Impact of chemotherapy delay on short- and long-term survival in younger and older AML patients: a Danish population-based cohort study

Article metrics

In recent decades, new strategies and molecularly targeted approaches have been developed to improve treatment in acute myeloid leukemia (AML) patients. To direct therapy, results of conventional cytogenetic and molecular mutation analysis are required.1 Results can be ready in 72 h, but are generally not available until after 5 days. A relevant question is whether the outcome is affected if intensive chemotherapy awaits results. Only two studies exist on this topic. In an observational US study of 1317 AML patients, delaying treatment beyond 5 days resulted in inferior complete remission (CR) rates and overall survival in younger patients, but not in patients 60 years.2 As a consequence, LeukemiaNet recommends treatment initiation before day 5 in younger patients.3 In a French single-center study of 599 patients, no association was found of treatment delays above 5 days compared with patients with a time to treatment initiation (TTT) below 5 days in terms of early death, CR rates and long-term survival.4 The two studies have important limitations, and the results may not be generalizable to other patients.5

We designed a national population-based cohort study to investigate the effect of TTT on CR rates and overall survival (30-day and 31-day–3-yearsurvival) in AML patients treated with curative intent in Denmark during 2000–2012. We used the Danish National Acute Leukemia Registry to identify all non-promyelocytic AML patients. The registry covers all AML patients diagnosed in Denmark, and it prospectively collects high-quality demographic, clinical, treatment and outcome data.6 Information on comorbidity was obtained through the Danish National Registry of Patients.7 Information on all-cause mortality was obtained from the Civil Registration System.8

We defined date of diagnosis as the date of first diagnostic bone marrow examination and TTT as the time from diagnosis until the initiation of intensive chemotherapy. Curative treatment intent was defined as allocation to a multiagent remission induction chemotherapy. Information on exposure was grouped into five categories. Patients treated on the day of diagnosis were expected to differ from patients treated later and were kept in a separate category. Factors associated with TTT 5 days and the chance of achieving CR according to TTT were assessed using logistic regression. Overall survival was calculated as the time from diagnosis until death or end of follow-up (18 July 2013). We used Cox proportional hazards regression analysis and restricted cubic splines to assess the influence of TTT on survival.9 We analyzed data at the time of initiated chemotherapy in an intention-to-treat manner and used the date of chemotherapy as a time-dependent covariate.10 All survival analyses were stratified by age (<60 years or 60 years), presence of AML secondary to myelodysplastic syndrome or myeloproliferative neoplasias (sAML) or treatment-related AML (tAML) and white blood cells (WBC) (<30 × 109/l and 30 × 109/l). All estimates were adjusted for potential prognostic factors, provided with corresponding 95% confidence intervals (CIs) and with TTT=3–4 days as a reference (detailed in Supplementary Methods).

The study population consisted of 1388 patients (74 patients with TTT >28 days were excluded). Eleven patients (0.8%) died before treatment initiation. The characteristics of the patients included in the study by TTT and age are presented in Supplementary Table S1. The median TTT was 3 days (interquartile range (IQR) 1–7) and 90% were treated within 13 days. The median TTT was shorter in young patients (3 days (IQR 1–6) vs 4 days (IQR 2–8) in patients 60 years). The median follow-up for all patients was 500 days (range 0–4915). The median TTT did not vary by calendar year or weekday, by inclusion into clinical trials or allogeneic transplantation status, but varied from 2.5 to 5 days between departments. Male gender, presence of comorbidity, sAML/tAML and low WBC predicted longer TTT in younger patients. In patients 60 years, a TTT >5 days was associated with a low WBC and the presence of sAML/tAML (Supplementary Table S2). In multivariate analysis, TTT >5 days was not adversely related to CR (Supplementary Table S3).

Crude estimates of overall survival according to TTT are shown in Supplementary Figure S1. Table 1 presents crude and adjusted 0–30-day and 31-day-to-3-year mortality rates (MRs), overall, stratified by age and sAML/tAML. In younger patients and patients 60 years, the adjusted 0–30-day MR did not increase with longer TTT. The association between TTT as a continuous variable and 31-day-to-3-year mortality is described in Figure 1. The adjusted 31-day-to-3-year MR for all patients combined and for patients 60 years showed an almost U-shaped relation with the lowest MR found in patients with TTT=3–4 days. In younger patients, only a TTT 10 days was associated with an increased MR (1.57, 95% CI (1.05–2.35)). In patients 60 years, we found a clear association between longer TTT and inferior survival beyond 5 days (adjusted MRs were 1.50, 95% CI (1.07–2.12) and 2.03, 95% CI (1.41–2.92) for TTT=5–9 days and 9 days, respectively). In de novo patients, TTT 10 days was associated with a worse prognosis. In older sAML/tAML patients, a strong association was seen for TTT beyond 5 days (adjusted MR 3.44, 95% CI (1.35–8.80) for TTT=5–9 days and 4.93, 95% CI (1.83–13.30) for TTT 10 days), whereas in younger sAML/tAML patients the MR was only substantially increased beyond day 10 (adjusted MR 2.58, 95% CI (0.81–8.21)). We tested the robustness of the study conclusions through sensitivity analyses (see Supplementary Results). These analyses did not change the interpretation of results.

Table 1 Association between TTT and survival in 1388 AML patients, overall results, and stratified by age and type of leukemia
Figure 1

Estimated 30-day-to-3-year MRs according to time from diagnosis to treatment (TTT). (a) Crude and (b) adjusted MR for younger patients, and (c) crude and (d) adjusted MR for patients 60 years. To obtain a graphical picture of the shape of the relation between TTT as a continuous covariate and overall survival, we used restricted cubic splines to smoothen the TTT mortality with linear restrictions imposed on both tails. It shows MRs (crude and adjusted 31-day-to-3-year MRs) stratified by age with 95% CI bands according to TTT as a continuous exposure. The four knots are placed at the 20th, 40th, 60th and 80th percentiles. The adjusted 30-day-to-3-year MRs reflect the results of Table X and show an almost U-shaped relation with the lowest MRs found in patients with TTT=3–4 days in both age groups. In patients 60 years, delaying chemotherapy had a significant impact on mortality for shorter TTTs than in younger patients (<60 years) (data not split at the day of initiated chemotherapy).

The null association between TTT and the CR rate in our study is consistent with results from the French study,4 whereas the US study2 found a negative association between longer TTT and CR rates. In line with the US study, we found an association between longer TTT and increased mortality in younger patients. However, in contrast to the US study, our results only showed increased mortality for those who initiated treatment 10 days. Also contrary to their results, we found an increased mortality in older patients from TTT 5 days. The definition of TTT may partly explain the different results for younger patients compared with the US study. The median TTT in our study was only 3 days compared with 4 days in the US study and 8 days in the French study. The US study used a different definition from our study and the French study; the authors defined the date of diagnosis as the date of confirmed diagnosis at the treating institution, which, per definition, will give a shorter TTT.

Our study population differed from the US study, in which the median age was higher, more patients had an adverse cytogenetic risk profile and 45% of all patients had sAML/tAML. This may reflect referral bias. We found that longer TTT was associated with a worse outcome within the sAML/tAML group than within de novo AML patients. Although the authors of the US study adjusted for sAML, residual confounding may explain their findings. Also, the sex distribution was not described or controlled for. We found that men had a significantly longer TTT, which could have acted as a confounder. Both prior studies excluded patients who died before the initiation of chemotherapy, which could lead to a spurious survival advantage for patients with longer TTT.11, 12 Older patients had a significantly longer TTT, and immortal person-time bias could explain the finding of no association in the French study and in older patients in the US study. Our results of the 31-day-to-3-year MRs are in sharp contrast to the findings of the French study. Their conclusion of no association was based on TTT as a dichotomous variable (5 days and >5 days). Our study demonstrated an almost U-shaped association of TTT and mortality, and their inclusion of patients with TTT up to 90 days, WBC >50 and TTT=0 probably amplified the immortal person-time bias for patients with the longest TTTs. This could explain the decrease in MR for patients with TTT >20 days.

In our study, the truly population-based design with virtually complete follow-up reduced selection bias. The coverage of the registry is almost 100% and the positive predictive values (PPVs) of variables that define TTT are high (‘date of diagnosis’ PPV=90% and ‘date of chemotherapy initiation’ PPV=100%).6 The study population was large, treatment strategies homogeneous and results were adjusted for possible confounders and stratified by age, WBC and sAML/tAML to minimize confounding by indication. The proportion of patients being treated with curative intent and the CR rates are comparable to those of other population-based studies.13, 14 Also, the study was designed and analyzed to avoid immortal person-time bias. Our study also has limitations. As in the US study, we did not obtain information on the reasons for delaying treatment. Although the data were adjusted for most factors known to affect the outcome, we could not fully control for early complications such as infections, bleeding and organ failure, which could have delayed treatment initiation.

In conclusion, we found that TTT affects the prognosis in both younger and older AML patients. Our results corroborate that intensive chemotherapy should be initiated as early as possible. In patients older than 60 years and especially in patients with sAML/tAML, delaying intensive treatment beyond day 5 seemed to have an adverse impact on survival. In these patients, cytogenetic analyses should be accelerated. In younger patients and older de novo patients, treatment can probably be delayed for up to 10 days without affecting the prognosis.


  1. 1

    Burnett A, Wetzler M, Lowenberg B . Therapeutic advances in acute myeloid leukemia. J Clin Oncol 2011; 29: 487–494.

  2. 2

    Sekeres MA . Treatment of older adults with acute myeloid leukemia: state of the art and current perspectives. Haematologica 2008; 93: 1769–1772.

  3. 3

    Dohner H, Estey EH, Amadori S, Appelbaum FR, Buchner T, Burnett AK et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood 2010; 115: 453–474.

  4. 4

    Bertoli S, Berard E, Huguet F, Huynh A, Tavitian S, Vergez F et al. Time from diagnosis to intensive chemotherapy initiation does not adversely impact the outcome of patients with acute myeloid leukemia. Blood 2013; 121: 2618–2626.

  5. 5

    van Walraven C, Davis D, Forster AJ, Wells GA . Time-dependent bias was common in survival analyses published in leading clinical journals. J Clin Epidemiol 2004; 57: 672–682.

  6. 6

    Østgård L, Nørgaard J, Severinsen M, Friis L, Jensen M, Nielsen O et al. Data quality in the Danish National Acute Leukemia Registry: a hematological data resource. Clin Epidemiol 2013; 5: 335–344.

  7. 7

    Østgård L, Nørgaard J, Sengeløv H, Severinsen M, Friis L, Marcher C et alComorbidity and performance status in acute myeloid leukemia patients: a nation-wide population-based cohort study, March 2014 (under submission).

  8. 8

    Frank L . Epidemiology. When an entire country is a cohort. Science 2000; 287: 2398–2399.

  9. 9

    Greenland S . Dose-response and trend analysis in epidemiology: alternatives to categorical analysis. Epidemiology 1995; 6: 356–365.

  10. 10

    Fisher LD, Lin DY . Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health 1999; 20: 145–157.

  11. 11

    Rothmann KJ, Greenland S . Cohort Studies. In: Rothman KJ, Greenland S, Lash TL (eds) Modern Epidemiology 3rd edn Lippincott Williams & Wilkins: Philadelphia, PA, USA, 2008; pp 100–110.

  12. 12

    Levesque LE, Hanley JA, Kezouh A, Suissa S . Problem of immortal time bias in cohort studies: example using statins for preventing progression of diabetes. BMJ 2010; 340: b5087.

  13. 13

    Juliusson G . Older patients with acute myeloid leukemia benefit from intensive chemotherapy: an update from the Swedish Acute Leukemia Registry. Clin Lymphoma Myeloma Leuk 2011; 11 (Suppl 1): S54–S59.

  14. 14

    Gangatharan SA, Grove CS, P'ng S, O'Reilly J, Joske D, Leahy MF et al. Acute myeloid leukaemia in Western Australia 1991–2005: a retrospective population-based study of 898 patients regarding epidemiology, cytogenetics, treatment and outcome. Intern Med J 2013; 43: 903–911.

Download references


This study was supported by research funding from the Danish Cancer Society, The University of Aarhus (Faculty of Health), the AV Lykfeldt’s Foundation, Gr Brogaard and Wife’s Foundation, F Ejner Willumsen’s Grant, Arvid Nilsson’s Foundation and Fraenkel’s Memorial Foundation to LGSØ. None of the funding sources had a role in the design, conduct, analysis or reporting of the study results.

Author information

Correspondence to L S G Østgard.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies this paper on the Leukemia website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Østgard, L., Nørgaard, J., Sengeløv, H. et al. Impact of chemotherapy delay on short- and long-term survival in younger and older AML patients: a Danish population-based cohort study. Leukemia 28, 1926–1929 (2014) doi:10.1038/leu.2014.157

Download citation

Further reading