Phase I study of dexamethasone, methotrexate, ifosfamide, l‐asparaginase, and etoposide (SMILE) chemotherapy for advanced‐stage, relapsed or refractory extranodal natural killer (NK)/T‐cell lymphoma and leukemia

Abstract

Extranodal natural killer (NK)/T‐cell lymphoma, nasal type, and aggressive NK‐cell leukemia are rare, and their standard therapy has not been established. They are Epstein–Barr virus‐associated lymphoid malignancies, and tumor cells express P‐glycoprotein leading to multidrug resistance of the disease. Patients with stage IV, relapsed or refractory diseases have a dismal prognosis, with survival measured in months only. To develop an efficacious chemotherapeutic regimen, we conducted a dose‐escalation feasibility study of a new chemotherapeutic regimen, SMILE, comprising the steroid dexamethasone, methotrexate, ifosfamide, l‐asparaginase, and etoposide. The components of SMILE are multidrug resistance‐unrelated agents and etoposide. Etoposide shows both in vitro and in vivo efficacy for Epstein–Barr virus‐associated lymphoproliferative disorders. Eligible patients had newly diagnosed stage IV, relapsed or refractory diseases after first‐line chemotherapy, were 15–69 years of age, and had satisfactory performance scores (0–2). Four dose levels of methotrexate and etoposide were originally planned to be evaluated. At level 1, six patients with extranodal NK/T‐cell lymphoma, nasal type, were enrolled. Their disease status was newly diagnosed stage IV (n = 3), first relapse (n = 2), and primary refractory (n = 1). All of the first three patients developed dose‐limiting toxicities, and one of them died of sepsis with grade 4 neutropenia. A protocol revision stipulating early granulocyte colony‐stimulating factor administration was made. Two out of three additional patients developed dose‐limiting toxicities that were all manageable and transient. For the six enrolled patients, the overall response rate was 67% and the complete response rate was 50%. Although its safety and efficacy require further evaluation, we recommend a SMILE chemotherapy dose level of 1 for further clinical studies. (Cancer Sci 2008; 99: 1016–1020)

Extranodal natural killer (NK)/T‐cell lymphoma, nasal type (ENKL), and aggressive NK‐cell leukemia (ANKL) account for 3–8% of malignant lymphomas in East Asia.^{( 1} , ^{2 )} Both are Epstein–Barr virus (EBV)‐associated lymphoid malignancies.^{( 3} , ^{4 )} Neoplastic NK cells, similar to their normal counterparts, express high levels of P‐glycoprotein, leading to the concern that multidrug resistance (MDR) might be an obstacle to successful treatment with chemotherapy.^{( 2} , ⁵ , ^{6 )} More than two‐thirds of patients with ENKL present with localized disease.^{( 7} , ⁸ , ^{9 )} Recent studies suggest that first‐line local radiotherapy of at least 45 Gy is effective for these patients.^{( 10} , ¹¹ , ¹² , ^{13 )} Concurrent chemoradiotherapy has also been reported to be efficacious,^{( 10} , ^{12 )} as supported by results of a recent prospective study.^{( 14 )} In contrast, the treatment results of stage IV, relapsed or refractory ENKL, and ANKL with conventional chemotherapy are extremely poor.^{( 4} , ⁷ , ⁸ , ^{9 )} Long‐term survival after high‐dose chemotherapy and hematopoietic stem‐cell transplantation (HSCT) has been reported for a small number of patients with advanced‐stage, relapsed or refractory disease.^{( 15} , ^{16 )} Successful disease control, an important prerequisite to HSCT, is however rarely achieved in most patients with relapsed or refractory diseases. Therefore, the development of an effective chemotherapy regimen for these patients is an important initial step in improving the treatment outcome.

To address this issue, we have formulated a new chemotherapeutic regimen comprising the steroid dexamethasone, methotrexate, ifosfamide, l‐asparaginase, and etoposide (SMILE). The design of the SMILE regimen was based on several considerations. Etoposide has demonstrated in vitro and in vivo efficacy for NK‐cell neoplasms,^{( 17} , ^{18 )} being effective for pediatric EBV‐related hemophagocytic syndrome,^{( 19 )} and pediatric EBV‐associated lymphoproliferative disease.^{( 20 )} l‐Asparaginase induces the selective apoptosis of NK lymphoma cells in vitro ^{( 21 )} Indeed, successful therapeutic results in NK‐cell lymphoma have been reported for l‐asparaginase, either alone,^{( 22 )} or in combination with other chemotherapy.^{( 23 )} Dexamethasone is better than prednisolone in ameliorating the adverse drug reactions of l‐asparaginase.^{( 24 )} Methotrexate and ifosfamide are unaffected by the MDR phenotype, and are components of regimens reported to be effective in NK/T‐cell lymphomas.^{( 10} , ¹⁸ , ^{25 )} Methotrexate was scheduled on day 1 to precede the other drugs because there is a possibility of it showing antagonistic effects on administration with etoposide and ifosfamide,^{( 26 )} but synergic effects when preceding etoposide.^{( 27 )} The other three drugs were scheduled for days 2–4 because the simultaneous use of etoposide and ifosfamide might lead to additive effects.^{( 26 )}

Because advanced‐stage ENKL and ANKL are rare and aggressive, a prospective therapeutic trial for these diseases is difficult to conduct. To overcome this problem, we designed a multicenter cooperative phase I study, the first of its kind, in East Asia where the incidence of ENKL and ANKL is higher than in other parts of the world. In the present report, we describe the results of a dose‐escalation feasibility study of SMILE in newly diagnosed stage IV, relapsed or refractory ENKL and ANKL.

Materials and Methods

Patient selection.  Patients of 15–69 years of age with ENKL or ANKL diagnosed according to the World Health Organization (WHO) classification,^{( 28 )} who were newly diagnosed with Ann Arbor stage IV disease, first relapsed or recurrent disease after remission, or refractory disease after first‐line chemotherapy, were eligible for the SMILE phase I study. Neither chemotherapy nor radiotherapy was given within 21 days before registration. Additional entry requirements included an Eastern Cooperative Oncology Group performance status of 0–2, at least one evaluable lesion, and laboratory parameters obtained within 7 days before registration within the following ranges: white blood cells (WBC)  3000/mm³, absolute neutrophil count ≥ 1200/mm³, platelet count ≥ 7.5 × 10⁴/mm³ (for patients with bone marrow involvement or hemophagocytosis, platelet count must be ≥ 5.0 × 10⁴/mm³), aspartate aminotransferase (AST) £ upper normal limit × 5, alanine aminotransferase (ALT) = upper normal limit × 5, total bilirubin £ 2.0 mg/dL, serum creatinine £ 1.5 mg/dL, left ventricular ejection fraction ≥ 50%, arterial blood gas ≥ 65 mmHg or O₂ saturation ≥ 90% (under room air). Patients with no ischemic change, atrial fibrillation, or ventricular arrhythmia requiring treatment carried out within 21 days were eligible. Patients who received corticosteroids alone were eligible for this study, but those under treatment had to discontinue it before registration. Patients who had a history of HSCT, only had cutaneous lesions, or clinical symptoms of central nervous system involvement were excluded.

Registration was conducted by facsimile from participating physicians to the regional Study Coordinators (M. Y. within and R. S. outside Japan). The protocol was approved by the Protocol Committee and the institutional review board at each participating institute. All patients gave written informed consent.

Study design.  The study was designed as a phase I dose‐escalation study conducted by the NK‐cell Tumor Study Group in Japan, and collaborative institutes in Hong Kong, Korea, and Taiwan. The primary endpoint was the maximum tolerated dose (MTD) of SMILE, and the secondary endpoints were the overall response rate (ORR) and complete response (CR) rate. Considering that the study was the first prospective multicenter trial for ENKL and ANKL in East Asia, the study was not designed as a phase I/II study. The protocol stipulated that a subsequent phase II study to examine the efficacy of SMILE chemotherapy would be projected after the resolution of the recommended dose.

The National Cancer Institute Common Toxicity Criteria 2.0 were used for safety evaluation. A standard 3 + 3 design was used to evaluate dose‐limiting toxicities (DLT). Four dose levels of methotrexate and etoposide were planned to be evaluated. DLT included grade 4 hematologic toxicities lasting 7 days or more; any non‐hematological toxicity of grade 3 or more except for nausea, vomiting, stomatitis, hypofibrinogenemia, and hyperglycemia; more than 28 days delay of the second course of SMILE; and patient refusal. Treatment efficacy was evaluated according to the WHO response criteria.^{( 29 )} Three to six patients were enrolled in each level. When all of the protocol treatments of the first three patients for each level were completed, registration was held, and all adverse events observed in the three patients were evaluated according to the criteria for DLT. When the initial three cases in level 1 developed DLT, the protocol committee reconsidered the continuation of this study. When one or two of the three patients in each level developed DLT, an additional three patients were enrolled at the same level. If three of the six patients developed DLT, the protocol committee reconsidered the continuation of this study. When the numbers of patients who developed DLT was two or lower in the six patients, the next cohort of patients was treated at the next level. When more than two patients in a cohort of three or six patients experienced DLT, no additional patients were enrolled and dose escalation ceased. When none of the three patients developed DLT, study registration was started with the next level. When all of the first three patients developed DLT at level n + 1, the MTD was determined as level n. When none of the three patients in level 4 developed DLT, the MTD was determined as level 4. When treatment‐related death occurred, registration was stopped when the severe adverse event report was submitted. The protocol committee discussed the continuation of the study, and their decision was reviewed by the Data and Safety Monitoring Committee.

Treatment.  The drug doses of level 1 and the administration schedule were as follows: dexamethasone, 40 mg/body intravenously on days 2–4; methotrexate, 2 g/m² intravenously over 6 h on day 1; ifosfamide, 1.5 g/m² intravenously on days 2–4; Escherichia coli l‐asparaginase (Leunase; Kyowa Hakko Kogyo, Tokyo, Japan), 6000 U/m² intravenously on days 8, 10, 12, 14, 16, 18, and 20; and etoposide, 100 mg/m² intravenously on days 2–4. Doses of methotrexate and etoposide were scheduled to be escalated to 2 g/m² and 150 mg/m² in level 2, 3 g/m² and 150 mg/m² in level 3, and 3 g/m² and 200 mg/m² in level 4. The second course was started from day 29 of the first course. Leucovorin was begun 30 h after the initiation of methotrexate. Mesna was given at 300 mg/m² simultaneously with ifosfamide, and at 4 and 8 h afterwards. Granulocyte colony‐stimulation factor (G‐CSF) was initiated if the WBC count decreased to less than 2000/mm³, and was discontinued if the WBC count exceeded 5000/mm³. If l‐asparaginase‐induced grade 1–2 allergic reactions or hypersensitivity were observed, the dose of l‐asparaginase was reduced by half. When l‐asparaginase was discontinued due to grade 4 thrombocytopenia or grade 3–4 non‐hematological toxicity in the first course, it could be resumed in the second course if the patient had recovered. Two courses of SMILE were planned.

Results

Patients.  Patient registration for the SMILE phase I study was started in July 2005. A total of seven patients were registered. All patients had ENKL, so that no patient with ANKL had been registered. The first patient enrolled (patient #01) was ineligible because of thrombocytopenia. We decided to exclude this patient from further evaluation, and our decision was approved by the Data and Safety Monitoring Committee. For the six eligible patients, there were five men and one woman, at a median age of 48 years (range 28–69 years). The disease status was newly diagnosed stage IV (n = 3), first relapse (n = 2), and primary refractory (n = 1). Serum lactate dehydrogenase was elevated in four patients, and performance status was higher than one in one patient (Table 1). EBV was identified in tumor cells by in situ hybridization of all six eligible patients. CD56 was positive in five patients. The other patient (#07) was CD56‐negative and cytotoxic molecule‐positive, hence fitting the diagnostic criteria of ENKL according to the WHO classification.^{( 28 )}

Table 1.

Patient characteristics ( n  = 6)

Patient no.	#02	#03	#04	#05	#06	#07
Age (years)	63	39	28	57	33	69
Sex	M	M	M	M	M	F
Disease state	Refractory	First relapse	Newly diagnosed	Newly diagnosed	Newly diagnosed	First relapse
Stage at diagnosis	IVB	IEA	IVB	IVB	IVB	IEA
Sites of involvement at diagnosis	Bone marrow, spleen	Nasal cavity	Waldeyer ring, lymph nodes, nasal cavity, bone marrow	Bilateral adrenal glands, pancreas, gallbladder	Nasopharynx, lymph nodes, small bowel	Nasal cavity
Initial treatment	CHOP × 2	CHOP × 2→RT →IMVP‐16	–	–	–	RT (42 Gy) →DeVIC × 4
Sites of involvement at registration	Bone marrow, spleen	Nasal cavity	Waldeyer ring, lymph nodes, nasal cavity, Bone marrow	Bilateral adrenal glands, pancreas, gallbladder	Nasopharynx, lymph nodes, small bowel	Nasal cavity, paranasal sinuses, cheek
PS	1	1	1	2	0	0
sLDH level	Elevated	Below the upper normal range	Elevated	Elevated	Below the upper normal range	Elevated

CHOP, cyclophosphamide, doxorubicin, vincristine, prednisolone; DeVIC, dexamethasone, etoposide, ifosfamide, carboplatin; IMVP‐16, ifosfamide, methotrexate, etoposide; PS, performance status; RT, radiotherapy; sLDH, serum lactate dehydrogenase.

Evaluation of DLT.  All of the three initial eligible patients in level 1 developed DLT. Of these, one patient (#02) experienced grade 5 infection accompanied by grade 4 neutropenia. For this patient, the initiation of G‐CSF was delayed (from day 15 with a WBC count of 100/mm³), which was evident by case report‐form monitoring and it was considered as a protocol violation. Other DLT in this patient were grade 4 leukopenia and grade 4 neutropenia lasting 7 days, febrile neutropenia, grade 3 AST elevation, and grade 3 ALT elevation. Another patient (#03) developed grade 3 hyponatremia (Na 129 mEq/L) of 1‐day duration alone. The DLT developing in the remaining patient (#04) were grade 4 leukopenia and grade 4 neutropenia lasting 7 days, and febrile neutropenia. We subsequently made a protocol revision stipulating mandatory initiation of G‐CSF from day 6 and cessation of l‐asparaginase if grade 4 thrombocytopenia or grade 3 or more non‐hematological toxicity developed during its administration. These revisions were approved by the Data and Safety Monitoring Committee. After these revisions, three additional patients were registered until October 2006. Of these patients, two developed DLT. One patient (#05) developed grade 3 hyponatremia and grade 3 activated partial thromboplastin time (APTT) prolongation. Another patient (#06) experienced grade 3 hyponatremia. All DLT that developed in these two patients were manageable and transient. According to the criteria for assessment of DLT, dose escalation to level 2 was not done.

Safety.  All six evaluable patients developed grade 3 or 4 leukopenia and grade 4 neutropenia (Table 2). Grade 3 non‐hematological toxicities associated with protocol treatment included hyponatremia (n = 3), febrile neutropenia (n = 2), APTT prolongation (n = 1), hypofibrinogenemia (n = 1), nausea (n = 1), AST elevation (n = 1), ALT elevation (n = 1), and hyperglycemia (n = 1) (Table 2). For the last three patients who were registered after protocol revision, the hematological toxicity was less severe, and no grade 4 non‐hematological toxicity was encountered. Grade 3 non‐hematological toxicities consisted of two for hyponatremia and one each for APTT prolongation and hyperglycemia; all resolved rapidly and were manageable. Dose modification was needed only in l‐asparaginase, and the maximum delay of the second course of SMILE was 7 days (Table 3). No severe adverse events such as allergy or thrombosis were observed.

Table 2.

Grade 3 and 4 toxicity profile ( n  = 6)

Adverse event	First three patients (#02–#04)		Additional three patients (#05–#07)
	Grade 3	Grade 4	Grade 3	Grade 4
Leukopenia	0	3 (2)	3	0
Neutropenia	0	3 (2)	0	3 (0)
Anemia	3	0	0	0
Thrombocytopenia	1	1 (1)	0	0
RBC transfusion	3	0	0	0
PLT transfusion	2	0	1	0
Febrile neutropenia	2	0	0	0
Infection with grade 3 or 4 neutropenia	0	1 †	0	0
Nausea	0	0	1	0
AST elevation	1	0	0	0
ALT elevation	1	0	0	0
Hypofibrinogenemia	1	0	0	0
Prolonged APTT	0	0	1	0
Hyperglycemia	0	0	1	0
Hyponatremia	1	0	2	0

Values indicate patient numbers with each adverse events. Values in parentheses show those with grade 4 hematological toxicities lasting 7 days or more.

^† Died of sepsis (grade 5 infection).

ALT, alanine aminotransferase; APTT, activated partial thromboplastin time; AST, aspartate aminotransferase; PLT, platelet; RBC, red blood cell.

Table 3.

Treatment profile and efficacy ( n  = 6)

Patient no.	#02	#03	#04	#05	#06	#07
Dose reduction of SMILE (course)	l‐Asp: 7→4 times (2nd)	l‐Asp: reduced at a dose of 50% (2nd)	–	l‐Asp: 7→4 times (1st)	–	–
Delay of the second course of SMILE	–	7 days	–	–	2 days	2 days
Planned treatment	Terminated	Completed	Terminated	Completed	Completed	Completed
Overall response	NE	CR	NR	PR	CR	CR
Additional treatment	–	SMILE × 1, SMI(L)E × 1 for PBSCH, HD‐auto PBSCT	DeVIC × 1	SMILE × 1, HD‐ETP for PBSCH, HD‐auto PBSCT	MILD × 2, HD‐ETP for PBSCH, MILD × 1, HD‐auto PBSCT	SMILE × 4 (from 4th course, E. coli l‐Asp was switched to Erwinia l‐Asp)
Outcome	TRD, 2M	AND, 15M	DOD, 3M	AWD, 7M	AND, 7M	AND, 7M

AND, alive with no evidence of disease; AWD, alive with disease; CR, complete response; DeVIC, dexamethasone, etoposide, ifosfamide, carboplatin; DOD, died of disease; ETP, etoposide; HD, high dose; M, months after registration; MILD, methotrexate, ifosfamide, l‐asparaginase, dexamethasone; NE, not evaluable; NR, no response; PBSCH, peripheral blood stem cell harvest; PBSCT, peripheral blood stem cell transplantation; PR, partial response; SMI(L)E, dexamethasone, methotrexate, ifosfamide, etoposide; SMILE, dexamethasone, methotrexate, ifosfamide, l‐asparaginase, etoposide; TRD, treatment‐related death.

Efficacy.  The efficacy of treatment is shown in Table 3. One patient died of infection and could not be evaluated. The responses were CR in three patients, partial response in one patient, and no response in one patient, giving a CR rate of 50% and an ORR of 67%. Three patients were treated with additional SMILE chemotherapy followed by high‐dose chemotherapy and autologous HSCT (Table 3). In patient #03, SMILE chemotherapy was terminated at the end of the first course because of prolonged pancytopenia and hypofibrinogenemia. He was treated with additional chemotherapy but died of disease 3 months later. Patient #07 was treated with six courses of SMILE. In this patient, allergic reaction to E. coli l‐asparaginase developed in the fourth course of SMILE. After that, Erwinia l‐asparaginase was used.

Discussion

In the present study, we have attempted to tackle several obstacles in the effective treatment of advanced or relapsed NK/T‐cell lymphoma and leukemia. ENKL and ANKL respond poorly to conventional chemotherapy designed for B‐cell lymphomas.^{( 30 )} Furthermore, the addition of anthracyclines, which are important in B‐cell lymphoma regimens, has not been shown to increase therapeutic efficacy in NK‐cell malignancies.^{( 30 )} Finally, the role of high‐dose chemotherapy with HSCT remains controversial for NK‐cell lymphomas because the ideal conditioning regimen for HSCT is currently unclear.^{( 30 )} Therefore, the definition of a chemotherapeutic regimen with high treatment efficacy is an important goal. The SMILE regimen reported herein is currently the only regimen formulated specifically for NK/T‐cell lymphomas.

The most common adverse event during SMILE chemotherapy was grade 4 neutropenia. At level 1, one treatment‐related death occurred due to a delay in G‐CSF administration. However, after the protocol revision that mandated G‐CSF administration from day 6, severe infection was not observed. Three out of six patients were able to undergo high‐dose chemotherapy with autologous HSCT. Based on these results and the fact that only six patients were evaluated in this trial, we considered that the hematological toxicities and severe infectious complications of SMILE in level 1 with G‐CSF support should be evaluated further in the setting of a prospective clinical trial.

The most frequent non‐hematological adverse event was hyponatremia. Hyponatremia might be related to the syndrome of inappropriate secretion of antidiuretic hormone or renal tubular damage, which was likely caused by chemotherapeutic agents in the SMILE protocol, particularly ifosfamide. Interestingly, hyponatremia was also observed in a recent clinical trial for localized ENKL.^{( 14 )} Therefore, other disease‐specific causes of hyponatremia would have to be examined in future studies.

The antitumor effect of two courses of SMILE, which resulted in a CR rate of 50% and ORR of 67%, was remarkable. To date, two promising results of nonanthracycline‐containing regimens in nasal ENKL have been reported. In a Mexican prospective study of newly diagnosed advanced‐stage ENKL with nasal involvement,^{( 25 )} patients were given six courses of cyclophosphamide, methotrexate, etoposide, and dexamethasone chemotherapy, with sandwiched radiotherapy of 55 Gy after three courses in cases with facial involvement. Another Chinese study of nasal ENKL relapsing or refractory to anthracycline‐containing chemotherapy used a salvage regimen containing l‐asparaginase, vincristine, and dexamethasone followed by involved‐field radiotherapy (median: 56 Gy).^{( 23 )} Although the CR rates in these studies exceeded 55%, high‐dose radiotherapy was used, making evaluation of the contribution of the efficacy of chemotherapeutic regimens difficult. Moreover, high‐dose radiotherapy may not be an option for patients with advanced disseminated diseases, and for relapsed patients who have already received involved‐field radiotherapy during primary treatment for nasal ENKL. The therapeutic efficacy of the SMILE protocol requires further evaluation in a larger number of patients with advanced ENKL.

We did not use the International Working Group (IWC) criteria to assess the response in this trial for the following reasons:^{( 31 )} (1) ENKL and ANKL usually show extranodal involvement that is often difficult to measure bidimensionally; and (2) given that the SMILE phase I study is the first prospective, international, multicenter‐based clinical trial for ENKL and ANKL in East Asia, simple and familiar response criteria were thought to be more appropriate. Recently, the IWC criteria have been revised, and 18‐fluoro‐deoxyglucose (FDG) positron emission tomography scanning was incorporated into the evaluation of lymphomatous involvement.^{( 32 )} Moreover, a recent report suggested that ENKL is an FDG‐avid lymphoma.^{( 33 )} In future clinical trials on ENKL, the appropriateness of the revised IWC criteria for ENKL needs to be examined.

According to the criteria for DLT assessment stipulated in the protocol, we did not escalate the dose to level 2 and could not determine the MTD in this trial. However, all grade 3 non‐hematological toxicities developing in patients who were enrolled after the protocol revision in level 1 were manageable and transient. In addition, no grade 4 hematological toxicities lasting 7 days or more were observed in patients who were enrolled after the protocol revision (Table 2). SMILE at dose level 1 was thought to be promising because three out of six patients in level 1 achieved a CR. From these results, we believe that dose level 1 of SMILE chemotherapy is appropriate for further clinical studies. Our findings are now further evaluated in a prospective phase II study of level 1 SMILE with G‐CSF support, which has started since July 2007.