|
|Survival (months)||6 (1–53)||6 (1–196)||15·5 (2–60)||0·03|
The use of alkylating agents or topoisomerase II inhibitors and/or radiotherapy for a primary malignancy may be complicated by the induction of a second neoplasm, the most frequent of which are MDS and sAML (Vaughan et al, 1983; Brusamolino et al, 1986; Greene et al, 1986; Boivin, 1990; Hawkins et al, 1992).
sAML, representing approximately 5–10% of all AML, has peculiar features and has been frequently described after HD, ALL and ovarian, testicular and breast cancer (Pedersen-Bjergaard & Larsen, 1982; Coleman et al, 1987; Andrieu et al, 1990; Kaldor et al, 1990a, b; Neglia et al, 1991; Pui et al, 1991; Winick et al, 1993; Valagussa et al, 1994; Van Leuween et al, 1994; Tallman et al, 1995; Brusamolino et al, 1998).
In HD treated with the ABVD (doxorubicin, bleomycin, vinblastine, decarbazine) protocol since 1975, sAML has been observed in 0·7% of patients at 15 years from diagnosis, whereas in those treated with MOPP protocol [mechlorethamine, vincristine (Oncovin), procarbazine, prednisone] the incidence ranged from 2·1% to 6·4% at 10 years (Pedersen-Bjergaard & Larsen, 1982; Kaldor et al, 1990a; Van Leuween et al, 1994; Brusamolino et al, 1998). A recent co-operative study indicated an overall 15-year actuarial risk of sAML of 4·2% among 1659 patients treated for HD. In particular, it was 0·3% after radiotherapy alone, 2·8% after chemotherapy alone (2·2% after MOPP, 4·4% after ABVD + MOPP + lomustine) and 5·4% in patients receiving combined modality therapy (10·2% for radiotherapy + MOPP, 15·6% for radiotherapy + MOPP + lomustine) (Brusamolino et al, 1998). In children cured of ALL, the risk of sAML has been evaluated in different series as between 3·8% at 6 years and 5·9% at 4 years (Neglia et al, 1991; Pui et al, 1991; Winick et al, 1993). In women who received alkylating agents for treatment of ovarian cancer the risk was 7% at 10 years (Coleman et al, 1987), whereas in those with breast cancer some authors have reported a lower risk of 2% at 10 years (Valagussa et al, 1994; Tallman et al, 1995). For patients treated for testicular cancer, the risk of sAML after the use of a topoisomerase II inhibitor is 4·7% at 5·7 years (Kaldor et al, 1990b; Pedersen Bjergaard & Philip, 1991).
The real incidence and the characteristics of AML onset after a PM in the absence of previous chemoradiotherapy have never been reported.
The outcome of sAML is generally poor with a CR rate ranging from 40% to 50%, and a short overall survival (Hoyle et al, 1989; De Witte et al, 1995; Anonymous, 1998; Pinto et al, 1998). This poor outcome was also confirmed in our study, in which the CR rate was 56% with a median overall survival of 7 months, even if the small subgroup of sAPL presented a better response with a higher CR percentage and a prolonged survival.
These last data suggest that sAPL responds to chemotherapy as well as de novo APL.
This study, despite the limits of a retrospective study, raises some interesting points.
The 5% proportion of AML occurring after a PM on the global AML population, recorded in our Archive, was similar to those reported by other authors (Koeffler & Rowley, 1985; Levine & Bloomfield, 1992; Smith et al, 1996; Thirman & Larson, 1996). In our series, the most frequent PM was breast cancer, followed by NHL and HD.
The high number of breast cancer as primary tumours in our population could be justified if we consider that this malignancy is the most frequent neoplasm in women (the incidence in Italy is 85·7 cases/100 000/year) (Belzi et al, 1997) and it is very frequently curable with therefore a long follow-up (Pritchard, 1997). In the past, Carey et al (1967) found that in breast cancer patients treated with surgery only the risk of AML was 30 times higher than in the general population. Metcalf et al (1969) clearly documented that in Wistar rats an intraperitoneal injection of methylcholanthrene was capable of inducing an acute leukaemia or a breast cancer. These data suggested to the authors the possibility of a common aetiological mechanism or at least a predisposing condition. Furthermore, Rosner et al (1978), in a review of the literature and from personal experience, observed that acute leukaemia occurred after breast cancer with a sevenfold increase in frequency over the expected number.
In our series, breast cancer represented 29% of all PM, and this percentage was even higher considering only the female patients (37/76 females; 49%). However, a very interesting finding was that 27% of all AML following breast cancer arose from patients treated only with surgery, and overall 30% of patients (41/127) in our series did not receive any kind of chemo- or radiotherapy. We found a significantly higher incidence of cancer in first-degree relatives of patients with AML occurring after a PM than in those of patients with de novo AML. Therefore, for these patients, it is possible to hypothesize that AML raised in the context of a natural predisposition to neoplastic disease is related to inherited alterations that are genetically determined.
These genetically predisposing factors have been recently confirmed in breast, colon and multiple endocrine neoplasia (MEN) syndromes (Marx et al, 1998). Moreover, a genetic predisposition to neoplasm has been also hypothesized in HD and in some chronic lymphoproliferative disease (Pottern et al, 1991).
However, comparing the available cytogenetic data on AML following chemo- and/or radiotherapy for a previous neoplasm (33 patients) with those of AML arising in patients treated only with surgery (10 patients), we observed that abnormalities of chromosomes 5 and 7 were more frequent in therapy-related AML (14/33 vs. 2/10), strongly supporting the role of some drugs in leukaemogenesis (Groupe Francais de Cytogenetique Hematologique, 1984; Le Beau et al, 1986; Dastugue et al, 1988; Philip & Pedersen-Bjergaard, 1988; Pedersen-Bjergaard et al, 1995).
The observation that a significant proportion of patients developed AML without previous chemoradiotherapy, as well as the increased occurrence of malignancies among first-degree relatives of sAML patients, suggests that, besides proven leukaemogenic mechanisms of chemotherapeutic agents and ionizing radiation, genetic factors favouring the onset of neoplastic diseases could be present. Furthermore, the presence of familial genetic syndromes that increase the susceptibility to radiation or chemotherapy inducing new cancer development are well described (Kony et al, 1997).
Molecular biology studies in patients with secondary acute leukaemia frequently showed the presence of germline mutation of the TP53 gene (Felix et al, 1998; Panizo et al, 1998). It has been hypothesized that chemoradiotherapy could induce mutations in DNA repair genes, causing DNA instability; this could predispose to secondary acute leukaemia allowing the repeated development of mutations in cancer-associated genes (Ben-Jehuda et al, 1996).
In conclusion, our study, confirming data on sAML already described in the medical literature, suggests that some so-called ‘sAMLs’ may arise in the absence of any medical or physical treatment. Therefore, further studies on patients with AML that developed only after surgical resection of PM are indicated to evaluate the presence of mutations, which could be responsible for leukaemogenesis in the absence of any chemo- or radiotherapy.
This work was supported in part by a grant from the Ministry of University and Technological Research (MURST) of 1taly.
The principal investigators and participating centres involved in this study were: Pagano, L., Mele, L., Equitani, F. and Leone, G., Cattedra di Ematologia, Università Cattolica S. Cuore, Roma; Pulsoni, A., Avvisati, G. and Mandelli, F., Cattedra di Ematologia, Università‘La Sapienza’, Roma; Tosti, M.L. and Mele, A., Reparto di Epidemiologia Clinica, 1stituto Superiore della Sanità, Roma; Martino, B., Divisione di Ematologia, Ospedali Riuniti, Reggio Calabria; Camera, A., 1stituto di Ematologia, 2a Università‘Federico 2’, Napoli; Cerri, R., Ematologia 1, Ospedale S. Martino, Genova; Di Bona, E., Divisione di Ematologia, Ospedale S. Bortolo, 5icenza; 1nvernizzi, R., 1stituto di Clinica Medica 2, Università di Pavia; Miraglia, E., Divisione di Ematologia, Ospedale ‘Nuovo Pellegrini’, Napoli; Almici, C., Cattedra di Ematologia, Università di Parma; 5isani, G., 1stituto di Ematologia, Università di Bologna; Levis, A., Divisione di Ematologia, Ospedale di Cuneo; Allione, B., Ematologia Molinette, Ospedale Maggiore ‘S. Giovanni Battista’, Torino; Chierichini, A., Divisione di Ematologia, Ospedale di Latina; Nosari, A., Divione di Ematologia, Ospedale ‘Cà Granda’ Niguarda, Milano; Melillo, L., Divisione di Ematologia, Ospedale S. Giovanni Rotondo; Clavio, M., Dipartimento Medicina 1nterna, Università di Genova; Coser, P., Ematologia, Ospedale ‘L. Boehlerstrass’, Bolzano; Candoni, A., Cattedra di Ematologia, Università di Udine, Italy.