In this retrospective comparison, the combination of docetaxel and trastuzumab was associated with higher response rate and longer overall survival compared with vinorelbine and trastuzumab, with no differences in time to progression. Obviously, because of retrospective nature of the study, several biases need to be taken into account when interpreting our findings. A major source of bias is the fact that the two populations were unbalanced with respect to factors that could have potentially affected clinical outcomes. For example, patients receiving docetaxel had more frequently features of aggressive disease (more patients with stage IV at first diagnosis of breast cancer, with visceral involvement and with ≥ 3 metastatic sites). By converse, patients in the vinorelbine group could be considered relatively chemoresistant because of a more frequent exposure to neo- or adjuvant chemotherapy, and to taxanes in particular. Although we performed multivariate analyses to correct the effect of treatment by these potential covariates, the real impact of these imbalances can not be estimated in this relatively small cohort of patients. Another possible source of bias in our study is the fact that no central revision of tumor response was performed and timing and modality of disease assessment might have been different according to the Institution and the treating physician. In this respect, the higher response rate in the docetaxel-trastuzumab treated group must be considered with caution.
The other relevant finding of our study was the overall survival advantage in the docetaxel group. The endpoint of mortality, which in our database is accurately collected and constantly updated, is much less subject to measurement bias than tumor response and time to progression. We might be conservative and consider also this finding as a possible result of the non-randomized nature of our study.
In spite of that, we believe that this difference in overall survival is worth additional analysis. Studying second-line options and outcomes, we found that time to progression during salvage chemotherapy was significantly longer in patients receiving docetaxel. This possibly implies higher sensitivity to subsequent chemotherapy in these patients, which could be the reason for the improvement in survival. In our opinion, this hypothesis could be true for two reasons, which are not mutually exclusive. The first is that, as already discussed, patients treated with vinorelbine had been more frequently exposed to adjuvant chemotherapy and to taxanes in particular. This exposure may have resulted in increased chemoresistance. The second reason is slightly more speculative. The two regimens that were compared differed in the duration of chemotherapy. All the patients in the docetaxel group received this drug on a 3-weekly basis and for 6-8 cycles, beyond which single agent trastuzumab was administered to non-progressing patients. Conversely weekly vinorelbine was administered with trastuzumab until tumor progression, resulting in longer duration of the combined treatment. If this could have delayed tumor progression, it is also possible that prolonged exposure to vinorelbine could have determined the development of a multidrug resistant phenotype, which limited the efficacy of subsequent salvage therapy.
Recently published guidelines indicate both docetaxel and vinorelbine among the preferred first-line agents with trastuzumab despite the absence of randomized comparisons among different combinations. If, on one hand, this implies an assumption of equal efficacy, on the other hand it points at toxicity and convenience as factors to guide therapeutic choices. Unfortunately, we were not able to collect accurate safety data in our study. However, differences between docetaxel and vinorelbine in combination with trastuzumab can be derived from the medical literature and from our previous published experience[2, 10, 18, 19]. Docetaxel is more frequently associated with grade 3 and 4 neutropenia, febrile neutropenia, thrombocytopenia (all grades), alopecia, fluid retention, nail changes and peripheral neuropathy. By converse, vinorelbine, is more frequently associated with constipation and gastrointestinal disorders. Overall, the docetaxel-trastuzumab regimen is, in fact, more toxic than vinorelbine-trastuzumab. Therefore, it is reasonable to prefer vinorelbine and trastuzumab, as many oncologists do, if the assumption of equal efficacy is true. However, we believe that our results, together with those of a randomized study in the adjuvant setting,  seem to challenge this assumption.
In high-risk, operable breast cancer patients, docetaxel followed by FEC (5-fluorouracil, epirubicin and cyclophosphamide) outperformed vinorelbine followed by FEC, with a 42% proportional reduction in the hazard of recurrence or death (p = 0.005) . The same study also randomized patients with HER2-positive disease to 9 weeks of trastuzumab given concomitantly with docetaxel or vinorelbine. A recent update of this study confirmed that the addition of trastuzumab to docetaxel, but not to vinorelbine, yielded a significant benefit in terms of time to distant relapse.
Based on our findings of an increased overall survival in patients receiving docetaxel-trastuzumab, we believe that the issue of optimal companion cytotoxic agent for trastuzumab in first-line regimens for HER2-positive metastatic breast cancer is worthy of further investigation in adequately designed prospective trials.