Gemtuzumab ozogamicin: first clinical experiences in children with relapsed/refractory acute myeloid leukemia treated on compassionate-use basis

Gemtuzumab ozogamicin (GO) is an immunoconjugate, consisting of a humanized anti-CD33 antibody, to which the cytotoxic compound N-acetyl-γ-calicheamicin dimethylhydrazine, a member of the enediyne antitumor antibiotic family, is linked.1,2 GO selectively targets CD33⁺ cells and was developed as an antileukemic drug for treatment of acute myeloid leukemia (AML), in which CD33 positivity occurs in 80% to 90% of cases.3After binding to the receptor, rapid internalization of the complex occurs, after which calicheamicin is released intracellularly. Calicheamicins are known for their extreme potency, and their cytotoxicity has been described to arise through DNA damage.4 Phase 1 and 2 studies of GO therapy in adults with relapsed AML have been performed and have shown response rates in approximately 30% of patients.1,2,5 Toxicity profiles were relatively mild when compared with classical multiagent chemotherapy, especially with regard to mucositis and infections. However, severe liver toxicity, that is, hepatic sinusoidal obstruction syndrome, may occur. Several factors increase the risk for hepatotoxicity, such as treatment in which GO is combined with conventional chemotherapy, or when GO is administered after previous stem cell transplantation (SCT).6,7 In addition, slow platelet recovery has been described, probably due to damage of the CD33-expressing platelet precursors.2 The drug has recently been approved for use in the United States for elderly patients with relapsed AML.5 

So far, no clear relationship was found between the CD33 positivity of the leukemic cells and the clinical response to GO.2 In a recent study, in which the exact number of CD33 antigens on the cell surface was precisely quantified, again no relationship was found.8 

Almost no data were available until now considering GO treatment of relapsed or refractory AML in children. Sievers et al report the preliminary data of a phase 1 study with GO in 18 children with relapsed or refractory AML, and they conclude that the adverse events are similar to those in adults.9 10 

We report here on our experiences in 15 children with relapsed/refractory AML, treated with GO monotherapy up to 3 doses on compassionate use basis.

Fifteen children were treated with GO on compassionate use basis, after approval was obtained from the local institutional review board where the patient was treated. Informed consent was provided according to the Declaration of Helsinki. The children were diagnosed with de novo AML refractory to standard induction therapy (n = 4), first relapse of AML refractory to reinduction treatment (n = 7), or AML in second (or greater) relapse (n = 4).

The AML-BFM (Berlin-Frankfurt-Münster) Study Group (Münster, Germany), the Dutch Childhood Leukemia Study Group (DCLSG; Den Haag, The Netherlands) and the Nordic Society of Pediatric Hematology and Oncology (NOPHO, Stockholm, Sweden) centrally reviewed the diagnosis of AML in these children, as well as the clinical and cell biologic data presented here. Patient characteristics at initial diagnosis and at the time of GO administration are given in Table1. All patients had CD33⁺ AML at diagnosis, ranging from 27% to 97% CD33⁺ blasts (positive defined as > 20% of the blasts positive for CD33, results shown in Table 2). FAB M0 and M7 cases were overrepresented, reflecting the poor risk characteristics of this patient group.

First-line chemotherapy was given according to 4 different protocols, all based on intensive chemotherapy consisting of cytarabine plus anthracyclines: AML-BFM 93, n = 3; AML-BFM 98, n = 9; MRC12/DCLSG ANLL 97 protocol, n = 2; NOPHO AML 93 protocol, n = 1.

Three of the 4 patients (Table 2: UPN 02, 05, and 09) with primary refractory de novo AML had been treated with standard induction therapy according to the AML-BFM 98 protocol and showed no response. Two patients were treated further with FLAG (a combination of fludarabine, cytarabine, and granulocyte colony-stimulating factor [G-CSF]) plus liposomal daunorubicin or idarubicin, again without response. One of these 2 received further treatment with a course of etoposide and topotecan, before he was treated with GO. The fourth patient (Table 2: UPN 15) with refractory disease was treated according to the MRC12/DCLSG 97 protocol with the standard induction treatment, and was further treated with a course of CLASP (high dose cytarabine plusl-asparaginase) without any response, before treatment with GO.

Most of the 11 patients having relapses had been treated with FLAG with or without anthracyclines (either idarubicin or liposomal daunorubicin) before receiving GO.

Two patients (Table 2: UPN 08 and 11) had undergone SCT before they were treated with GO; in both cases this concerned a matched unrelated donor (MUD) SCT. There were no signs of active graft-versus-host disease of the liver and there was no transaminase or bilirubin elevation at the time of GO administration in these patients, which was given because of subsequent relapse. In one patient (UPN 08) this SCT preceding GO treatment was complicated by veno-occlusive disease (VOD). Twelve months later she had a relapse and was treated with GO.

GO was given at dosages of 4 to 9 mg/m²/course. Seven patients only received one infusion of GO, 5 patients received 2 infusions, and 3 patients received 3 infusions (data summarized in Table 2). Dose levels and frequency were extrapolated from the adult studies and the preliminary pediatric data from Sievers et al,10 and were decided on by the physician in charge of the patient. In some patients the schedule was also based on availability and time needed for preparation of SCT.1,2 10 

We defined response to GO according to the following criteria: a bone marrow blast percentage of 5% or less, in the absence of leukemia in the peripheral blood or elsewhere. To diagnose a complete remission (CR) sufficient recovery of peripheral blood values (> 1000 × 10⁶/L granulocytes and > 100 × 10⁹/L platelets) was required. A CRp was defined as response plus incomplete regeneration of platelets but with platelet transfusion independency.2 Side effects were described according to the National Cancer Institute common toxicity criteria (NCI-CTC; revised version 2.0 of 1999).

After GO treatment a response was observed in 8 of 15 patients, which included a CRp in 5 patients. In 3 patients no change in bone marrow blast count was observed, and in 4 progressive disease occurred. The response and major toxicity data are summarized in Table2.

In the 8 patients with a response, the side effects of GO were moderate with the exception of hematologic toxicity in all patients (NCI-CTC grade 3-4). Two other patients had a febrile reaction during infusion, and one patient (UPN 01) had an infusion-related drop in blood pressure (NCI-CTC grade 3), which needed fluid replacement for 2 days. One other patient (UPN 05) experienced transient NCI-CTC grade 3 hyperbilirubinemia, with normal transaminases and without ascites or weight gain suggestive of VOD. One patient (UPN 08) developed severe GO-related liver toxicity.

Six of the 8 patients with a response received further treatment with SCT. In 3 children with CRp (UPN 01, 02, and 03) further platelet regeneration was not awaited, and they received a transplant (2 matched related donor and 1 MUD-SCT) almost directly following their last GO course. In the 2 other children (UPN 04 and 05) with CRp the interval to transplantation was longer. In one child (UPN 04) receiving a MUD-SCT the interval was 44 days, and the bone marrow showed 8% blasts directly prior to SCT, suggestive of subsequent relapse after GO therapy. In the other patient (UPN 05) allogeneic SCT was performed 2 months after GO treatment.

So far, of the 5 CRp patients, 2 are in continuous CR 6 and 9 months after SCT (UPN 01 and 02). Two patients (UPN 03 and UPN 04) have had a relapse after SCT; UPN 03 died 6 months thereafter from progressive leukemia. One patient died shortly after SCT due to septic shock without peripheral blood regeneration (UPN 05). One other child (UPN 06) who responded to GO received an autologous SCT and died from fungal sepsis 6 weeks following SCT, but without signs of leukemia.

The 2 other patients (UPN 07 and 08) showing a response did not receive further therapy after GO. Interestingly, one of these children (UPN 07) was treated with GO at relatively large time intervals. After the first course of GO (9 mg/m²) he was in aplasia for 28 days and platelets regenerated to a maximum of 27 × 10⁹/L. The bone marrow blasts percentage dropped from 7 to 0 after this course. Three months later he had a relapse and was re-treated with GO (6 mg/m²), showing the same response without any additional toxicity, which was repeated again 2 months later (6 mg/m²). He died of progressive disease 4 months after the last dose. The other patient (UPN 08) with a response was the child with preceding VOD complicating a MUD-SCT. Due to the known high risk for VOD with GO after SCT, the patient was started on prophylactic defibrotide. Despite the prophylaxis, 4 days following GO treatment, she developed NCI-CTC grade 4 liver problems, with the clinical picture of VOD. She was treated with defibrotide and standard supportive care, and recovered fully after 10 days of treatment. Although she responded to GO with a bone marrow blast reduction from 25% to 4%, she relapsed quickly and died 7 months after GO infusion.

The 3 patients (UPN 09, 10, and 11) with no change in the bone marrow blast count received 1, 2, and 3 infusions of GO, respectively, but in all cases no responses were obtained and all 3 children died of progressive disease, after intervals of 2 weeks, 3 weeks, and 3 months after the last infusion, respectively. Apart from NCI-CTC grade 3-4 hematologic toxicity, probably related to the underlying leukemia, no side effects occurred from GO treatment in these children.

The 4 patients with progression (UPN 12-15) after GO treatment all showed grade 4 hematologic toxicity related to underlying leukemia, and one also showed NCI-CTC grade 2 transient transaminase elevation. Two patients died after further palliative therapy and one underwent transplantation and was treated with donor lymphocyte infusion for a subsequent relapse. One child (UPN 15) died at day 6 after GO infusion from progressive disease. This patient also developed grade NCI-CTC grade 4 hepatotoxicty. A postmortem liver biopsy showed massive leukemic infiltration of the liver, without any signs of VOD.

Fifteen children, diagnosed with relapsed or refractory de novo AML, were treated with gemtuzumab ozogamicin (4-9 mg/m² up to 3 courses) on compassionate use basis. The 11 patients having relapses were either refractory to reinduction therapy after relapse or suffered from subsequent relapse. Four patients with newly diagnosed AML were refractory to several different induction regimens. Outcome in this group of children is known to be extremely poor and almost no curative treatment options are available, which is further limited by the significant toxicity that these patients usually experience from previous intensive therapy.11 12 Of the 15 patients, 8 showed a response, that is, a marked reduction of bone marrow blasts to less than or equal to 5% after GO monotherapy. In 5 of these 8 patients a CR, although without full platelet recovery (CRp), was diagnosed. No CRs were diagnosed, that is, absence of leukemia with full hematologic regeneration.

Whether CR and CRp are equivalent in terms of long-term outcome is not fully established as yet. Sievers et al reported that the relapse-free survival between 23 patients in CR (median 7.2 months) and 19 in CRp (median 4.4 months) was not significantly different.2 When GO therapy was followed by SCT as postremission therapy, the 8 CR patients showed similar survival times when compared with the 7 CRp patients (14.5 versus 5.4 months, respectively, P = .272). Although these differences were nonsignificant, the numbers preclude any firm conclusions. In 3 of the 5 children with CRp reported here, further platelet regeneration was not awaited and hence no clear discrimination between CR and CRp could be made.

Our results are in line with the data on treatment of adults with relapsed AML.1,2 However, the adult studies reported on patients treated with GO at first relapse only, and according to a fixed schedule, whereas the children included in this report were treated at a later stage in their disease (which implies higher cumulative toxicity) and with different dosages of GO varying from 4 to 9 mg/m.2 In the preliminary report of the phase 1 study with GO in CD33⁺ relapsed/refractory AML in 18 children by Sievers et al, 4 patients had less than 5% bone marrow blasts after the second dose of GO.9 10 No data are provided considering postremission therapy or longer follow-up of these children.

In our patients reported here, 2 children are still alive in remission after treatment with GO monotherapy followed by intensive postremission therapy with SCT, although follow-up is short. They both achieved a partial complete remission (CRp) following GO treatment. The fact that these children had not undergone transplantation prior to GO treatment may have contributed to the relative paucity of treatment- and transplant-related toxicity or morbidity. In addition, these data suggest the need for intensive postremission therapy following remission induction by GO, to induce a more durable therapy response (now lasting 6 and 9 months in these 2 children). This is similar to the experience in adults.13 

Considering GO side effects, hematologic toxicity was difficult to assess due to subsequent SCT or underlying leukemia. With regard to nonhematologic toxicity, GO was relatively mild with the exception of liver toxicity. No mucositis or severe infections were documented. In this series only one patient developed GO-related VOD after a prior SCT, which is a well-known risk factor to develop VOD after subsequent GO treatment.7 Two others developed transient hepatic toxicity, which resolved spontaneously. In one child the hepatic toxicity could be attributed to progressive leukemia and infiltration of the liver. In the phase 1 study by Sievers et al, 4 of 18 patients experienced grade 3 or 4 side effects, including respiratory failure and hyperbilirubinemia, prolonged pancytopenia, gastrointestinal bleeding and congestive heart failure, and transient transaminase elevation.10 

It is noteworthy that one of our patients (UPN 07) had been treated repeatedly with GO, with relatively long intervals between the infusions, and responded each time without showing any signs of additional toxicity. Although our experience is limited to this patient only, it suggests that palliative treatment of some patients with AML with repeated dosages of GO at relatively long time intervals is feasible, and needs to be explored further.

In conclusion, this report on compassionate use of GO in children with relapsed/refractory CD33⁺ AML demonstrates that GO has clinical activity in these children. Further studies in children are needed to clearly establish the clinical efficacy and safety of GO in pediatric AML with more stringent eligibility and dose and scheduling criteria. A phase 2 trial by the AML Committee of the International BFM Study Group is currently recruiting patients in such a study.

The authors wish to thank all hospitals and clinicians who treated the patients and provided us with the treatment data. We are indebted to Wyeth Pharmaceuticals (St Davids, PA) for providing us with GO for the treatment of our patients on compassionate use basis. We would also like to thank Mrs A. Heus for her much appreciated secretarial help. This is a report of the AML Committee of the International BFM Study Group.