The value of event-free survival as a clinical endpoint for AML

The most valued indicators of clinical benefit during trials for acute myeloid leukemia (AML) are overall survival (OS) and complete remission (CR) rates.^1,2 These are frequently used as primary endpoints in AML trials but do hold caveats in their interpretation. When used as single endpoints, they may lead to data misinterpretation or hinder statistical significance.¹ Thus, the concomitant use of good surrogate indicators is key to correct clinical benefit evaluation. Event-free survival (EFS) is a commonly reported outcome that has not yet been thoroughly evaluated for its effectiveness to indicate clinical benefit in AML. This has led to its frequent consideration as a poor surrogate for OS.²

For this reason, Abhishek Maiti and colleagues published in Blood Advances an in-depth retrospective investigation of the clinical value of EFS in AML trials.² The authors hypothesized that improvement in EFS may correlate with a reduction in the use of healthcare for AML treatment.

Study design²

• Retrospective study with medical review of adult patients with newly diagnosed AML who started frontline AML treatment at The University of Texas MD Anderson Cancer Center, Texas, US, between 2003–2013
• Patients with OS from 2–36 months were included
• EFS was defined as the time from study start until the time of primary refractory disease confirmation, relapse, or death
• Patients included in the study must have had an EFS ≥ 2 months, an adverse event, and suffered death or progressed to stem cell transplantation (SCT) by the time of data collection:
  • The 2-month EFS cutoff was used, as patients usually needed two cycles of therapy to achieve response
  • EFS was censored at the time of SCT, as SCT incurs a great need for healthcare and would interfere with study outcomes
• A review of medical records was used to determine healthcare use from AML diagnosis until death or SCT
• Healthcare usage was defined as clinic visits, emergency room (ER) visits, hospitalizations, inpatient or outpatient consultations from other services, blood transfusions, laboratory studies, imaging studies, and invasive procedures
• Patients were stratified for AML risk based on the 2017 European LeukemiaNet guidelines³

Results

• In total, 400 patients were retrospectively assessed. The baseline patient characteristics are shown in Table 1

Table 1. Baseline patient characteristics²

AML, acute myeloid leukemia; ECOG PS, European Cooperative Oncology Group performance status; FTL3, FMS-like tyrosine kinase 3; IDH, isocitrate dehydrogenase; ITD, internal tandem duplication; TKD, tyrosine kinase domain *AML risk stratification was based on the 2017 European LeukemiaNet guidelines
Baseline characteristic	All patients (N = 400)
AML risk*, % Favorable Intermediate Adverse	7 58 35
Median age, years (range)	65 (55–72)
Male patients, %	56
ECOG PS, % 0–1 ≥ 2 Unknown	84 10 6
Mutations, % FTL3 ITD/TKD IDH1/2	19 11
First-line treatment, % Intensive chemotherapy only + non-chemotherapy agents Hypomethylating agents only + chemotherapy + non-chemotherapy agents Non-intensive chemotherapy only +/or non-chemotherapy agents	29 19 15 2 11 21 5

• The response rates achieved are shown in Table 2, with a total of 87% of patients achieving any type of response and 84% of patients eventually relapsing

Table 2. Response outcomes of patients included in this retrospective study²

CR, complete response; CRi, CR with incomplete hematological response; EFS, event-free survival; MLFS, morphologic leukemia-free status; ORR, overall response rate; PR, partial response; SCT, stem cell transplantation
Response outcome	All patients (N = 400)
ORR, % CR CRi/MLFS PR	84 72 11 1
Relapse, %	83
Median months of EFS censored at SCT	9.7
Median months of OS censored at SCT	17.6

• Linear regression analyses showed a statistically significant correlation between increasing EFS and decreasing overall use of healthcare per month of OS (r = −0.45; 95% CI, −0.53 to −0.37; p < 0.0001)
• • Similar results were observed when breaking down overall healthcare to the following subgroups:
    • Clinic visits + ER visits + hospitalizations + consolations per OS month (r = −0.51; p < 0.001)
    • Invasive procedures + laboratory and imaging studies per OS month (r = −0.60; p < 0.001)
    • Blood transfusions per OS month (r = −0.20; p < 0.001)
    • Hospitalizations per OS months (r = −0.36; p < 0.001)
    • ER visits per OS month (r = −0.17; p = 0.003)
• Patients who achieved a CR had a steeper decline in the use of healthcare with prolonged EFS (r from −0.17 to −0.60)
• Similarly, a decrease in overall healthcare use with increasing EFS was observed for
  • patients with intermediate- or adverse-risk AML (r = −0.40; p < 0.001 and r = −0.55; p < 0.001, respectively)
  • patients aged < 65 or ≥ 65 (r = −0.43; p < 0.001 and r = −0.41; p < 0.001, respectively)
  • patients receiving intensive chemotherapy or hypomethylating agents with or without other chemotherapy agents (r = −0.47; p < 0.001 and r = −0.47; p < 0.001, respectively)
• Irrespective of OS
  • a nonlinear decline in overall healthcare use was also observed with increasing EFS per month of EFS (r = −0.40; p < 0.001)
  • the use of healthcare per month of EFS was higher in patients with shorter EFS duration (1–6 or 6–12 months) when compared to those with longer EFS duration (> 12 months; p < 0.0001)

Conclusion

These retrospective results indicate that EFS correlates well with healthcare use in patients with newly diagnosed AML, both when combined and irrespective of OS. Monthly EFS improvement correlated significantly with a decline in the use of healthcare. This decline was even more marked for patients achieving CR but was seen across all AML risk, age, and treatment subgroups. Despite the retrospective nature of this study, the results do highlight the potential effectiveness of EFS as an independent and OS-surrogate endpoint for healthcare use in patients with newly diagnosed AML.