|Home | About | Journals | Submit | Contact Us | Français|
The prognosis of patients with relapsed and refractory acute leukaemia (RRAL) is very poor. Forty patients with RRAL were enrolled (28 acute myeloid leukaemia [AML], 12 acute lymphoblastic leukaemia [ALL]) in this Phase 1 dose-escalation trial of daily-infused clofarabine (CLO) followed by cyclophosphamide (CY) for 4 consecutive days (CLO-CYx4). The median age was 48.5 years. The median number of prior regimens was 2 (range 1–5), and 6/40 patients (15%) had prior allogeneic haematopoietic stem cell transplant. 28/40 patients (70%) had adverse genetic features. 6/40 patients (15%) died within 60 days of induction (2 infections, 4 progressive disease). The average time to neutrophil recovery (absolute neutrophil count ≥ 0.5 × 109/l was 34 days, (range, 17–78). The overall response rate (ORR) was 33% (13/40), with 7 complete remissions (18%), 4 complete remissions with incomplete recovery of blood counts (10%), and 2 partial remissions (5%). ORR was 25% (7/28), and 50% (6/12), for AML and ALL, respectively. Notably, the clinical responses were independent of dose level. 7/17 patients (41%) exhibited CLO-mediated enhancement of CY-induced DNA, which was associated with, but not necessary for, improved clinical outcomes. In summary, the CLO-CYx4 regimen was well tolerated and had activity in patients with RRAL, especially relapsed ALL. Therefore, CLO-CYx4 can be considered a salvage therapy for adults with RRALs, and warrants further investigations.
Refractory and relapsed acute leukaemias (RRALs) are characterized by poor response to salvage therapy and dismal outcomes. Clearly, development of more effective therapeutic modalities is vital to improve the outcomes of these patients. Clofarabine (CLO) is a second-generation purine nucleoside analog that was rationally designed to overcome the limitations and incorporate the best qualities of the first-generation deoxyadenosine analogs fludarabine and cladribine (Carson et al, 1992). CLO has activity as a single agent and in combination with other drugs against refractory and relapsed acute lymphoblastic leukaemia (rrALL) (Jeha et al, 2006, McGregor et al, 2009, Advani et al, 2010). Similarly, CLO is active, as a single agent or in combination, in refractory and relapsed acute myeloid leukaemia (rrAML) and in poor-risk newly diagnosed AML (Faderl et al, 2005, Faderl et al, 2008b, Faderl et al, 2008a, Krawczyk et al, 2010, Agura et al, 2011, Tse et al, 2011). The cytotoxic effects of CLO are mediated by multiple mechanisms of action, with the major effects being inhibition of ribonucleotide reductase and DNA polymerases following the incorporation of CLO into DNA, leading to inhibition of DNA synthesis and repair (Parker et al, 1991 Xie & Plunkett, 1996). CLO depletes cellular deoxynucleotide triphosphate pools and enhances the incorporation of CLO 5′-triphosphate (CLO-TP) into DNA, which leads to termination of DNA chain prolongation (Parker et al, 1991, Lotfi et al, 1999). Cyclophosphamide (CY) is an alkylating agent that primarily induces DNA inter-strand cross-links, but these cross-links can be quickly repaired in malignant cells (Yamauchi et al, 2001). Phosphorylation of histone variant H2AX occurs as an early event in response to DNA double-strand breaks (Banath & Olive, 2003). Therefore, the degree of accumulation of phosphorylated H2AX (H2AX) has been used as a proxy for persistently-damaged and unrepaired DNA (Byrd et al, 2002). We have previously shown that administration of CLO prior to CY in a timed-sequential fashion enhanced the levels of DNA double-strand breaks induced by CY and resulted in enhanced blast cytotoxicity and apoptosis in RRAL, leading to objective responses in 67% and 25% of patients with rrALL and rrAML, respectively (Karp et al, 2007).
Unfortunately, the timed-sequential CLO-CY regimen was associated with significant haematological toxicity independent of CLO dosing. This observation was consistent with the possibility of stem cell toxicity related to recruitment and resulting in increased sensitivity to CLO-CY mediated cytotoxicity. Therefore, we hypothesized that a conventional dosing regimen of CLO followed by CY on a daily basis for 4 days (CLO-CYx4) would reduce haematological toxicity while maintaining clinical effectiveness. In this paper, we report the results of the 40 patients enrolled prospectively on this regimen.
This study was a single-institution, modified “3+3” dose-escalation Phase 1 trial of daily CLO followed by fractionated CY in a 4-day regimen (CLO-CYx4) in adults ≥ 18 years of age with RRAL. The original study and its subsequent amendments were approved by the Johns Hopkins Medical Institutional Review Board (IRB). All patients provided informed consent, and the study was conducted in accordance with regulations of the local IRB and the Declaration of Helsinki. The study was registered with the National Cancer Institute at http://clinicaltrials.gov with identifier NCT00293410. The eligibility and exclusion criteria have been described previously (Karp et al, 2007). Patients with prior allogeneic haematopoietic stem cell transplantation (allo-HSCT) were eligible if they exhibited no evidence of graft-versus-host disease. There were no upper limits on age or on the number of prior relapses.
This study was designed to determine the maximum-tolerated dose (MTD) and assess the toxicity of CLO-CYx4 regimen in patients with RRAL. Four cohorts of patients received escalating doses of CLO followed by fractionated CY at 4 dose levels (DLs). The 4 DLs were chosen based on our previous results (Karp et al, 2007), and each DL enrolled 6 to 7 patients. If dose-limiting toxicities (DLTs) were noted in < 2 patients in any DL, the next DL would enroll patients. Intra-patient dose escalation was not permitted. An expansion cohort was planned to enroll an additional 14 patients at the optimal DL selected for further exploration. The treatment plan and the dose-escalation schema are shown in Table 1. On days 1 to 4, CLO was administered intravenously (IV) over 2 h. Two hours after finishing the CLO infusion, CY was infused IV over 2 h. The first dose of CY was split in half and given on day 0 and day 1 of cycle 1 to allow for comparison in the correlative studies between CY alone and the CLO-CY sequence.
Each cycle was defined as 28 days. A second cycle at the same DL was planned for patients who achieved an objective response without experiencing unacceptable toxicity.
Using the National Cancer Institute Common Terminology Criteria for Adverse Events version 3.0 (NCI-CTCAE 3.0; http://ctep.cancer.gov/protocolDevelopment/electronic_applications/docs/ctcaev3.pdf), all adverse events were reported and graded at the discretion of the treating physician. DLTs were assessed during the first cycle of therapy and were defined as the following: 1) Any grade 4 drug-related non-haematological toxicity, 2) Any grade 3 drug-related non-haematological toxicity that did not resolve to ≤ grade 2 within 24 h with the following exceptions: a) grade 3 elevations of one or more hepatic enzymes were considered a DLT only if resolution to ≤ grade 2 required> 7 days, b) grade 3 diarrhoea or oral mucositis were considered DLT only if resolution to ≤ grade 2 required > 48 h, and c) grade 3 neurological toxicity of any duration. If ≥ 2 patients in any DL developed DLTs, then the MTD would be defined as one DL below which the DLTs occurred.
Haematological toxicity was not considered in evaluating toxicity except where bone marrow (BM) aplasia lasted for ≥ 42 days with an overall BM cellularity ≤ 5% or lower with no evidence of leukaemia. Dose delays or dose reductions of CLO for haematological toxicity were not allowed. Myelosuppression-related DLT occurred when the absolute neutrophil count (ANC) did not return to 0.5 × 109/l or the platelet count did not return to 75 × 109/l or 20% of baseline values (whichever was less) by day 42 of therapy, in the absence of leukaemia in BM and in setting of BM cellularity of ≤ 5%.
Responses were evaluated following cycle 1 according to the International Working Group (IWG) 2006 criteria (Cheson et al, 2006) for complete remission (CR), CR with incomplete blood count recovery (CRi), and partial remission (PR).
Demographics and baseline patient characteristics were summarized descriptively. The non-haematological toxicities were recorded according to NCI-CTCAE 3.0 and summarized by DLs. The haematological recovery was described by the median and range of duration to recovery of ANC and platelet count in responding patients. The overall response rate (ORR) was defined as the combination of CR, CRi, and PR. The duration of CR was calculated from the first documentation of CR. Survival was calculated from the first day of salvage chemotherapy to death from any cause or to time of last follow-up.
Samples from peripheral blood (PB) were obtained on day 0 before the infusion of CY (D0 Pre-Tx), day 0 at the end of the infusion of CY (D0 End-CY), day 0 at 2 h after the end of the infusion of CY (D0 Post-CY), day 1 before the infusion of CLO (D1 Pre-CLO), and on day 1 at 2 h after the end of the infusion of CY (D1 Post-CY). When feasible, BM samples at D0 Pre-Tx, D0 Post-CY, and D1 Post-CY were obtained. PB and BM blasts were isolated and prepared for flow cytometry as described previously (Karp et al, 2007). The standard 1 × 104 events were acquired from each sample during flow cytometry. Cellular debris was gated out of both the non-specific staining control and the experimental sample. The debris-corrected, non-specific staining, control histogram was subtracted from the debris-corrected histogram of the experimental sample, yielding the histogram of cells stained specifically for γH2AX. The event count represented by this histogram of cells stained specifically for γH2AX was divided by the event count represented by the debris-corrected histogram of the experimental sample, which included the total of cells stained specifically and non-specifically for γH2AX. This value was multiplied by 100 to give the percentage of blasts with detected DNA-damage.
Previous generations of adenine-based nucleoside analogs were associated with T-cell depletion and immunosuppression. To evaluate the potential effects of CLO on immune function, in particular T-cell mediated immunity, we monitored absolute lymphocyte count (ALC), and CD4+ T-cell lymphocyte subset count at baseline, day 30 (D30) ± 3 days, and weekly thereafter until day 60 (D60), or until recovery to ALC of ≥ 0.2 ×109/l. Quantitative immunoglobulin levels (IgA, IgG, and IgM) were monitored at baseline, D30 ± 7 days, and D60 ± 7 days.
Between December 2007 and March 2009, 26 adults with histologically confirmed RRAL were enrolled across the 4 dosing cohorts. Subsequently, 14 more patients were enrolled between April 2009 and March 2010 in the dose expansion cohort at the optimal DL determined by the Phase 1 trial (DL1). Baseline characteristics of the patients are summarized in Table 2. All 40 patients received at least one cycle of planned therapy, and 5 patients received a second cycle at the same DL. Thirty patients (75%) had refractory leukaemia, while 10 (25%) had relapsed disease. Of 10 patients with relapsed disease, 4 had first complete remission (CR1) duration of ≤ 12 months..
Table 3 summarizes the non-haematological, non-infectious, grade ≥2 toxicities according to the dosing cohort. During the dose-escalation phase, grade ≥3 non-haematological, non-infectious toxicities occurred in 0%, 0%, 14%, and 67%, of patients in DL1, DL2, DL3, DL4, respectively. After cohort expansion at DL1, grade ≥3 non-haematological, non-infectious toxicities were noted in 29% of all patients treated at DL1, and in 28% of the entire 40-patient cohort. The 30-day induction mortality was 2.5% (1 death due to polymicrobial sepsis). No DLTs were observed at DL1 or DL2, one patient had DLT at DL3, and 2 patients had DLTs at DL4. At DL4, one patient had DLT with grade 3 pericarditis leading to atrial fibrillation with pericardial and pleural effusions, while another patient developed a DLT related to a grade 3 non-oligouric acute renal failure (not requiring dialysis), and hand-foot syndrome. Although the MTD was determined to be DL3, it was decided to proceed with DL1 as the optimal dose to be used in the expansion cohort. The reasons included the higher response rates in DL1 (ORR 71%, compared to 17% and 14%, at DL2 and DL3, respectively), the increased incidence of toxicities at DL3 compared to DL1, and the lack of evidence of increased DNA damage in relation to increasing DLs (see below). A notable toxicity was pericarditis or pleuritis with or without associated effusions or arrhythmias in 5 patients (12.5%). This toxicity was seen at all 4 DLs, and was of grade 2 in 4 patients, and grade 3 in 1 patient (a DLT at DL4).
Neutropenic fevers developed in 68% of patients, and were more common in DL4 compared to DL1 (100% vs. 62%). Documented bacteraemias occurred in 9 patients (22.5%), while fungaemias (4 patients) and definitive or probable fungal pneumonias (5 patients) were seen in 19%, 17%, 14%, and 50% of patients in DL1, DL2, DL3, and DL4, respectively. There was no increased incidence of infections with pneumocystis carinii or viral infections.
The prolonged aplasia that was previously reported in 7/18 patients (39%) with the timed-sequential CLO-CY regimen was not seen with the CLO-CYx4 regimen. The median duration of recovery of ANC ≥ 0.1 × 109/l and ≥ 0.5 × 109/l for the 13 patients who achieved clinical responses was 25 days (range, 14–59), and 34 days (range, 17–78), respectively. Of the 2/40 (5%) patients that needed >42 days to recover to ANC ≥ 0.5 × 109/l, one was treated at DL1 with baseline ANC 0.04 × 109/l, and the other had a prior allo-HSCT treated at DL4 with baseline ANC 0.04 × 109/l. The median time of recovery to an untransfused platelet count of ≥ 20 × 109/l, ≥ 50 × 109/l, and ≥ 100 × 109/l for patients who achieved clinical responses was 29 (range 15–67), 39 (range 16–67), and 48 days (range 19–99), respectively.
The objective clinical responses are summarized in Table 4. The ORR was 33% (13/40 patients). Seven patients had CR (18%), 4 patients had CRi (10%) and 2 had PR (5%). For AML, the ORR was 25% (7/28), with 3 CR (11%), 2 CRi (7%), and 2 PR (7%). For ALL, the ORR was 50% (6/12), including 4 CR (33%), and 2 CRi (17%). The clinical responses were not dependent on the DL. The ORR for DL1, DL2, DL3, and DL4 was 48%, 17%, 14%, and 17% respectively. The median duration of CR was 6 months (range, 2–28). All patients who had responses less than CR (CRi or PR) relapsed within 3 months.
Of the 11 patients with CR or CRi, 5 had adverse genetics (including Philadelphia-chromosome [Ph]-positive ALL and FLT3-internal tandem duplication-positive AML). Responses were seen across a wide age range (median 52 years, range 23–73). The median number of prior regimens in responders was 2, but responses were seen in patients who had up to 4 prior regimens (range 1–4). Seven of the 10 patients (70%) with relapsed disease had objective responses (2 AML, 5 ALL). Among patients with relapsed ALL, 5/6 (83%) responded. In contrast, for patients with refractory disease, 6/30 (20%) had clinical responses (5 AML, 1 ALL). One of the 6 patients who had prior allo-HSCT achieved CR and subsequently received donor lymphocyte infusion (DLI). In addition, 3 of the refractory patients who achieved CR were able to proceed to allo-HSCT after CLO-CYx4. Six patients (15%) died within the first 60 days of therapy, 2 from infections and 4 from progressive disease. The median survival for the entire 40-patient cohort was 3.8 months (range, 0.9– 43.7). Among the 13 patients with clinical responses, the median survival was 6.4 months (range, 2.1–43.7). Three patients survived >12 months from induction date, and one remains alive after 43.7 months of follow-up.
Samples from PB were available for analysis of DNA damage in 17 patients (42.5%), 6 of whom had eventual clinical responses (Table 5). Seven of the 17 assayed patients (41%) exhibited greater increases in γH2AX in PB blasts when pretreated with CLO (D1) than when treated with CY alone (D0), with a median D1 net γH2AX accumulation 18% (range 5 – 45%) greater than D0 net γH2AX accumulation. Four of 6 clinical responders (67%) exhibited greater increases in γH2AX when pretreated with CLO (D1) than when treated with CY alone (D0), compared to 3/11 non-responders (27%). The patients who demonstrated CLO-mediated enhancement of CY-induced DNA damage had an ORR of 57% (4/7) compared with the subpopulation in whom CLO-mediated enhancement of DNA damage was not observed (20%=2/10). Neither the presence nor the magnitude of increased net DNA damage in PB blasts was dependent on the CLO or CY dose (Table 5a). Serial BM samples for measurement of γH2AX were available from 6 patients (15%) (Table 5b), of whom 1 achieved an eventual clinical response. Interestingly, this responder had a 3-fold increase in CY-induced DNA damage in BM blasts with CLO pretreatment.
The median ALC before initiation of therapy for the 13 responders was 1.393 × 109/l (range, 0.11–11.43 × 109/l), while the median absolute CD4+ lymphocyte count in this group was 0.352 × 109/l (range, 0.039–1.023 × 109/l). The lowest ALC achieved after therapy was zero × 109/l for 10 of the 13 patients, and 0.01–0.1 × 109/l for the other 3 patients, with a median percentage decline of 100% in ALC. The median duration of recovery of ALC ≥ 0.2 ×109/l was 46 days (range, 15–75), with all 13 patients recovering to ALC of ≥ 0.2 × 109/l. Median ALC on D30 of therapy (±3 days) for the group was 0.130 × 109/l (range, 0– 0.630 × 109/l), while the median absolute CD4+ lymphocyte subset count on D30 (±3 days) was 0.01 × 109/l (range, 0.003–0.156 × 109/l, n=10).
The quantitative immunoglobulin levels were evaluated pre-therapy, on D30 (±7 days), and on D60 (±7 days). For the patients with clinical responses, the median IgG levels pre-therapy, on D30, and on D60 were 10.04 g/l (range, 2.26–19.30), 9.19 g/l (range, 4.44–16.90), and 7.81 g/l (range, 2.84–16.40), respectively. The median IgM levels pre-therapy, on D30, and on D60 were 0.62 g/l (range, 0.1–1.91), 0.47 g/l (range, 0.09–0.98), and 0.38 g/l (range, 0.08–0.73), respectively. The median IgA levels pre-therapy, on D30, and on D60 were 1.5 g/l (range, <0.07 −3.08), 0.74 g/l (range, <0.07–3.82), and 0.78 g/l (range, <0.07–2.65), respectively. The median percentage reduction in IgG levels from pre-therapy levels to D60 was 6% (range, −27 to 31%), for IgM 39% (range, −21 to 72%), and for IgA 32% (range, −18 to 83%).
The outcomes of the majority of patients with RRAL with commonly used salvage chemotherapeutic regimens continue to be dismal, with limited clinical responses and poor long-term survival (Estey, 2000, Fielding et al, 2007, O’Brien et al, 2008). CLO has been studied extensively as a single agent and in combination with other drugs in RRAL in both adults and children. Table 6 lists published prospective trials that evaluated CLO-based combination regimens for salvage therapy of RRAL in adults. Most of these Phase I and II trials have combined CLO with cytarabine or with CY (Karp et al, 2007, Advani et al, 2010, Faderl et al, 2005, Faderl et al, 2008a, Agura et al, 2011, Tse et al, 2011, Becker et al, 2011). The ORR in these trials ranged from 17% to 61%, with CR in the range of 13% to 47%. The wide variability in results can be partly attributed to the different prognostic characteristics of patients enrolled in each study, such as the relative proportions of patients with refractory disease, multiple relapses, and short CR1 (Estey et al, 1996, Estey et al, 1997, Estey, 2000, Becker et al, 2011).
We have previously reported an ORR of 67% and 33% in rrALL and rrAML, respectively, using a timed-sequential CLO-CY regimen, but the haematological toxicity was significant, consistent with the notion of possible haematopoietic stem cell recruitment and enhanced chemotherapy susceptibility (Karp et al, 2007). In the current study, we used a more conventional dosing regimen of CLO followed by CY on a daily basis for 4 days (CLO-CYx4) to decrease the putative stem cell recruitment and potential damage. This regimen was well tolerated, with a manageable side-effect profile, and no haematological DLT was seen. Notable non-haematological toxicities with the CLO-CYx4 regimen were pericarditis or pleuritis, which occurred in 12.5% of patients independent of DL, but were reversible in all cases. Out of 40 patients in this study, 75% had refractory leukaemia, 70% of patients had adverse cytogenetics or molecular features, and 15% had prior allo-HSCT. The use of CLO-CYx4 regimen in this heavily pretreated population resulted in an ORR of 33% (CR, 18%). The ORR for rrALL was 50% (CR, 33%), while for rrAML it was 25% (CR, 11%). While we observed more toxicity with increased DL, there was no clear correlation between the DL and clinical response rates. As a result of these findings, we used DL1 for the dose expansion group. In DL1 (CLO 10 mg/m2/day →CY 600 mg/m2/day × 4 days), the ORR was 48% (CR, 38%).
These findings confirm the clinical activity of CLO-CY combination in RRAL that we previously reported, especially for rrALL (ORR 50% in this trial, and 67% in previous trial) (Karp et al, 2007). In addition, our current findings confirm that CLO-pretreatment can be instrumental in augmenting CY-induced DNA damage in some patients, as measured by γH2AX, albeit not CLO dose-dependent. The immune reconstitution analysis of patients with clinical responses showed a profound and protracted decline in ALC, including the CD4+ subset, with a median recovery duration of ALC of ≥ 0.2 × 109/l of 46 days (range, 15–75). There were no significant changes in the level of IgG between pre-therapy and D60 values for most clinical responders, but there was a trend for reduction in IgA and especially for IgM levels. These findings suggested a more pronounced negative impact on early immunoglobulin responses (IgM level). Further immunological correlates should accompany clinical studies using CLO.
In summary, CLO-CYx4 regimen is well tolerated in patients with RRAL, and has a meaningful clinical activity, especially against relapsed ALL. In line with our results, Vitale et al (2009) reported meaningful clinical responses to a regimen of CLO followed by CY in 2 adult patients with relapsed/refractory Ph-positive ALL who were tyrosine-kinase inhibitor-resistant, one of whom had the highly-resistant ABL1 T315I mutation. Therefore, CLO-CYx4 can be considered as a salvage therapy for adult patients with RRAL, especially for relapsed ALL, and should be considered for further investigations.
This study was supported in part by an investigator-initiated grant from Genzyme (JEK), NCI Cancer Center Support Grant 2P30 CA06973-47, and by philanthropic funds from Dr. Robert Fischell in memory of his wife Marian (JEK).
This study was presented in part at the 53rd Annual American Society of Hematology meeting in San Diego, December 10–13, 2011.
A.M.Z. analysed the data and wrote the manuscript. J.E.K. and R.M.R. designed and performed research, analysed data, and helped with writing the paper. H.E.C. and H.D.Y. analysed the data and revised the manuscript. J.M.G. performed research, collected data, and assisted with data analysis. B.D.S., M.J.L., S.D.G., M.A.M., K.W.P., M.M.S., and D.E.G., performed research and reviewed the manuscript.