Effect of Complete Remission on Survival in Patients With Acute Myelogenous Leukemia Receiving First Salvage Therapy

To the Editor:

There is controversy regarding whether patients with AML receiving initial reinduction therapy for relapse after a complete remission (CR) of less than 24 months should receive investigational single agents rather than proven therapies such as high-dose ara-C (HDAC)-containing regimens. This debate reflects uncertainty as to whether a second CR, presumably more likely following standard therapies, translates into longer survival. Empirical proof of the relationship between CR and survival in newly diagnosed patients has been provided by Freireich et al1 and, more recently, by Estey et al.2 However, while Keating et al3suggested a similar relationship existed in patients receiving “first salvage therapy” (first therapy for initial relapse or for failure of initial induction therapy) between 1974 and 1985, Keating et al’s3 comparison between the CR and no CR patients could be faulted because only the CR patients had to live a certain amount of time to be included, ie, the time needed to achieve CR. To address the relationship between CR and survival in patients receiving first salvage therapy, we analyzed data from 326 such patients seen here between 1990 and 1998. Patients with acute promyelocytic leukemia and patients given an allogeneic or autologous transplant either to induce CR or as postremission therapy were excluded. Eighty-five percent of the patients had received HDAC (≥1 g/m² per dose) at diagnosis or in first CR. First salvage therapy consisted of known active regimens in 76% (HDAC 55%, others 21%) and investigational single agents, often in a phase I trial, in 24%. In CR (standard definition), patients continued to receive their salvage induction regimen at reduced dose for 6 months or until recurrence, whichever came first. At recurrence, 64% of the patients received a second salvage attempt, which produced CR in 15%. Sixty-five percent of the patients who did not achieve a CR but survived their first salvage attempt received a second salvage attempt, which produced a CR in 2%. CR rates with first salvage therapy were higher in patients given the known active regimens than in patients given the investigational single agents, eg, 16 of 25 versus ½ in patients with first CRs exceeding 2 years, 20 of 43 versus 2 of 13 (P = .04) in patients with first CRs of 1 to 2 years and 21 of 181 versus 1 of 62 (P = .02) in patients with shorter first CRs or with primary refractory AML. We compared survival in patients who went into CR (n = 61) with survival in patients who did not but lived at least 10 weeks after start of first salvage (n = 188), by which time 97% of the patients who ultimately achieved CR were in CR. Only 3 of the patients in CR died before week 10. The log-rank P value for the comparison was <.001, suggesting a survival advantage from CR. This advantage was seen both in patients with first CR durations of 0 to 12 and 12 to 24 months, at least as judged by the Gehan test. This test is more robust for earlier parts of the survival curve than the log-rank test.4 Indeed, the difference between the CR and no CR groups is almost exclusively in the first 2 years. This can be seen in Table 1. The probability of survival at 24 months even among patients achieving a second CR after an initial CR of 0 to 24 months is <.2 (Table 1). However, patients achieving a second CR after an initial CR of 12 to 24 months had a .73 ± .10 probability of survival at 12 months versus a .27 ± .1 probability for similar patients whose disease failed to respond to first salvage therapy, whereas survival expectations were very similar for patients with first CR durations less than 1 year or who were primary refractory regardless of whether they achieved a CR with first salvage therapy (Table 1). Forty-one of the 48 patients who died after obtaining a CR with first salvage therapy had a recurrence before death. The other 7 died in CR. Actuarial median second remission duration was approximately 12, 28, and 30 weeks for patients with first CR durations of less than 12, 12 to 24, and greater than 24 months, respectively. Including only the 57 of the 61 second CR patients who received known active agents does not materially affect the results presented above.

Our results suggest that achieving a CR confers a survival benefit to patients with AML receiving first salvage therapy, although the data only permit analysis in patients with initial CR durations less than 2 years (Table 1). Furthermore, our data suggest that investigational single agents usually produce lower CR rates than accepted therapies such as HDAC. However, whether patients with AML recurrent after initial CR durations less than 2 years should always receive accepted therapies for first salvage depends on the value the reader places on the survival probabilities for CR patients shown in Table 1. In our opinion, provided that an allogeneic transplant is not an option, administration of investigational single agents is justified in patients with initial CR durations of 1 to 2 years and is indicated in patients with shorter first CRs including, of course, patients with primary refractory disease.