An attractive option could be to continue the anti-angiogenic agent as maintenance therapy in patients who are responder to first-line bevacizumab-based chemotherapy. This approach has an interesting preclinical rationale derived from studies that have suggested the power to increase tumor growth following withdrawal of VEGF inhibitors. In fact, although the VEGF inhibitors can destroy as much as 80% of tumor vasculature, tumor vessels may rapidly re-grow after cessation of treatment with these inhibitors [11,12]. Recently, Mancuso et al. [13] have also shown that when tumors in transgenic mouse were inhibited by VEGF tyrosine kinase receptor the same tumors were completely re-vascularized within the first week after stopping treatment, indicating that surviving pericytes and the empty sleeves of vascular basement membrane contributed to the rapid restoration of the tumor vasculature.

These preclinical data and the emerging results in advanced colo-rectal cancer [14,15], findings suggesting the clinical benefit of bevacizumab beyond first-line therapy due to permanent suppression of VEGF, could be an assumption for the clinical use of the biological agent as maintenance treatment also in mBC patients.

Preclinical models suggested that addition of anti-vascular endothelial growth factor therapy could improve the efficacy of anti-estrogens in hormone-sensitive breast cancer [16]. Recently a phase II trial in advanced breast cancer evaluated the feasibility and efficacy of bevacizumab added to either anastrozole or fulvestrant in postmenopausal hormonal receptor-positive patients already resistant to the adjuvant aromatase inhibitor. Both regimens showed a good response rate and encouraging progression free survival (ORR 28%, median PFS 18months) without severe toxicity registered in both regimens [17].

In the preclinical scenario, estrogen may play a key role in the regulation of angiogenesis. A direct effect of endocrine therapy on tumor vasculature has been reported. In a male mouse model of androgen-dependent breast cancer, castration led to tumor shrinkage and vascular regression [16]. However, endocrine resistance in this tumor model was suggested by a wave of neo-vascularization and tumor re-growth. On the other hand, endocrine castration initially caused a reduction in VEGF mRNA levels while, at the time of tumor re-growth, VEGF mRNA levels simultaneously rebound. Moreover, data coming from retrospective studies in patients with breast cancer indicated that high VEGF levels in breast tumor tissue were associated with decreased responsiveness to hormonal therapy in both adjuvant and metastatic breast cancer patients [20,21]. Overall, these data suggested that anti-VEGF therapy may delay or prevent the onset of endocrine therapy resistance in patients with hormone-sensitive breast cancer, supporting the use of the combined therapy with anti-angiogenic plus anti-hormonal agents in breast cancer. In our series, among patients with tumors expressing hormonal receptors, a better mBev PFS was seen when they were given an aromatase inhibitor plus bevacizumab compared with those who did not received any hormonal agent (mBev PFS 13months vs 4.1months, p=0.05),

To date, few data exist regarding the efficacy and safety of bevacizumab as maintenance treatment in mBC. Smith et al. [22] have recently reported a subgroup analyses in patients enrolled in the Athena study who continued single agent bevacizumab after stopping chemotherapy. They found a median TTP and OS of 18.4months and 30.0months, respectively. In the majority of these patients, maintenance bevacizumab was administered alone for the extended periods and no informations were given about use of bevacizumab plus HT in hormonal receptors positive patients. In our prospective analysis on HER2-negative breast cancer patients treated with mBev beyond controlled disease after first-line BT therapy, the ORr (CR+PR, 62.8%) resembled those reported in large randomized phase III trials [7-9]. We administered bevacizumab in mBC beyond the end of chemotherapy until disease progression or unacceptable toxicity. The maintenance therapy with bevacizumab was effective, reporting the ORr of 48.5%, the median mBev PFS of 6.8months (95%CI 0.8-12.7) and the duration of clinical benefit of 17.1months (95%CI 12.2-21.9).

According to data coming from a recent meta-analysis, it could also be of interest that prior treatment with adjuvant taxanes was predictive of better mBev PFS [23]. In our analysis, the median mBev PFS was 10.2months among patients who received a taxane as adjuvant treatment compared with 6.6months among those who did not (P=0.57). These findings are strengthen by the use of bevacizumab plus paclitaxel as first-line treatment also in women previously administered adjuvant taxane-including treatments.

Bevacizumab-associated adverse events are well known and include arterial thrombosis, wound-healing complications, gastrointestinal GI perforation, bleeding, hypertension, and proteinuria. A recent meta-analysis [24] of randomized, controlled bevacizumab-containing trials reported all grade hypertension in 3%to 36%of patients and proteinuria in 21%to 63%of patients with nephrotic-range proteinuria up to 2%of patients. The degree of proteinuria was bevacizumab dose-dependent. The pathogenesis of bevacizumab-associated hypertension and renal toxicity is poorly understood; however, one clinical study has suggested an association between these two adverse events [25]. An interesting explanation of the higher rate of proteinuria is that the continued inhibition of local VEGF in glomerular podocytes, as occurs in the prolonged bevacizumab treatment, leads to renal thrombotic microangiopathy, which causes the final event of proteinuria [26]. In our study, no evidence of increased risk of developing severe toxicities with longer bevacizumab treatment was reported, except for hypertension and proteinuria. According to data reported by Smith et al. (22), the onset of grade 3 proteinuria was common in the maintenance period. Similar findings were observed in advanced ovarian cancer, in patients receiving bevacizumab for prolonged time, after treatment with carboplatin, paclitaxel and bevacizumab [27]. In our study all patients recovered proteinuria toxicity after 2weeks of bevacizumab discontinuation and all patients continued treatment. Grade 3 hypertension was infrequent (8.5%) and in that cases bevacizumab was continued in association with antihypertensive therapy. Adding hormonal therapy to bevacizumab did not cause any further toxicity.

Our study, despite the small number of patients, is the first prospective report on the activity and feasibility of a maintenance combination of bevacizumab with hormonal therapy in bevacizumab-pretreated women. In absence of data comparing association of HTBev versus HT alone, our study could be the basis for a comparison trial. Unfortunately, some questions remain still unanswered: 1) What are the most useful biomarkers to predict the efficacy during bevacizumab treatment? 2) What is the optimal subset of patients could benefit by mBev? 3) Is there the optimal hormonal treatment to combine with bevacizumab?