Forty-two RCTs (n=11,941) of maintenance therapy were identified. IPD were available for 38 RCTs (n=10,742) and published data were available for 4 RCTs (n=1,199). Seven other trials were identified, but they were not included in the review because information was not yet available.
No relevant trials were excluded from the analyses following the data checking procedures.
Duration of maintenance therapy.
IPD were available for 14 RCTs (n=3,115) and published data were available for 2 RCTs (n=746). Data for one RCT were unavailable.
Based on IPD (n=3,115) from 14 RCTs, death during first remission was significantly more likely to occur in patients who received a longer duration of maintenance chemotherapy (2.7%) than in those who received a shorter duration (2.7% versus 2.1%; 2P=0.004). However, there was a significant 21% reduction in the odds of a relapse or death with longer maintenance compared with shorter maintenance (2P=0.0003), corresponding to an absolute reduction in risk of 4.3% based on IPD and published data (n=3,861) from 16 RCTs. The absolute reduction in risk was 3.8% (2P=0.003) when the analysis was based only on IPD (n=3,115) from 14 trials.
When analysing the different types of events separately from IPD, there was a significant decrease in relapse during first remission at bone marrow (2P=0.00005) and testes (2P=0.009) for longer duration of maintenance therapy compared with shorter duration. There was no significant difference in relapse at CNS, death after relapse, or death during first remission (2P>0.1).
The subgroup analysis by gender, age and initial WBC did not impact on the effects of treatment on survival in first remission or on survival. There was no evidence of statistical heterogeneity (chi-squared 19.6, P>0.1).
VP pulses.
IPD were available for 4 RCTs (n=1,052) and published data were available for 1 RCT (n=199). Data for 2 RCTs were unavailable.
Based on IPD (n=1,052) from 4 RCTs, death during first remission was not significantly more likely to occur in patients given VP pulses than in those who were not given VP pulses (4% versus 3.2%; 2P>0.1). However, there was a significant 29% reduction in the odds of a relapse or death with the addition of VP pulses compared with no VP pulses (2P=0.0004), corresponding to an absolute reduction in risk of 9.2% based on IPD and published data (n=1,251) from 5 RCTs. The absolute reduction in risk was 10.4% (2P=0.003) when the analysis was based only on IPD (n=1,052) from 4 trials.
When analysing the different types of events separately from IPD, there was a significant decrease in relapse during first remission at bone marrow (2P=0.0008) and testes (2P=0.0003), and a reduction in death after relapse (2P=0.08), for VP pulses compared with no VP pulses. There was no significant difference in relapse at CNS (2P>0.1) or any death during first remission (2P>0.1).
The subgroup analysis by gender, age and initial WBC did not impact on the effects of treatment on survival in first remission or on survival. There was no evidence of statistical heterogeneity (data not given).
Addition of re-induction therapy during maintenance.
IPD were available for 4 RCTs (n=3,442) and published data were available for 2 RCTs (n=254). Data for 1 RCT were unavailable.
Based on IPD (n=3,442) from 4 RCTs, death during first remission was significantly more likely to occur in patients receiving intensive re-induction therapy than in those not receiving re-induction therapy (4.8% versus 3.3%; 2P=0.08). However, there was a 27% reduction in the odds of a relapse or death during first remission with the addition of intensive re-induction compared with no re-induction therapy (27.8% versus 35.8%; 2P<0.00001), corresponding to an absolute reduction in risk of 8% based on IPD and published data (n=3,696) from 6 RCTs. The absolute reduction in risk was 7.8% (2P=0.00001) when the analysis was based only on IPD (n=3,442) from 4 RCTs.
When analysing the different types of events separately from IPD, there was a significant reduction in relapse during first remission at bone marrow (2P=0.001), CNS (2P=0.003) or testes (2P=0.06). There was also a significant reduction in death after relapse (2P=0.003) and the overall number of deaths (2P=0.01). There was no significant difference in death without remission (2P>0.1).
The subgroup analysis by gender, age and initial WBC did not impact on the effects of treatment on survival in first remission or on overall survival. There was no evidence of statistical heterogeneity (data not given).
Other maintenance therapy.
IPD were available for 15 RCTs (n=3,133).
There was a non significant 1% reduction in the odds of relapse or death during first remission (2P>0.1). The authors reported that the total numbers randomised were small and no clear conclusions were apparent.