News & Views | Published:

Translational medicine

Cancer lessons from mice to humans

Nature volume 471, pages 316317 (17 March 2011) | Download Citation

New clinical trials report the efficacy of two mechanism-based therapies for treating human pancreatic neuroendocrine tumours. Studies in mouse models have contributed to these success stories, and continue to do so.

Advances in cancer medicine have reset our clinical and social expectations: the aim now is to effectively combat formidable tumours — an effort that was previously deemed improbable. Writing in The New England Journal of Medicine, Raymond et al.1 and Yao et al.2 report phase III clinical trials of two drugs that target distinctive cancer-associated signalling pathways. The results suggest an impressive efficacy of both drugs (sunitinib and everolimus) for treating pancreatic neuroendocrine tumours. It is therefore likely that these drugs, which are already standard treatments for other cancers, will become the first new approvals in 25 years by the US Food and Drug Administration (FDA) for treating these cancers, a remarkable milestone.

Pancreatic neuroendocrine tumours (PNET) are uncommon, but difficult to diagnose and treat. These cancers, which originate from the hormone-producing pancreatic islet cells, stand in stark contrast to another type of pancreatic cancer, pancreatic ductal adenocarcinoma, which is much more prevalent and deadly: a larger proportion of patients with PNET undergo surgical excision, and the clinical course of the disease is highly variable. Nonetheless, patients with advanced PNET who are not candidates for surgery have a terminal illness, and their tumours are difficult to manage; the FDA-approved chemotherapeutic agent streptozotocin shows only modest activity in these patients.

A vast number of potential anticancer drugs are currently in the pipelines of biopharmaceutical companies. Indeed, the scope of mechanism-based targeting is broad, often with several potential drugs affecting the same target. Consequently, it is challenging to decide which targets and candidate drugs might be of value in particular forms of human cancer, especially those that are rare but deadly like PNET. There is growing optimism that genetically engineered mouse models, which can mimic the progression of specific types of human cancer at the genomic and tissue levels, can contribute to this prioritization3. The hope is that preclinical trials of candidate drugs in representative mouse models could help to motivate and guide clinical trials of targeted therapies in the related human tumours (Fig. 1). The two new papers1,2 reflect proof of this concept.

Figure 1: Linking preclinical and clinical trials.
Figure 1

(1) Preclinical trials on cohorts of mice engineered to develop a particular type of cancer are a good starting point for evaluating mechanism-based drugs. (2) If the mice show detectable therapeutic benefits, such as increased survival and/or tumour shrinkage, the preclinical trials can motivate and guide the design of clinical trials in the same type of cancer. (3) Clinical benefit, such as tumour shrinkage, increased progression-free-survival and overall survival, can justify drug approval for clinical use. (4) Relapses and clinical failures, however, can be translated back into refined preclinical trials aiming to understand and circumvent the limitation.

The mouse model of PNET, called RIP-Tag2, shows similar tissue-level features to the human tumours4. However, the cancer in the animal does not follow the same — currently obscure5 — initiating events that lead to human PNET; it is instead driven by a viral oncogene that abrogates the function of two generic tumour-suppressor pathways commonly lost in human tumours.

Preclinical trials in this model had predicted that both sunitinib, a pan-specific inhibitor of tyrosine-kinase enzymes, and everolimus, which inhibits another kinase, mTOR, would be effective in treating human PNET. Several studies6,7,8 showed that sunitinib, and other kinase inhibitors that target signalling associated with angiogenesis through receptors for the growth factors VEGF and PDGF (thus inhibiting angiogenesis), cause tumour shrinkage. Sunitinib also produced increased survival in the animal studies. These results motivated Raymond and colleagues to perform two phase II trials9,10 and now the phase III trial1 of sunitinib in patients with PNET (Box 1). Similarly, a separate study11 reported the efficacy of another mTOR inhibitor, rapamycin, in treating PNET in the mouse model, presaging the clinical success of everolimus, a refined mTOR inhibitor, which Yao et al.2 now describe.

Box 1: A closer look at the trials

Both trials1,2 enrolled patients with advanced pancreatic neuroendocrine tumours (PNET) who had already received much treatment (including surgery and chemotherapy).

Raymond et al.1 compared the response of 86 randomly selected patients given sunitinib with that of 85 control patients on a placebo drug. Neither the patients nor their doctors were aware of who was receiving which treatment — a double-blind trial.

The median tumour-progression-free survival of patients on sunitinib (11.4 months) was appreciably longer than that of the control group (5.5 months). Moreover, patients treated with sunitinib showed early signs of an increase in overall survival. The side effects were typical of previous experience with sunitinib and included gastrointestinal disturbances and fatigue.

Yao and colleagues2 compared the response of 207 randomly selected patients on everolimus with that of 203 patients who received only the best supportive care. In this study, the patients and their physicians were aware of the treatment used, with patients who were receiving only supportive care having the option to switch to everolimus if their cancer progressed. Nonetheless, all patients were classified by the treatment they initially received (intention to treat).

Everolimus was beneficial, extending the median time of progression from 4.6 months to 11 months. The incidence of rash, gastrointestinal disturbances, fatigue, anaemia and infections was higher in patients treated with this drug, but these side effects were largely manageable. Although no increase in overall survival of patients treated with everolimus was noted, this conclusion is tentative because of the cross-over of patients and the continuance of the trial for the best responders to everolimus.

It should be determined whether these treatments are also beneficial to other groups of patients with PNET, including those who have just had surgery, those who have early-stage disease and those who were marginally too ill to qualify for these trials1,2. D.T. & D.H.

Although drug efficacies seen in the preclinical trials were encouraging, the trials also revealed limitations — in tumour shrinkage and long-term survival of the mice — that may well influence how these drugs are most effectively used to treat human PNET. Yao and colleagues also find that, whereas everolimus delays time to progression of the disease (progression-free survival), it seemingly does not increase overall survival rates. This trial is still ongoing, however, so the lack of effect on overall survival is not yet conclusive.

Pertinent to this clinical observation is an intriguing result with translational potential from preclinical trials of rapamycin in the mouse model of PNET. Rapamycin on its own produced only a modest overall survival benefit, and the animals also showed evidence of rapamycin resistance following treatment, in the form of regrowth of the previously responding tumours9. But when rapamycin was given in combination with another approved drug — erlotinib, which inhibits the growth factor receptor EGFR — the animals' overall survival rate improved significantly and there was a decrease in relapse during treatment11.

These outcomes in the PNET mouse model are consistent with the possibly limited overall survival of patients with PNET following treatment with everolimus only2. The preclinical results therefore encourage clinical trials on everolimus in combination with erlotinib (or with other drugs that target downstream effectors in the same signalling pathway). A small clinical trial12 combining the two drugs to treat PNET is already under way.

For sunitinib, the tumour shrinkage and increased overall survival seen in preclinical trials8 are recapitulated in the human trial: Raymond et al.1 report improved both progression-free survival and overall survival after administration of this drug. But, as with everolimus, preclinical trials revealed limitations to the effectiveness of sunitinib in the form of adaptive resistance in PNET. In other words, faced with sunitinib's potent effect in blocking angiogenesis, the tumours not only adapt after a period of shrinkage, but also survive the treatment better, inducing alternative pro-angiogenic signalling circuits13 and becoming more invasive and metastatic8; this reflects a phenomenon seen in other preclinical models as well as in clinical trials14,15,16.

The mouse data therefore predict eventual failure of therapy with sunitinib alone, and should motivate preclinical and clinical trials to circumvent the evasive resistance — an iterative and bidirectional process of translational therapeutic oncology.

The clinical results with everolimus and sunitinib1,2 are landmarks for treating PNET. The approach that led to this — aligned preclinical trials in a representative mouse model and human clinical trials — could also be used to test the efficacy of other anticancer drugs and may well replicate this success story. Indeed, this approach heralds a future in which preclinical trials in genetically engineered mouse models, and in other representative animal models, could guide the development of more effective therapies for human cancers, revealing efficacy, beneficial drug combinations and (potentially surmountable) mechanisms of resistance.


  1. 1.

    et al. N. Engl. J. Med. 364, 501–513 (2011).

  2. 2.

    et al. N. Engl. J. Med. 364, 514–523 (2011).

  3. 3.

    & Nature Rev. Cancer 7, 645–658 (2007).

  4. 4.

    Nature 315, 115–122 (1985).

  5. 5.

    et al. Science 10.1126/science.1200609 (2011).

  6. 6.

    , , , & J. Clin. Invest. 111, 1287–1295 (2003).

  7. 7.

    & J. Clin. Oncol. 23, 939–952 (2005).

  8. 8.

    et al. Cancer Cell 15, 220–231 (2009).

  9. 9.

    et al. J. Clin. Oncol. 24, 25–35 (2006).

  10. 10.

    et al. J. Clin. Oncol. 26, 3403–3410 (2008).

  11. 11.

    , & J. Clin. Oncol. 28, 4425–4433 (2010).

  12. 12.

  13. 13.

    , , & Cancer Cell. 8, 299–309 (2005).

  14. 14.

    & Nature Rev. Cancer 8, 592–603 (2008).

  15. 15.

    , & Eur. J. Cancer 46, 1323–1332 (2010).

  16. 16.

    & Nature Rev. Clin. Oncol. 10.1038/nrclinonc.2011.21 (2011).

Download references

Author information


  1. David Tuveson is at the Cancer Research UK/Cambridge Research Institute, University of Cambridge, Cambridge CB2 0RE, UK.

    • David Tuveson
  2. Douglas Hanahan is at the Swiss Institute for Experimental Cancer Research, Swiss Federal Institute of Technology Lausanne, CH-1015 Lausanne, Switzerland.

    • Douglas Hanahan


  1. Search for David Tuveson in:

  2. Search for Douglas Hanahan in:

Competing interests

D.H. is a member of the Therapeutic Area Scientific Advisory Panel for Oncology, Pfizer, Inc., and a consultant for Onyx Pharmaceuticals and for Aveo Pharmaceuticals

Corresponding authors

Correspondence to David Tuveson or Douglas Hanahan.

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing