Treosulfan and fludarabine: a new toxicity-reduced conditioning regimen for allogeneic hematopoietic stem cell transplantation

For many years, conditioning therapy prior to allogeneic transplantation has been based on the combination of the alkylators busulfan and cyclophosphamide or total body irradiation (TBI) and cyclophosphamide. Stem cell toxicity to induce myeloablation, immunosuppression to improve engraftment/prevent rejection, and antineoplastic activity are the major goals of the supralethal doses of chemotherapy and radiation. With the introduction of new conditioning approaches such as nonmyeloablative treatment¹  or mini-transplantation,²  the perception of the aims of conditioning for certain indications has changed: conditioning now is necessary to provide a platform for the development of complete donor chimerism, and the graft-versus-leukemia/lymphoma/tumor effect is considered to be more important for cure than the antineoplastic activity of the conditioning regimen itself. Therefore, severe mucositis and leukopenia are no longer unavoidable in the early posttransplantation period, when doses of conditioning therapy are significantly reduced. However, other side effects such as veno-occlusive disease (VOD) induced by busulfan^1,3-5  may not be excluded even after reduced-intensity conditioning. In addition, with the use of “new” drugs like melphalan, adverse events that were not normally associated with these drugs, such as cardiotoxicity, have been reported.^6,7  Therefore, the introduction of alternative conditioning agents such as treosulfan may further improve the therapeutic options in allogeneic transplantation.

Treosulfan (l-threitol-1,4-bis-methanesulfonate; dihydroxybusulfan; NSC 39069) is a prodrug of a bifunctional alkylating cytotoxic agent that is approved for the treatment of ovarian carcinomas in a number of European countries.^8,9  Hematotoxicity is dose limiting in conventional therapy.¹⁰  Only recently, in two phase 1 protocols with autologous blood stem cell rescue, was it possible to escalate the dose to almost 5 times the maximum tolerated dose defined by conventional therapy before mucositis/stomatitis, diarrhea, skin toxicity, and acidosis became dose limiting.^11,12  In vitro data confirm the pronounced effect of treosulfan against primitive hematopoietic stem cells,¹³  and repeated dosing of treosulfan is at least as effective as busulfan or TBI in inducing reliable donor-cell engraftment in mice.^14,15  The antitumor activity of treosulfan has been shown in a variety of solid tumors.^16-19  In myeloma and in chronic and acute myelogenous leukemia, treosulfan is a potent inducer of cell death, being equal or even superior to melphalan or busulfan.^20-22  Recently, antileukemic activity superior to that of equitoxic doses of cyclophosphamide or busulfan has been shown in in vivo human acute lymphatic leukemia models.²³  Treosulfan may even have advantages over orally administered busulfan because of its pharmacologic characteristics and intravenous route of administration. The active metabolites, monoepoxide intermediate and l-diepoxybutane, are formed by a nonenzymatic, pH- and temperature-dependent intramolecular nucleophilic substitution, resulting in a predictable pharmacokinetic profile.^24,25  Both epoxide species are supposed to be responsible for DNA alkylation, interstrand crosslinks, chromosomal aberration, and induction of apoptosis.^26,27  Because of its limited nonhematologic (organ) toxicity, treosulfan may therefore offer a promising alternative to conventional conditioning agents in the setting of toxicity-reduced allogeneic transplantation.

Fludarabine has already been incorporated into a number of nonmyeloablative or reduced-intensity conditioning regimens.^1,28,29  Fludarabine is known to inhibit DNA repair.³⁰  Besides its activity against lymphoid neoplasms,³¹  its favorable toxicity profile and assumed immunosuppressive characteristics were the reasons for its introduction into several conditioning regimens.¹ 

In the current phase 1/2 study, treosulfan and fludarabine were used as the preparative regimen before allogeneic transplantation, combining the effects of DNA damage caused by treosulfan with inhibition of repair by fludarabine. The treosulfan dose of 30 g/m² (3 times the maximum tolerated dose of conventionally dosed treosulfan) was chosen in order to ensure sufficient stem cell toxicity and antineoplastic efficacy and at the same time to avoid serious nonhematologic toxicity.

Patients with acute myeloid leukemia (AML), myelodysplastic syndrome (MDS), chronic myeloid leukemia (CML), multiple myeloma (MM), non-Hodgkin lymphoma (NHL), or chronic lymphocytic leukemia (CLL) who were between the ages of 18 and 65 years and at unacceptable risk for conventional conditioning were enrolled by 3 German transplantation centers. The first 5 patients were treated in a pilot phase according to the protocol, which was approved by the ethics committees of all 3 centers thereafter. The study was conducted in accordance with the Declaration of Helsinki protocol. All patients gave written informed consent. The characteristics of the patients and the risk factors for conventional conditioning are listed in Table 1. Patients were required to have an HLA-identical family or unrelated donor, as determined by serologic typing for HLA A/B and molecular typing for HLA DRB1.

Patients received fludarabine (Schering, Berlin, Germany) 30 mg/m² intravenously each day over 30 minutes from day –6 to day –2 and treosulfan (Medac, Hamburg, Germany) 10 g/m² intravenously over 2 hours daily from day –6 to day –4. For patients with unrelated donors, rabbit antithymocyte globulin (Fresenius, Gräfelfing, Germany) 10 mg/kg was given intravenously daily from day –4 to day –2.

Patients were managed in laminar airflow rooms. All patients received prophylactic cotrimoxazole against Pneumocystis carinii infection. Acyclovir and fluconazole or itraconazole prophylaxis was routinely used. Red cell and platelet transfusions were given to maintain hemoglobin levels higher than 80 g/L and platelet counts higher than 20 × 10⁹/L. Blood products were irradiated. Neutropenic patients received broad-spectrum antibiotics according to each hospital's policy for the management of febrile neutropenia.

Filgrastim at 5 μg/kg per day was administered intravenously or subcutaneously from day 1 until the leukocyte count had been at least 1.0 × 10⁹/L for 3 consecutive days.

Prophylaxis for graft-versus-host disease (GVHD) consisted of 2 doses of cyclosporine A (CSA) 1.5 mg/kg every 12 hours starting on day –1. CSA was administered intravenously until discharge and continued afterward as oral medication. Doses were adjusted to maintain whole-blood steady-state trough concentrations between 200 ng/mL and 250 ng/mL and modified as clinically indicated for nephrotoxicity. The CSA was administered at full dose through day 42. Thereafter, if GVHD did not occur, the protocol allowed a tapering of the CSA dose by 10% every week.

Chimerism was evaluated using polymerase chain reaction (PCR) to amplify microsatellites. Genomic DNA was prepared from blood and bone marrow using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany). Amplification of short tandem repeats (STRs) (D3S1358, VWA, FGA, Amelogenin, D8S1179, D21S11, D18S51, D5S818, D13S317, D7S820) was performed by multiplex-PCR using the AmpFlSTR Profiler Plus-Kit (Applied Biosystems, Weiterstadt, Germany). Fragments were detected by high-voltage capillary electrophoresis using fluorescent multicolor dye technology (ABI Prism 310 Genetic Analyzer; Applied Biosystems). Monitoring of recipient and donor alleles started at day 14. Further controls were performed at days 28, 56, 84, 112, and 180 and then every 6 months or if clinically indicated. In cases of sex mismatch of donor and recipient, a fluorescent in-situ hybridization method (Vysis, Downers Grove, IL) was used to analyze the X/Y composition of 500 cells.

Patients who showed incomplete hematopoietic donor-type chimerism on or after day 28 or had evidence of relapse or disease progression were candidates for adoptive immunotherapy. In patients still on immunosuppressive therapy but without signs of GVHD, immunosuppression was reduced. In patients with nonetheless persisting incomplete hematopoietic donor-type chimerism or increasing proportions of recipient cells, the immunosuppression was terminated. Patients without any concurrent immunosuppression and no signs of GVHD were candidates for donor lymphocyte infusion (DLI).

The 2-sample exact Wilcoxon Mann-Whitney test was used to identify any significant difference in the distribution of GVHD with respect to the donor type (related or unrelated). Actuarial curves and survival rates were estimated by the Kaplan-Meier method. Overall survival was measured from transplantation until death from any cause. Event-free survival was measured from transplantation until graft failure, progression, relapse, or death from any cause. Treatment-related mortality was determined from the date of transplantation until death related to transplantation. Patients who died from other causes were censored at the time of death. Toxicity was evaluated according to the Cancer Therapy Evaluation Program Common Toxicity Criteria (CTC) version 2.0 (National Cancer Institute, Bethesda, MD).

Between March 1999 and January 2002, 30 patients (20 men and 10 women) with a median age of 49 years (range, 20-60 years) received a transplant. Of these, 16 patients had a matched unrelated donor, and 14 had a matched related donor. Of the donors, 14 donated bone marrow and 16 donated peripheral blood stem cells with a median of 2.02 × 10⁶ CD34⁺ cells/kg or 3.02 × 10⁶ CD34⁺ cells/kg of body weight of the recipient, respectively (range, 0.39-3.33 × 10⁶ cells/kg or 1-15.9 × 10⁶ cells/kg, respectively). Of the patients, 6 had CML, 3 had MDS, 7 had AML (5) or secondary AML (2), 5 had MM, 4 had chronic lymphocytic leukemia, and 5 had NHL (1 T-cell NHL, 1 diffuse large-cell NHL, and 3 follicular cell lymphoma) (Table 1).

There were 29 patients (97%) who had primary engraftment, as defined by leukocyte counts higher than 1.0 × 10⁹/L and platelet counts higher than 50 × 10⁹/L without transfusion for at least 3 consecutive days. The median leukocyte nadir (median, 0.5 × 10⁹/ L; range, 0.03-59 × 10⁹/L) occurred at day 7. In 17 patients with a normal platelet count at the beginning of the conditioning, the median time to platelet transfusion dependency—defined as the first day with platelet count lower than 20 × 10⁹/L or substitution of platelet concentrates—occurred by day 7. A total of 8 were transfusion dependent, and 5 already had decreased platelet counts at the beginning of the conditioning. Overall, leukocytes recovered to higher than 1.0 × 10⁹/L at a median of day 10 (range, days 4-19), neutrophils reached 0.5 × 10⁹/L at a median of day 11.5 (range, days 6-19), and platelets exceeded 50 × 10⁹/L by a median of day 18 (range, days 9-35).

Complete donor chimerism, defined as more than 95% donor cells in the bone marrow, was reached on day 14 by 70% (16/23) of the patients evaluated, and on day 28 by 81% (22/27). In addition, 2 patients had complete donor chimerism when first tested on day 56, and 1 patient had reached complete chimerism by day 56. In a patient with CML (no. 10), complete chimerism was achieved only after reduction of immunosuppression (100% donor cells by day 130), and a patient with CLL (no. 18) remained a mixed chimera. The increasing CLL clone contributed 85% of recipient hematopoiesis by day 140. In the following 5 months, CLL progressed despite DLI and treatment with gemcitabine until a combination of alemtuzumab and docetaxel was started. This patient has now reached complete donor chimerism and complete remission of his CLL. Therefore, in total, 90% (27/30) of the patients reached complete donor chimerism at some time after transplantation.

One patient (no. 11) had a primary graft failure. This patient suffered from CML diagnosed 9 years before transplantation. At 3 months before transplantation, a first blastic phase of lymphoid origin occurred. Prior to the conditioning therapy, he had been leukopenic and transfusion dependent. Leukocytes exceeded 1 × 10⁹/L on day 4. Transfusion continued to be required to maintain an adequate platelet count. Chimerism analysis on days 14 and 28 showed no signs of donor hematopoiesis.

One patient with CLL (no. 28) reached transient chimerism of approximately 75% donor cells in the bone marrow by day 28 but had lost all signs of donor hematopoiesis by day 45. Pretransplantation blood samples, evaluated retrospectively, showed a nonspecific positive crossmatch resulting from multiple autoantibodies.

One patient (no. 17) died of septic shock on day 29 before an evaluation of donor chimerism was possible.

As shown in Table 2, the conditioning regimen with treosulfan and fludarabine was well tolerated. Acute nausea and vomiting reached CTC grade 3 in only one patient and did not represent a clinical problem when patients were given 5HT₃ antagonists and dexamethasone. Patients were able to eat at all times after conditioning. Mucositis did not exceed CTC grade 2, and a third of the patients remained free of any signs of mucositis. Nephrotoxicity was attributable mainly to CSA and antibiotic/antimycotic therapy. There were 10 patients who suffered from liver enzyme (serum alanine aminotransferase [ALT] or aspartate aminotransferase [AST]) elevations reaching CTC grade 3. However, 7 of these patients had histories of liver toxicity and showed CTC grade 1 or 2 elevations before the conditioning regimen was started. Bilirubin elevations could not always be clearly attributed to the toxicity of the conditioning regimen, comedication after transplantation, GVHD, or ABO incompatibility. Of the patients, 4 (13%) had CTC grade 3 and 1 had a CTC grade 4 bilirubin elevation. At the time of rising bilirubin values (days 8-14), the latter patient (no. 16) suffered from hemolysis caused by minor ABO incompatibility and had a concomitant CSA concentration considerably higher than the therapeutic range. Bilirubin values dropped after the hemolytic event resolved and GVHD prophylaxis was changed to tacrolimus. Diarrhea CTC grade 3 developed in 3 patients (10%) 3 to 4 weeks after transplantation and could not be clearly differentiated from cytomegalovirus reactivation (1 patient) or GVHD (2 patients). The single case of constipation reaching CTC grade 3 did not result in clinical problems.

No infections or fever were found in 20% of the patients. However, 2 patients suffered from CTC grade 4 infections, 1 of whom (no. 17) died, as noted above. Peripheral polyneuropathy (CTC grade 3) occurred in one patient who had suffered from polyneuropathy during previous therapies and also displayed disturbed consciousness in response to medication given for its treatment. A virus infection was the suspected cause of the continuing problems, and she recovered slowly after antiviral treatment. Of the patients, 15 (50%) complained of pain after transplantation mainly as a result of catheter-related problems (3; 10%), filgrastim and engraftment (7; 23%), or headache (4; 13%).

Of the patients, 6 (20%) have died without signs of a relapse of the previous malignant disease. In 2 patients, severe fungal infections resistant to liposomal amphotericin B and caspofungin (died on day 146 and day 232) were seen. In 3 patients, infections of unknown origin were observed. All 3 patients were still under immunosuppression with CSA (died on days 29, 87, and 112). The suspected causes were bacterial sepsis in the first 2 patients and fungal infection in the third patient. One patient was lost to sudden cardiac arrest most likely caused by his severe accompanying cardiac amyloidosis (day 62).

All patients were evaluated for acute GVHD (aGVHD; Table 3). Of these, 43% (13/30) remained without signs of aGVHD, 27% had grade I disease, and 17% had grade II aGVHD. There was a lower incidence in the patients receiving grafts from a matched related donor (grade 0: 50% vs 38%, grade I: 21% vs 31%, grade II: 14% vs 19%). Grades III and IV aGVHD each occurred in 7% of the patients (2/30) without any major difference according to the donor-recipient relation. No significant difference in the distribution of GVHD according to donor type was found (P = .669). The primary immunosuppression with CSA was maintained in median until day 138 (range, days 72-476).

There were 26 patients evaluable for chronic GVHD (cGVHD). Limited cGVHD was found in 2 patients (8%) and extensive cGVHD in 8 (31%). There were no major differences between the donor types and the occurrence or extent of cGVHD. The DLI induced GVHD in 2 patients.

Disease status and transplantation outcome are shown in detail in Table 1. There were 10 patients with AML or MDS who received transplants; 3 patients with AML in the first, second, or third CR did not relapse. However, 2 of 4 patients who received transplants while not in complete remission (in the first relapse in one case and the first partial remission [PR], as defined by 5% to 25% blasts after chemotherapy, in the other) relapsed, and 1 of them died of this relapse on day 186. The other 2 patients who received transplants during the first relapse or in the second PR died of a septic episode (day 29) and an invasive fungal infection (day 112) without signs of leukemia progression or relapse. All 3 patients with untreated MDS achieved a complete remission after transplantation. None of these patients relapsed, but one died on day 146 from an invasive fungal infection.

Six patients with CML received transplants a median of 5 years after first diagnosis (range, 3-9 years). All 5 patients in a first chronic phase reached a molecular (4 patients) or complete cytogenetic (1 patient) remission after transplantation. The sixth patient (no. 11) received a transplant 9 years after diagnosis but experienced primary graft failure. Of the 5 patients with CR after transplantation, 3 had cytogenetic relapses (days 56, 163, and 430). After reduction of immunosuppression (1 patient) or DLI (2 patients), all 3 patients are now in molecular remission. A fourth patient died on day 232 secondary to infection, and the fifth patient suffers from cGVHD but remains in molecular remission after 807 days.

None of the 5 patients with multiple myeloma had a relapse or progression. However, one patient (no. 20) showed a transient increase in aneuploid plasma cells, as judged by fluorescence-activated cell sorter analysis. After reduction of the immunosuppression, aGVHD developed, and the aneuploid plasma cell fraction was cleared from the bone marrow. The patient has remained in CR since then. One patient (no. 9) with secondary amyloidosis showed a significant reduction of related symptoms (cardiac and skin involvement), and a second patient with multiple organ involvement by amyloidosis died of sudden cardiac arrest on day 62, as mentioned above.

There were 9 patients with NHL treated. Of the 4 patients with CLL, 2 reached CR. One patient (no. 18) required further treatment (refractory CLL before transplantation) before reaching CR, and the final patient (no. 28) was not evaluable for response, as there was only a transient engraftment of 75% on day 28 and no signs of donor hematopoiesis by day 45. At days 331, 470, and 631, 3 patients with follicular center lymphoma remain in CR. One patient (no. 26) with a diffuse large-cell NHL relapsed on day 113 and died of his disease on day 371. One patient (no. 24) with a T-cell NHL died on day 87 secondary to infection without any signs of disease progression.

After transplantation, 8 patients have died: 6 (20%) of non–relapserelated causes and 2 (6.7%) of relapse on day 186 (AML) and day 371 (NHL). Thus, the estimated overall survival rate is 73% and the event-free survival rate is 49% after a median follow-up of 22 months (range, 7.4-33.4 months) (Figure 1). Taking into account that the success of a less-toxic conditioning regimen is in part related to a graft-versus-malignancy effect after transplantation, reduction of the immunosuppression or DLI resulted in a current survival rate in complete remission of 63%.

There were no treatment-related deaths by day 28. By day 100, the treatment-related mortality reached 7% and did not exceed 17% by one year or thereafter. In the evaluation of treatment-related mortality, the patient (no. 8) suffering from sudden cardiac arrest secondary to severe accompanying amyloidosis was censored at the time of death (day 62) (Figure 2).

This study shows that treosulfan can be combined with fludarabine for conditioning of patients prior to allogeneic transplantation. A short myelosuppression comparable with that produced by reduced-intensity or nonmyeloablative regimens such as busulfan and fludarabine or melphalan and fludarabine^5,7,32  was observed. Certain toxic side effects reported after reduced-intensity conditioning were less prevalent or did not occur in our patients. While reduced-dose combinations with busulfan may still lead to severe hepatic toxicity (Bearman grade II)^33,34  or to VOD,^1,5  the liver toxicity (ALT/AST elevation) seen in this study was transient and did not exceed CTC grade 3. No VOD occurred in the patients treated with treosulfan and fludarabine. Pneumonitis and pulmonary toxicity^33-36  or severe left ventricular failure/cardiac toxicity^6,34  associated with reduced-dose busulfan and fludarabine or melphalan combinations was likewise not observed. The treatment-related mortality was not a consequence of nonhematologic side effects of the conditioning regimen but resulted in 4 of the 6 cases from infections associated with direct immunosuppression or GVHD. Considering the selection of poor-risk patients for this study, the treatment-related mortality rate of 17% compares favorably with that of reduced-intensity approaches.^34,35,37-39 

Despite the low nonhematologic toxicity, the combination of treosulfan and fludarabine may provide a rather high antineoplastic activity. Of the 7 patients with AML, 6 had a high relapse risk (transplantation in second or later CR, first or second PR, or relapse), and only 2 patients suffered from relapse. No relapses were seen in the MDS or follicular center lymphoma patients, and none of 5 patients with multiple myeloma (4 in the first through the fourth PR or with progressive disease) relapsed or progressed. This first impression is supported by the in vitro data of the antineoplastic activity of treosulfan in multiple myeloma²⁰  but has to be confirmed in a larger patient population with these diseases. The rate of molecular or cytogenetic relapse in CML will have to be followed closely in the future; however, it has to be taken into account that these patients had long histories of disease or advanced CML. An additive effect of imatinib and treosulfan has been described in vitro and may be a subject for further investigations.²¹ 

It has been reported that as many as a third of patients have mixed chimerism after conditioning with regimens such as fludarabine and melphalan³² ; fludarabine, cyclophosphamide, and idarubicin³⁹ ; or even fludarabine and busulfan.¹  In comparison, treosulfan and fludarabine conditioning induced initial full-donor chimerism in 81% of the patients. In addition, the rate of primary graft failures was low and in the range of that obtained with other conditioning regimens.^34,35,38,39 

Cyclosporine A was the only immunosuppressive GVHD prophylaxis given after transplantation. This practice did not result in a higher frequency of aGVHD (14% grades III + IV) than has been reported with other combinations, which frequently are associated with grades III and IV GVHD in more than 15% of patients.^1,7,34,35  The rate of cGVHD (8% limited and 31% extensive) was likewise low.^1,7,35,40 

The comparison of the survival rates of patients treated with different reduced-intensity conditioning regimens is difficult because disease status at transplantation and the heterogeneity of patient populations differ considerably among studies. The Kaplan-Meier estimates for overall survival of 73% and event-free survival of 49% in the present series encourage further evaluation of the treosulfan-containing regimen. Reduction of immunosuppression or DLI (with additional chemotherapy in one patient with refractory CLL) cured an additional proportion of the patients who relapsed or had persistent disease after transplantation and resulted in an estimated 63% probability of continuing CR after a median follow-up of 22 months.

In the context of the new conditioning approaches ranging from very light protocols (“mini-transplantation”) to those with only attenuated-intensity combination chemotherapy derived from existing conventional transplantation protocols, the treosulfan and fludarabine regimen has to be considered a more intensive conditioning approach.⁴¹  Application of almost two thirds of the maximum tolerated dose (47 g/m² after single injection^11,12,42 ) over 3 days obviously induces sufficient myeloablation/stem cell toxicity. In combination with fludarabine, potent pretransplantation immunosuppression is given to guarantee prompt engraftment and complete donor chimerism in nearly all patients. At the same time, the toxicity of this conditioning regimen has to be considered to be comparably low and the dreaded side effects of busulfan- or melphalan-based conditioning regimens such as VOD and pulmonary or cardiac toxicity have not been observed. Taking into account the low relapse rate even of patients in disease stages with high relapse risk, the conditioning with treosulfan and fludarabine may be considered a full-intensity but reduced-toxicity conditioning regimen.

On the basis of the experience with other autologous or allogeneic treosulfan-containing protocols,^11,12,42  a study to escalate the treosulfan dose to a total of 42 g/m² is being conducted to further increase the antineoplastic potential while maintaining the low toxicity profile. Considering the suitability of combining treosulfan with, for example, cyclophosphamide, melphalan, or etoposide,^12,42,43  this may ultimately lead to new reduced-toxicity but full-intensity conditioning regimens. From the results reported here, promising antineoplastic activity may be anticipated in myelodysplastic syndrome, AML, multiple myeloma, and NHL. In vitro/in vivo data for multiple myeloma, chronic myeloid leukemia, and acute myeloid leukemia support the potential activity of treosulfan in these indications also. This study has already shown that treosulfan and fludarabine is a favorable combination for conditioning with respect to toxicity, achievement of complete donor chimerism, and low GVHD rate and with a low treatment-related mortality and relapse rate. Consequently, the overall and disease-free survival rates are promising in a poor-risk patient population.

We would like to thank the staff members of the bone marrow transplantation (BMT) units for their excellent care of the patients, Mrs B. Sachse for documentation, and Dr J. Baumgart for his helpful discussion of the use of treosulfan.