Objective To determine factors associated with disease progression or death after autotransplantation in women with metastatic breast cancer.

Design Analysis of data collected retrospectively (January 1989 to 1992) and prospectively (1992 through January 1995) for the Autologous Blood and Marrow Transplant Registry.

Setting Sixty-three hospitals in North America, Brazil, and Russia.

Participants A total of 1188 consecutive women aged 18 to 70 years receiving autotransplants for metastatic or locally recurrent breast cancer, with a median follow-up of 29 1/2 months.

Main Outcome Measure Time to treatment failure (disease progression, disease recurrence, or death) after autotransplantation.

Results Factors associated with significantly (P<.05) increased risk of treatment failure in a Cox multivariate analysis included age older than 45 years (relative hazard, 1.17; 95% confidence interval [CI], 1.02-1.33), Karnofsky performance score less than 90% (1.27; 95% CI, 1.07-1.51), absence of hormone receptors (1.31; 95% CI, 1.15-1.51), prior use of adjuvant chemotherapy (1.31; 95% CI, 1.10-1.56), initial disease-free survival interval after adjuvant treatment of no more than 18 months (1.99; 95% CI, 1.62-2.43), metastases in the liver (1.47; 95% CI, 1.20-1.80) or central nervous system (1.56; 95% CI, 0.99-2.46 [approaches significance]) vs soft tissue, bone, or lung, 3 or more sites of metastatic disease (1.32; 95% CI, 1.13-1.54), and incomplete response vs complete response to standard-dose chemotherapy (1.65; 95% CI, 1.36-1.99). Receiving tamoxifen posttransplantation was associated with a reduced risk of treatment failure in women with hormone receptor-positive tumors (relative hazard, 0.60; 95% CI, 0.47-0.87). Women with no risk factors (n = 38) had a 3-year probability of progression-free survival of 43% (95% CI, 27%-61%) vs 4% (95% CI, 2%-8%) for women with more than 3 risk factors (n = 343),

Conclusion These data indicate that some women are unlikely to benefit from autotransplantation and should receive this treatment only after being provided with prognostic information and in the context of clinical trials attempting to improve outcome.