Fossati and colleagues (2009) performed a systematic review of randomized clinical trials in metastatic breast cancer reported between 1975 and 2009 that were identified by a MEDLINE and EMBASE search. The data extracted from each report included tumor response, the hazard ratio for mortality, and proportion of patients with severe side effects. A total of 189 randomized trials was identified, of which all provided response data and 133 (70%) provided data or Kaplan-Meier curves necessary for calculation of the hazard ratios. A total of 165 were two-arm trials, and 24 studies included three or more arms. The trials were categorized into twelve separate groups by their primary comparison, in­cluding six groups that included chemotherapy as a component of therapy :

Polychemotherapy versus single agent therapy: Polychemotherapy was associated with a higher objective response rate (48 versus 34%) and a significant reduction in the hazard rate for death whether the comparison was anthracycline combinations versus single-agent anthracyclines (HR 0.87; 95%), or nonanthracycline combinations versus single agents (HR 0.70; 95%).

Anthracycline versus nonanthracycline chemotherapy: Anthracyclines were associated with a significantly higher response rate (51 versus 45%) but had no significant effect on survival. Anthracyclines produced more nausea and vomiting, leukopenia, alopecia, and neurological and cardiac toxicity. There was a modest reduction rate in the hazard rate for death (HR0.89; 95%) if the comparison regimen did not contain predni­sone, whereas there was a disadvantage for anthracyclines (HR 1.16; 95%) when compared with a regimen that contained prednisone.

Other chemotherapy versus CMF: There was a slightly higher response rate for the non-CMF regimens (49 versus 44%), but the non-CMF regimens produced significantly more nausea and vomiting, leukopenia, alopecia, and neurological toxicity. There was no significant difference in the hazard rate for death.

Epirubicin versus doxorubicin: There was no significant difference in response rate (44 versus 47%). Epirubicin produced less leukopenia and cardiac toxicity. Epirubicin was associated with a trend toward a higher risk for death (HR 1.13; 95%), although this was not statistically significant.

Standard-dose versus low-dose chemotherapy: Standard-dose chemotherapy was asso­ciated with a significantly higher response rate (44 versus 33%) but also with more nausea and vomiting, leukopenia, mucositis, and alopecia. Standard-dose therapy was associated with a significant reduction in the hazard rate for death (HR 0.90; 95%).

Chemotherapy versus chemohormonal therapy: Hormonal therapy consisted of ta­moxifen, medroxyprogesterone acetate, estrogen, oophorectomy, and other hormones. There was a higher response rate for chemohormonal therapy (56 versus 46%) but also a higher risk of cardiac toxicity, hot flashes, and edema. There was no significant effect on the hazard rate for death.