Acute myeloid leukaemia in human immunodeficiency virus-infected adults: epidemiology, treatment feasibility and outcome

Authors


Laurent Sutton, M.D., Service d'Hématologie, Hôpital Pitié-Salpétrière, 47 Boulevard de l'Hôpital, 75651 Paris Cedex 13, France. E-mail: laurent.sutton@psl.ap-hop-paris.frMembers of the French Study Group on Acute Myeloid Leukaemia in HIV-infected patients are listed in the appendix.

Abstract

The epidemiology and clinical outcome of acute myeloid leukaemia in human immunodeficiency virus (HIV)-infected adults is poorly documented. We retrospectively surveyed all French haematology centres for adult acute myeloid leukaemia (AML) cases diagnosed between January 1990 and July 1996 who were found to be HIV-seropositive before or at the time of AML diagnosis. Medical charts were reviewed to determine the stage of HIV infection, the characteristics of AML and the response of AML to chemotherapy. Sixteen cases of AML (13 men, three women) were reported by 12 haematology units. Based on assumptions on the size, age and sex distribution of the HIV-infected population in France, the estimated risk of AML in 1990 to 1996 among HIV-infected adults was twice that of the general population (standardized incidence ratio = 2·05; 95% confidence interval, 1·17–3·34). Two other cases occurring before 1990 were spontaneously notified to the authors and were included in the clinical analysis. At AML diagnosis, the median CD4+ cell count was 275 × 106/l and nine patients had acquired immune deficiency syndrome (AIDS). Fifteen patients were scheduled for remission-induction therapy of AML. No deaths were related to AML treatment. Complete remission was obtained in 11 out of 15 patients. Three patients were long-term survivors: two remain alive in complete remission at 8 years and 9 years, respectively, and the third died of AIDS at 8 years. A CD4+ cell count above 200 × 106/l at AML diagnosis was predictive of longer survival (log-rank test: P = 0·004). Like many other malignancies, the incidence of AML appears to be increased in HIV-infected patients. Our results show a twofold higher incidence, although this needs to be confirmed in a specifically designed prospective epidemiological study. Such patients, especially those with CD4+ cell counts above 200 × 106/l at AML diagnosis, should receive remission-induction therapy, which can confer long-term survival.

The link between human immunodeficiency virus (HIV) infection and an increased risk of B-cell non-Hodgkin's lymphoma and Kaposi's sarcoma is well established (Beral et al, 1991; Levine, 1993). These malignancies are acquired immune deficiency syndrome (AIDS)-defining conditions in the Centres for Disease Control (CDC) 1993 revised classification (Castro et al, 1993). With the prolonged survival of HIV-infected patients conferred by the use of effective anti-retroviral and prophylactic strategies, other malignancies, including Hodgkin's disease and solid tumours, have also been reported in this setting (Monfardini et al, 1989; Schulz et al, 1996; Franceschi et al, 1998), although their relationship with HIV infection is unclear (Beral & Newton, 1998).

Rare cases of acute myeloid leukaemia (AML) have been reported during HIV infection and no clear relationship with the characteristic immune deficiency has been established (Napoli et al, 1986; Willumsen et al, 1987; Garavelli & Azzini, 1988; Gold et al, 1989; Peters et al, 1990; Wijermans & ten Kate, 1990; Mansberg et al, 1991; Murthy et al, 1991; Puppo et al, 1991; Rivers et al, 1992; Farber et al, 1993; Al-Bahar et al, 1994; Kane et al, 1997; King et al, 1998). Thus, the epidemiology and clinical outcome of AML in HIV-infected patients have not been adequately documented. We surveyed all haematology centres in France over a 6-year period to estimate the incidence of AML during HIV infection in adults, to describe the clinical and biological features of such cases, and to study the response to chemotherapy and factors influencing outcome.

Patients and methods

Study design We conducted a retrospective survey of all haematology units in France. Centres were asked to report all cases of AML observed in HIV-infected adults (> 18 years of age) between January 1990 and July 1996. A total of 54 clinical haematology units and 88 haematology laboratories belonging to university and non-university hospitals throughout France were contacted by mail and all answered. The cases reported by haematology centres were cross-checked with those notified to the Centre Coopérateur de Données Epidémiologiques sur l'Immunodéficience Humaine (INSERM SC4, Paris, France), which collects epidemiological information on HIV infection and its complications in France. An anonymous datasheet was sent to each haematology centre in which cases had been diagnosed, to collect the information listed in Tables I, II, III and IV. The records were reviewed by a scientific committee composed of infectious-diseases specialists, haematologists and a statistician specializing in clinical haematology. Data were updated in January 2000.

Table I.  Characteristics of patients at diagnosis of AML.


Patient

Age
(years)


Sex
Risk factor
for HIV
infection
Time between
HIV and AML
diagnosis (months)
CD4+
cells
106/l

CDC
stage

FAB
type

Leucocytes
106/l


Karyotype
  1. AML, acute myeloid leukaemia; CDC, Centres for Disease Control classification; del, deletion; FAB, French–American–British classification; inv, inversion; IVDU, intravenous drug user; ND, not done; t, translocation.

124MaleIVDU0202A2M121246, XY
248MaleHeterosexual27460A2M224ND
336MaleHomosexual/IVDU0400A2M34t(15,17)(q22,q12)
429MaleIVDU47647A1M2446, XY
535MaleHomosexual3396C3M497ND
627MaleHomosexual/IVDU381000B1M52246, XY
747MaleHomosexual5327C3M25ND
848MaleHeterosexual015A3M45546, XY
933FemaleHeterosexual25536A1M43inv(16)(p13,q22)
1029FemaleHeterosexual67560A1M133Failed
1126MaleIVDU13110C3M540del(11)(q23)
1236MaleHomosexual6950C3M2346, XY
1341MaleHomosexual51608C1M582ND
1456MaleHomosexual55260A2M551ND
1540MaleIVDU7082A3M22ND
1627MaleHeterosexual50153B3M21·3t(8,21)(q22,q22)
1725FemaleHeterosexual97126B3M534ND
1847MaleHomosexual61290A2M10·946, XY
Table II.  First-line treatment.

Patient
Induction chemotherapy
(doses)
Post-induction
neutropenia (d)

1st CR
Consolidation (± maintenance)
chemotherapy (doses,d from 1st CR)
Relapse
(d from 1st CR)
  1. ABMT, autologous bone marrow transplantation; ATRA, all-trans retinoic acid; Am, amsacrin; Bu, busulphan; Cpm, cyclophosphamide; CR, complete remission; Cy, cytarabine; Dau, daunorubicin; Dox, doxorubicin; Eto, etoposide; Id, idarubicin; L, lomustin; Mp, 6-mercaptopurine; Mtg, mitoguazone; Mtx, methotrexate; NA, not applicable; ND, not determined; Tg, thioguanine. Doses are given per m2 (except when specified per kg).

1Dau; Cy; Tg (210 mg; 1·4g; 1·2g)27YesAm; Cy (0·45g; 6g; d 19)
Dau; Tg; Cy (70 mg; 1g; 0·5g; d 79)
468
2Dau; Cy (135 mg; 1·8g)30YesDau; Cy (90 mg; 1·4g; d 1), then three courses
every 2 months of Tg; Cy; Eto; Mtg (60 mg;
0·48g; 0·3g; 0·8g)
No
3ATRA (25 mg/d for 2 months)0YesMp; Mtx (90 mg/d; 15 mg/w)303
4L; Dox; Cy (80 mg; 105 mg; 1g)40YesL; Dox; Cy (40 mg; 35 mg; 0·4g; d 55)
Tg; Cy (2 mg/kg/d; 1·5 mg/kg/w; d 85–127)
L; Dox; Cy (40 mg; 35 mg; 0·4g; d 146)
No
6L; Dox; Cy (80 mg; 105 mg; 1g)23YesL; Dox; Cy (40 mg; 35 mg; 0·4g; d 13)
Tg; Cy (2 mg/kg/d; 1·5 mg/kg/w; d 43–85)
L; Dox; Cy (40 mg; 35 mg; 0·4g; d 104)
154
8L; Dox; Cy (80 mg; 105 mg; 1g)31YesNoND
9Am; Cy (0·8g; 1·4g)24YesAm; Cy (0·4g; 12g; d 13)
Am; Eto (0·5g; 0·5g; d 48)
383
10Dau; Cy (135 mg; 1·4g)25YesAm; Cy (360 mg; 6g; d 4)
ABMT: Bu; Cpm (16 mg/kg; 200 mg/kg; d 65)
278
11Am; Cy (0·2g; 8g)NDNo NA
12Id; Cy (40 mg; 1·4g)35YesId; Cy (20 mg; 24g; d 84)No
13Dau; Cy (180 mg; 0·6g)26YesDau; Cy (120 mg; 0·5g; d 19)56
14Dau (60 mg)NANoNANA
16Id; Cy (40 mg; 1·4g)53YesId; Cy (20 mg; 8g; d 1)181
17Dau; Cy; Eto (100 mg; 1g; 0·4g)NDNo NA
18Dau; Cy (240 mg; 1·4g)NDNo NA
Table III.  Treatment of relapse.

Patient
Relapse-induction
chemotherapy (doses)

2nd CR
Consolidation chemotherapy
(doses; d from 2nd CR)

2nd relapse (d from 2nd CR)
  1. ABMT, autologous bone marrow transplantation; Am, amsacrin; Bu, busulphan; Cpm, cyclophosphamide; CR, complete remission; Cy, cytarabine; Dau, daunorubicin; Eto, etoposide; Mx, mitoxantrone. Doses are given per m2 (except when specified per kg).

1Dau; Cy (135 mg; 1·4g)YesMx; Cy (24 mg; 1g; d 25)310
3No   
6No   
9Mx; Cy (48 mg; 10g)YesEto; Cy; Am (0·5g; 18g; 150 mg; d 16)
ABMT: Bu; Cpm (16 mg/kg; 200 mg/kg; d 76)
No
10Mx; Eto; Cy (36 mg; 0·6g; 3g)No  
13Mx; Eto; Cy (36 mg; 0·6g; 3g)No  
15No   
Table IV.  Outcome.

Patient
Follow-up: d from
AML diagnosis
AML status at cut-off
date, vital status

AIDS-related events at cut-off date
  1. AP, acute phase; AML, acute myeloid leukaemia; CR, complete remission, FU, follow-up.

19132nd relapse, deadKaposi's sarcoma, lung tuberculosis
228511st CR, deadGastro-intestinal candidiasis
33501st relapse, dead 
434241st CR, aliveCerebral toxoplasmosis
5141st AP, dead 
64011st relapse, dead 
7241st AP, deadLung Kaposi's sarcoma, thrush, cytomegalovirus oesophagitis
8461st CR, lost to FU 
931952nd CR, alive 
103501st relapse, dead 
11941st AP, dead 
122741st CR, deadNon-Hodgkin's lymphoma
131981st relapse, dead 
1421st AP, dead 
15331st AP, lost to FU 
162971st relapse, deadGastro-intestinal cytomegalovirus, thrush, candida septicaemia
171281st AP, dead 
184061st AP, dead 

HIV serological testing was always based on an enzyme-linked immunosorbent assay (ELISA) with Western blot confirmation. The stage of HIV disease was classified according to 1993 CDC criteria (Castro et al, 1993). Acute myeloblastic leukaemia was classified according to the French–American–British (FAB) classification on the basis of cytological, cytochemical and immunophenotyping data, defining subtypes M1 to M7 (Cheson et al, 1990). When available, results of cytogenetic studies were collected. Complete remission was defined by normalization of peripheral blood cell counts, with less than 5% blast cells on bone marrow smears (Cheson et al, 1990). Other patients were considered as having progressive AML. Relapse was defined as the recurrence of AML in patients who had entered complete remission after induction chemotherapy. The cause of death was documented in every case.

Statistical analyses The objective was to compare the incidence of AML in the HIV-infected population and general population in France. The expected number of cases in the HIV-infected population was estimated on the basis of an incidence similar to that of the French general population. The ratio of the observed to the expected number of AML cases in the HIV-infected population in France was calculated after adjustment for age and sex, giving the standardized incidence ratio (SIR). This can be interpreted as a relative risk, with a value above 1 indicating an increased incidence. The 95% confidence interval (CI) of the SIR was calculated by assuming a Poisson distribution of the observed cases (Breslow & Day, 1987).

The age- and sex-specific incidence rates of AML in France were obtained from the Eurocim database, which included all incident cases from nine regional cancer registries between 1988 and 1992, covering 6 million people representative of the French population (Eurocim, Réseau Francim, Grenoble, France, personal communication). The total number of HIV-infected people was obtained from a mathematical mixing model simulating the epidemic (according to this model, the number of HIV-infected persons in France from 1990 to 1996 ranged from 100 000 to 112 000 each year, with the maximum in 1993; Nadal et al, 1995); the age and sex distribution was obtained from a hospital-based information system that included all HIV-seropositive patients consulting 50 hospitals selected at random throughout France (DMI2 data base, Direction des Hôpitaux, Paris, France, personal communication).

The overall survival analysis took into account all deaths, regardless of the cause, from the date of AML diagnosis. Patients lost to follow-up were censored at the date of last contact. Distribution over time was estimated using the Kaplan–Meier product limit method. The following variables were studied: sex, anti-retroviral treatment before AML diagnosis, age, leucocytosis, the CD4+ cell count (above or below 200 × 106/l) at AML diagnosis, and time between HIV infection and AML diagnosis. Groups were compared using the log-rank test. For the study of the complete remission rate, categorical data were compared using Fisher's exact test, and continuous variables were compared using the Mann–Whitney test. Correlations between the CD4+ cell count and the interval between the diagnoses of HIV infection and AML were sought using Spearman's test. All analyses were run on SAS software (SAS Institute, Cary, NC, USA; V 6·12).

Results

Demographic data and relative risk of AML

Sixteen cases of AML (13 men and three women) occurring in HIV-infected patients between 1990 and 1996 were reported by 12 French hospitals. The expected number was 7·79, giving a standardized incidence ratio of 2·05 (95% CI, 1·17–3·34).

Characteristics of HIV infection and haematological data

Two spontaneously reported cases (patients 1 and 2) occurring before 1990 were included in the outcome analysis, giving a total of 18 patients (Table I). None of the cases was diagnosed post mortem. Three of the patients (6, 12 and 13) have been described previously (De la Salmonière et al, 1994; Rabaud et al, 1995; Guillemain et al, 1996). Seventeen patients were infected with HIV type 1 and one (patient 8) with HIV type 2. The stage of HIV disease in the nine patients with AIDS at AML diagnosis was C3 in four cases, with cutaneous Kaposi's sarcoma associated with recurrent oral candidiasis and disseminated varicella zoster infection in patient 5, cutaneous, pulmonary and gastrointestinal Kaposi's sarcoma and cytomegalovirus oesophagitis in patient 7, Burkitt's lymphoma 1 year before AML onset in patient 11, and cytomegalovirus retinitis in patient 12. The patient with Burkitt's lymphoma had received polychemotherapy and was in complete remission. Another patient (13) was in stage C1 (cutaneous Kaposi's sarcoma), two were in stage B3 (patient 16 had recurrent oral candidiasis and had developed testicular carcinoma 16 months before AML diagnosis, and patient 17 had localized cutaneous zoster). Two patients were in stage A3 (8 and 15).

At AML diagnosis, the median CD4+ cell count was 275 × 106/l (range, 15–1000 × 106/l) and the median CD8+ cell count (available for 11 patients) was 720 × 106/l (range, 39–5000 × 106/l). Two patients were hepatitis B surface-antigen carriers (all were tested) and three out of 10 patients tested had chronic hepatitis C (two had both hepatitis virus infections). Prophylaxis of opportunistic infections consisted of monthly aerosolized pentamidine isethionate in two patients and cotrimoxazole in four patients. The other patients were not receiving prophylaxis. At AML diagnosis, nine patients were on at least one anti-retroviral agent (zidovudine alone in six patients, zidovudine plus zalcitabine in two patients and zidovudine plus didanosine in one patient). Two patients (11 and 16) had previously received chemotherapy including etoposide. Patient 17 had a myelodysplastic syndrome consisting of refractory anaemia with ring sideroblasts, 4 months before developing M5 subtype AML.

Treatment and outcome

First-line treatments and response are summarized in Table II. Three patients (16·5%) only received palliative care, because of ‘poor clinical status’ in two cases (patient 5 received oral hydroxycarbamide plus mercaptopurine and died of gastrointestinal bleeding 14 d after AML diagnosis, patient 7 died of pulmonary Kaposi's sarcoma on d 24) and because patient 15 refused treatment and was lost to follow-up on d 33 after AML diagnosis.

Patient 14 received a single dose of daunorubicin (60 mg/m2), but died of adult respiratory distress syndrome caused by leucostasis on d 2 of treatment. Fourteen patients (78%) received at least one complete course of induction chemotherapy (Löwenberg et al, 1999). At the end of the induction course, three patients (6, 13, 18) received recombinant granulocyte colony-stimulating factor (G-CSF).

Chemotherapy failed in four (27%) of the 15 treated patients (14, 11, 17 and 18), who died from leukaemia on d 2 and at 3, 4 and 13 months respectively. Eleven patients (73%) entered complete remission after a median duration of neutropenia (< 0·5 × 109/l) of 27 d (range, 0–53 d) from the outset of chemotherapy. No HIV-related opportunistic infections occurred during aplasia. Infectious complications during neutropenia were documented microbiologically in seven patients. They consisted of Staphylococcus epidermidis bacteraemia (five episodes), bacteraemia owing to gram-negative bacilli (two episodes, one owing to Pseudomonas paucimobilis and one to Escherichia coli) and Candida albicans septicaemia (one episode). These infections were controlled by appropriate anti-infective therapy and by neutrophil recovery in patients who entered complete remission.

Among the 11 patients who entered complete remission, nine received miscellaneous consolidation or maintenance therapies (Table II) (Löwenberg et al, 1999). The two remaining patients, one with M3 leukaemia (3) and one who was lost to follow-up after 46 d (8), did not receive further treatment. Seven of the 11 patients who entered complete remission relapsed (Table III). One (10) relapsed 10 months after AML diagnosis despite consolidation with autologous haematopoietic stem cell transplantation after conditioning with busulphan (16 mg/kg) and cyclophosphamide (200 mg/kg). He was not re-treated and died from leukaemia 2 months later. Six other patients (1, 3, 6, 9, 13 and 16) relapsed after 16, 11, 6, 13, 2 and 7 months respectively. Three of these six patients did not receive further induction chemotherapy and died, two of progressive AML 15 d and 7 months after relapse (3 and 6 respectively), and one of leukaemia with Cytomegalovirus oesophagitis and Candida krusei septicaemia, 116 d after relapse (16). Three other patients received a new course of induction chemotherapy; this failed in one patient (13) who died of leukaemia 4 months after re-induction and yielded a new complete remission in the other two patients (1 and 9). Patient 1 had a second AML relapse and died of leukaemia with cutaneous Kaposi's sarcoma and pulmonary tuberculosis 14 months after the first relapse. Patient 9 underwent autologous haematopoietic stem cell transplantation conditioned with busulphan (16 mg/kg) and cyclophosphamide (200 mg/kg) while in second complete remission. She recovered from HIV-related polyradiculoneuritis. She remains alive in complete remission 8 years after AML diagnosis and is currently receiving highly active anti-retroviral treatment.

Among the first four patients who entered complete remission and did not relapse, one (8) was lost to follow-up on d 46 while in complete remission and two died of AIDS-related events (patient 2, from severe malnutrition owing to recurrent Candida oesophagitis at 8 years, and patient 12, 7 months after the diagnosis of AML, from non-Hodgkin's pulmonary lymphoma, with no evidence of AML at autopsy). The fourth patient (4) remains alive in first complete remission at 9 years, having recovered from an episode of cerebral toxoplasmosis; he is currently receiving highly active anti-retroviral combination and secondary prophylaxis against toxoplasmosis.

The outcome of the patients is summarized in Table IV.

Overall survival and prognostic factors

The Kaplan–Meier overall survival estimate for the 18 patients was 19% (± 19%, 95% confidence interval) at 5 years, with a median survival time of 11 months (Fig 1).

Figure 1.

Kaplan–Meier estimate of overall survival in the 18 patients from the diagnosis of acute myeloid leukaemia (median survival time: 11 months).

The number of cases was too small for multivariate analysis. None of the parameters tested was significantly predictive of complete remission, but a higher CD4+ cell count at diagnosis approached the threshold of significance (P = 0·06). A known duration of HIV infection exceeding 48 months (median value) before AML diagnosis (P = 0·033) and a CD4+ cell count below 200 × 106/l (P = 0·004) were both significantly associated with shorter survival, although no correlation between the CD4+ cell count and the interval between HIV and AML diagnoses was found (Figs 2 and 3).

Figure 2.

Kaplan–Meier estimates of overall survival according to CD4+ cell counts at diagnosis of acute myeloid leukaemia (CD4+ > 200 × 106/l, n = 10; CD4+ < 200 × 106/l, n = 8; P = 0·004).

Figure 3.

Kaplan–Meier estimates of overall survival according to the time between the diagnoses of HIV infection and acute myeloid leukaemia (> 48 months, n = 9; < 48 months, n = 9; P = 0·03)

Discussion

This is the only published series of AML occurring in HIV-infected adults. The calculated incidence of AML among HIV-infected patients was compatible with a twofold increased risk, relative to the incidence of AML in the general French population, which itself is in keeping with incidence rates in other Western countries (Eurocim data, personal communication; McKinney et al, 1989; McNally et al, 1997). This finding is, however, highly dependent on the assumptions made, as the size, sex and age distribution of the HIV-infected population in France between 1990 and 1996 may have been over- or underestimated, leading to an under- or overestimate of the SIR. Therefore, these results need to be confirmed in a specifically designed prospective epidemiological study.

In a large series, Biggar et al (1994) found no increase in the risk of AML after the onset of AIDS-related Kaposi's sarcoma. Similarly, the incidence of AML in never-married San Francisco men aged 25–54 years (a population with a high prevalence of HIV infection) did not change between 1970 and 1979, and between 1980 and 1990 (respectively before and during the HIV epidemic) (Rabkin & Yellin, 1994). This low incidence of AML contrasts strongly with that of Kaposi's sarcoma and non-Hodgkin's lymphoma in HIV-infected adults (Beral et al, 1991). However, our data suggest that, like many other malignancies, the incidence of AML is increased during HIV infection (Franceschi et al, 1998).

The slight increase in the risk of AML found here raises the possibility of a pathogenic link with HIV infection. Myelodysplasia is a well-known complication of HIV infection (Schneider & Picker, 1985) and has been found at a high frequency in systematic studies of bone marrow specimens from such patients (Karcher & Frost, 1991). In contrast to immunocompetent patients, myelodysplasia in HIV-infected patients is usually associated with a low bone marrow blast count with little, if any, acute transformation. However, Napoli et al (1986) and King et al (1998) each reported a case of true blastic transformation following a myelodysplastic period in an HIV-infected patient, and one of our patients also had a myelodysplastic syndrome 4 months before developing AML. Recently, Moses et al (1998) reported that infection of haematopoietic progenitor cells was infrequent and was probably not involved in the pathogenesis of marrow failure, but rather resulted from the infection and perturbation of regulatory functions of auxiliary cells. Whether or not HIV facilitates transformation or is simply associated with a failure of the host to control transformed cells is unclear. The immune depression induced by HIV is probably a risk factor for AML, but 55% of our patients had a CD4+ cell count exceeding 200 × 106/l, indicating relatively preserved immunity. This contrasts with data on non-Hodgkin's lymphoma in patients with depressed immunity related or unrelated to HIV infection (Penn, 1981; Pluda et al, 1990). Few authors have investigated the potential relationship between AML and HIV infection, and they obtained divergent results (Murthy et al, 1991; Farber et al, 1993; Guillemain et al, 1996). In two of our patients, AML may have been secondary to previous chemotherapy rather than to HIV infection. Overall, it is probable that multiple pathways operate individually or in concert within the context of HIV infection to promote leukaemogenesis in certain patients, as suggested for HIV-associated lymphomas (Pluda et al, 1990; Herndier et al, 1994).

Fifteen of the 18 patients in our series received conventional treatment for AML and 73% entered complete remission. Of the seven treated patients reported in the literature, 85% entered complete remission. These results are similar to those obtained in HIV-seronegative patients (Löwenberg et al, 1999), suggesting that HIV infection does not influence AML treatment outcome, at least in the short term. Indeed, the number of relapses was high in our study, affecting 64% of the patients within 15 months (median time to relapse, 9 months). A high AML relapse rate was also observed in the seven patients reported in the literature. This might be indirectly related to difficulties in administering consolidation or maintenance chemotherapy in the setting of HIV infection. However, it is important to note that no HIV-related opportunistic infections occurred during the post-induction neutropenic phase in our series or in other reports, and that no chemotherapy-related deaths occurred. Although the median duration of neutropenia was 27 d among patients who entered complete remission, we observed only one episode of candidaemia (related more to neutropenia than to HIV infection) and no cases of invasive aspergillosis; both infections have been described as life-threatening in the advanced stages of AIDS (Lortholary et al, 1993; Launay et al, 1998). It is, however, difficult to distinguish infections caused by HIV from those caused by neutropenia.

Until now, the decision to treat AML in HIV-infected patients was difficult, as quality of life was potentially poor and the chances of success were unknown. In the present report, a higher CD4+ cell count (above 200 × 106/l), although not significantly associated with a higher remission rate, was predictive of longer survival. Known long-lasting HIV infection or full-blown AIDS at AML diagnosis were associated with poorer survival. The same factors predictive of survival were found by Levine et al (1991) in patients with HIV-associated non-Hodgkin's lymphoma. In the same way, Bermudez et al (1989) found that a history of opportunistic infections and/or Kaposi's sarcoma had a negative impact on the likelihood of achieving a complete response in patients with lymphoma. Thus, we consider that AML in AIDS-free HIV-infected patients with a CD4+ cell count above 200 × 106/l and good performance status should be treated with standard cytotoxic regimens, as we obtained complete remissions in 73% of patients, including durable remissions in three patients, who were probably cured of the leukaemia. One of these latter patients died of AIDS nearly 8 years after AML diagnosis. Of note, two patients underwent autologous bone marrow transplantation without life-threatening complications and one remains alive 8 years later. Solely palliative treatment or low-dose chemotherapy with cytosine arabinoside has been proposed for patients in poor condition (Wijermans & ten Kate, 1990), but this was before the advent of highly active anti-retroviral regimens; the outcome of AML in severely immunodepressed HIV-infected patients might now be different.

In conclusion, this study shows an approximate twofold increase in the risk of AML in HIV-infected patients compared with the general population in France, a finding that needs to be confirmed by a large prospective cohort study. No HIV-related opportunistic infections occurred during AML chemotherapy-induced aplasia. Complete remissions and cases of long-term leukaemia-free survival were obtained in patients with CD4+ cell counts above 200 × 106/l at AML diagnosis, indicating that intensive treatment of AML is feasible in HIV-infected patients. The influence of the CD4+ cell count on overall survival might therefore simply reflect the natural history of HIV disease. It remains to be determined whether the use of highly active anti-retroviral therapies will extend the therapeutic indications for AML in HIV-infected patients.

Acknowledgments

We are particularly grateful to the following for their contribution to this study: Dominique Costagliola and Murielle Mary-Krause who gave us access to the database of the Centre Coopérateur de Données Épidémiologiques sur l'Immunodéficience Humaine (INSERM SC4, Centre Hospitalier Universitaire Saint Antoine, Paris, France); François Bourdillon and Sophie Courtial-Destembert for providing demographic data from the DMI2 database (Mission SIDA, Direction des Hôpitaux, Sous-Direction de l'Evaluation, Paris, France); François Ménégoz who provided us with the incidence rates of acute myeloid leukaemia in nine French departments, based on the EUROCIM database (Réseau FRANCIM, Registre des tumeurs de l'Isère, Grenoble, France); Anne Laporte (Institut de Veille Sanitaire, Hôpital national, Saint-Maurice, France); Laurence Brice (Observatoire Régional de la Santé d'Aquitaine, Bordeaux, France); The members of the Société Française d'Hématologie; and David Young for editing the manuscript.

Appendix

Members of the French Study Group on AML in HIV-infected patients: Faïza Ajana, Centre Hospitalier, Tourcoing; Fréderic Bauduer, Hôtel Dieu, Paris; Réda Bouabdallah, Institut Paoli-Calmettes, Marseille; François Boué, Hôpital Antoine Béclère, Clamart; Anne Coutelier, Hôpital Pitié-Salpétrière, Paris; Hervé Dombret, Hôpital Saint-Louis, Paris; Virginie Éclache, Hôpital Jean Verdier, Bondy; Raoul Herbrecht, Hôpital de Hautepierre, Strasbourg; Mathieu Kuentz, Hôpital Henri-Mondor, Créteil; Jean-Philippe Laporte, Hôpital Saint-Antoine, Paris; Véronique Leblond, Hôpital Pitié-Salpétrière, Paris; Yves Lévy, Hôpital Henri-Mondor, Créteil; Charles Mayaud, Hôpital Tenon, Paris; Éric Oksenhendler, Hôpital Saint-Louis, Paris; Xavier Thomas, Hôpital Édouard Herriot, Lyon; and Brigitte Witz, Hôpitaux de Brabois, Nancy, France.

Footnotes

  1. Members of the French Study Group on Acute Myeloid Leukaemia in HIV-infected patients are listed in the appendix.

Ancillary