Twelve studies were included for the review (n=1,089): one RCT (n=100); two non-randomised controlled trials (n=49); and nine single arm trials (n=940). Methodological quality of studies was generally low. For the RCT, the method of randomisation was not reported and allocation concealment was inadequate, but data were treated on intention-to-treat principles, sample-size calculations were reported and withdrawals and dropouts were adequately described. All but one of the non-randomised studies reported withdrawals and dropouts, but only three reported a priori sample size calculations.
Controlled studies (n=149): Conventional chemotherapy was superior to AHCT in overall survival (HR 1.79, 95% CI 1.11 to 2.91, p=0.018; three studies, n=149). There was no significant difference between conventional chemotherapy and AHCT in CHR, partial haematological response or renal response. AHCT was associated with a significantly higher risk of treatment-related mortality (RR 22, 95% CI 1.324 to 365.5, p=0.03; one RCT, n=100), a higher incidence of treatment-related infections (21% versus 0%; one study, n=31) and a higher rate of neutropenic fever and mucositis (100% versus 0%; one study, n=18) compared to conventional chemotherapy.
Single-arm studies (n=940): Pooled proportion was 0.35 (95% CI 0.25 to 0.46; seven studies, n=598) for mortality, 0.35 for CHR (95% CI 0.26 to 0.44; eight studies, n=864) and 0.12 for treatment-related mortality (95% CI 0.09 to 0.14; nine studies, n=678). Results for other outcomes in single-arm trials were also presented.
There was no evidence of statistically significant heterogeneity for meta-analyses of controlled trials. However, there was significant statistical heterogeneity for meta-analyses of single-arm trials (I2 range=71% to 86%). Sensitivity analyses did not significantly alter the results for any outcomes other than CHR. There was no evidence of publication bias for the single-arm trials.