Sir

The article 'Clinical translation of angiogenesis inhibitors' by Robert Kerbel and Judah Folkman published in the October issue (Nature Rev. Cancer 2, 727–739) was highly appreciated and included many valuable opinions. Some important clinical results, however, were overlooked by the authors in their discussion of the effects of combining anti-angiogenic agents with chemotherapeutics.

Recent large-scale clinical studies of Zarnestra1, SU-54162, Avastin3 and Iressa4, combined with chemotherapy, have failed to show an effect on survival despite convincing rationales and apparently promising preliminary data. The International Collaborative Ovarian Neoplasm Group (ICON3) study5, which enrolled 2,074 patients with ovarian cancer and compared the effect of first-line treatment with carboplatin to carboplatin plus paclitaxel, also showed that combining the two drugs did not improve survival. Other studies with similar results6,7 warn that simultaneous combination with optimal chemotherapy has no effect on survival.

In the review article, Folkman and Kerbel discussed possible problems of simultaneous administration of anti-angiogenic agents and chemotherapy, mentioning possible reductions in blood flow, drug delivery and DNA synthesis, which would reduce sensitivity to chemotherapy. However, like other authors on this subject4,7, they do not state outright that combining these two therapeutic approaches could actually be detrimental to patients.

News stories published optimistic headlines and abstracts of the reports of Iressa therapy in animal studies8, stated that “Efficacy of cytotoxic agents against human tumor xenografts is markedly enhanced by coadministration of ZD1839 (Iressa), an inhibitor of EGFR tyrosine kinase”. They did not mention, however, that combining this drug with platinums, taxanes, gemcitabine, doxorubicin and vinorelbine caused fatal multifocal ulcerative enteritis in the mice. It has now been shown that combining it with gemcitabine and paclitaxel caused similar GI toxicity in humans4. Furthermore, the claimed 'synergistic effect' of cytotoxic agents and smaller doses of Iressa on tumour growth seemed to be no greater than the effect of Iressa alone, when administered at maximally effective doses.

Interestingly, studies of Tarceva9 in animal models showed that 9 mg/kg of this drug had an additive effect with cisplatin, but the combination therapy is less effective than 100 mg/kg of Tarceva alone. The authors do not provide a reason for dose reduction, but toxicity could be a candidate. There are no data published on the therapeutic effects of SU-5416, although a Phase I study of a combination with gemcitabine and cisplatin10, reported after completion of the unsuccessful Phase III study, showed a high incidence of serious thrombo-embolism.

A study of Genentech's Herceptin combined with doxorubicin, cyclophosphamide or paclitaxel reported large increases in response rates, progression-free survival and overall survival in metastatic breast cancer patients11. This can be cited as contrary evidence to the idea that these drug combinations are unsafe and ineffective. Herceptin, however, showed significant cardiotoxicity when combined with anthracyclines (16% versus 3%), but this was not sufficient to overcome its benefits. Genentech has also recently announced the failure of Avastin (VEGF humanized monoclonal antibody), when combined with capecitabine, to prolong survival of patients with refractory metastatic breast cancer.

In conclusion, anti-angiogenic agents cannot always be safely and effectively combined with cytotoxics. Anti-angiogenic drugs might be more effective when given as monotherapies at maximally effective doses, after patients have already received cytotoxic agents. To successfully translate the promising preclinical effects of these anticancer agents into the clinic, it is necessary to rethink general concepts about drug combinations. We must first devise appropriate pre-clinical and clinical study designs to fully determine the effects of monotherapy, and probably agree alternative end points to survival for measuring efficacy.