Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.


Mixing cocktails

A major hurdle in treating cancer is that tumour cells acquire drug resistance. To overcome this problem, one strategy might be to fine-tune the right mixture of drugs that target specific molecules.

Certain cancers are caused by oncogenic primary or 'driver' mutations in specific kinases — enzymes that regulate the activity of other proteins. Consequently, kinase inhibitors have been used in the clinic as effective single-agent drugs to shrink tumours. Kinase 'addiction' persists in advanced cancer, and patients who relapse after initially responding to kinase-inhibitor therapy often develop secondary mutations in the target kinase that confer drug resistance without impairing the kinase's oncogenicity1. Two reports2,3 in Science now show that lung cancers and glioblastoma — a malignant tumour of the central nervous system — use another option to escape drug treatment. Rather than mutate the drug target further, these cancers recruit other kinases that are not affected by the inhibitor to substitute for the pharmacologically impaired kinase and to restore downstream molecular signalling cascades that contribute to tumour growth.

Engelman and colleagues2 stumbled across this new resistance mechanism by isolating cells from a lung-cancer cell line in which the driver mutation is in a kinase known as EGFR. This belongs to a group of kinases known as receptor tyrosine kinases (RTKs). The authors found that these cells, which initially responded to an EGFR-inhibitor drug, gefitinib, subsequently became resistant, but the resistance was not due to a secondary mutation in EGFR. Instead, in these cells, the levels of another RTK, MET, were unusually high owing to amplification of the gene that encodes it. The researchers then inhibited MET activity and found that, in the presence of EGFR inhibitors, this treatment restored the cancerous cells' drug sensitivity — an observation that makes a case for combination therapy in the treatment of drug-resistant forms of lung cancer.

Before this work, the only known cause of resistance to EGFR inhibitors in lung cancer was a secondary mutation — from threonine to methionine — at position 790 in the amino-acid sequence of EGFR, abbreviated to T790M (ref. 4). The T790 residue is evolutionarily conserved among kinases and is implicated in drug resistance with other kinase–inhibitor pairings in other cancers. The T790M mutation occurs in more than half of patients with non-small-cell lung cancer who relapse after treatment with gefitinib or another EGFR-inhibitor drug, erlotinib. Engelman et al.2 suggest that amplification of MET activity may account for 20% of the remaining cases of resistance.

The work of Engelman et al. highlights the principle that, in cancer, signal-transduction pathways are remarkably flexible and amenable to rewiring. In this case, MET doesn't overcome EGFR inhibition by simply driving the canonical MET-mediated signalling pathways. Instead, it undergoes a form of molecular morphing, whereby it becomes an illicit partner of the EGFR co-receptor ErbB3 and acquires the signalling properties of EGFR (Fig. 1).

Figure 1: The battle of tumours and drugs.

Receptor tyrosine kinases (RTKs) have essential functions in mediating communication between a cell and its environment, and consequently activate downstream signalling cascades. One such RTK is EGFR, which following a primary, or driver, mutation couples with another RTK, ErbB3, promoting tumour-cell growth. EGFR-inhibitor drugs, such as gefitinib, block the oncogenic activity of EGFR, leading to tumour shrinkage. However, with time, tumour cells become resistant to RTK inhibitors by one of two mechanisms. a, It is well established that drug resistance can be caused by a secondary mutation in EGFR (EGFR), which prevents drug binding and restores tumour growth. One way to tackle this might be to develop second-generation EGFR inhibitors. b, Two new studies2,3 reveal an alternative mechanism of drug resistance. They show that, although gefitinib inhibits EGFR, other RTKs, such as MET or PDGFR, which are unaffected by this drug, might instead couple with ErbB3 and take over the task of tumour regrowth. In this case, a cocktail of drugs to simultaneously inhibit EGFR, MET and PDGFR is required.

If resistance options are viewed from the perspective of a cancer cell, this RTK switch mechanism seems an easy strategy by which to escape drug inhibition, compared with the evolutionary work involved in selecting a rare second-site mutation that would impair drug activity without crippling the oncogenicity of the original RTK. Yet, so far, second-site mutations are much more commonly documented. This presumably reflects the fact that very few tumour cell types have the molecular circuitry to allow RTK switching. Mathematically, the probability of generating secondary mutations is low5, but the probability of acquiring the still-unknown genetic makeup required to rewire a signal-transduction pathway must be lower.

In another study, Stommel and colleagues3 describe a conceptually related RTK switch in glioblastoma. As in the lung study2, the authors show that MET can substitute for EGFR function, but with a new twist. They find that the tumour cells are resistant to EGFR inhibitors upfront (rather than acquiring resistance after prolonged drug exposure), but become sensitive when MET and EGFR are inhibited in parallel. So it is as if these tumours have already been through the cycle of drug response and relapse described by Engelman et al., when in reality they have never been exposed to these drugs. Furthermore, inhibition of a third RTK — PDGFR — is required for maximal drug efficacy. It therefore seems as though these cancers have conspired to implement several back-up systems against single-agent kinase-inhibitor treatment.

Cancer cells are admittedly wily, but the more likely interpretation of this behaviour is that the orchestra of RTKs collectively contributes to disease progression. Each player presumably makes an individual contribution, but can pick up the slack if one partner is crippled. Indeed, further analysis of acquired resistance in lung cancer indicates a similar scenario, whereby MET amplification and mutation in the T790 residue of EGFR coexist in patients and in cell lines that have never been exposed to EGFR-inhibitor drugs (J. Bean and W. Pao, personal communication). Together with another study6 showing that some drug-resistant forms of the cancer-associated kinase BCR–ABL have enhanced oncogenicity, there is growing evidence that these drug-resistant proteins also have a role in tumour progression.

The most important implication of the work of Stommel et al.3, which is yet to be clinically tested, is that a much larger spectrum of cancers, not just those with driver kinase mutations, might be sensitive to cocktails of kinase inhibitors if such mixtures are delivered in the right combination. But how will we recognize these multiple-kinase-dependent cancers? Borrowing from the approach used by Engelman et al.2 and Stommel and colleagues3 to uncover the RTK-switch phenomenon, one possibility is to look at the cellular profile of RTK activation in tumour biopsies, thereby distinguishing between single-kinase-dependent and multiple-kinase-dependent cancers.

Technologies for conducting extensive profiling of proteins that carry signs of kinase activity in tumour cells, such as mass spectrometry and antibody-based approaches, are readily accessible. So 'signatures' of RTK activation in tumour biopsies could complement information gained through analysis of gene mutation status — information that is currently used to define tumours with mutations in genes encoding kinases.

Preclinical strategies to estimate the fraction of multiple-kinase-dependent cancers might also be feasible using large panels of tumour-cell lines. Such high-throughput, cell-based screens are already gaining popularity as a tool to determine the relationship between genotype and response to single-drug inhibitors. But, within reason, they could be expanded to evaluate the genotype relationship with multiple-kinase inhibition.

The new findings2,3 add fuel to the argument that, using rationally chosen combinations of kinase inhibitors, successful cancer treatment can be achieved. Prior to these reports, this view had originated from the knowledge that second-site kinase mutants are a principal cause of single-agent resistance in tumours with driver mutations in kinases. Therefore, second-generation compounds must be effective against these drug-resistant proteins and should be used in combinations that prevent the emergence of resistant subclones. A good example is the proposed combined use of the ABL-kinase-inhibitor drugs imatinib and dasatinib to treat chronic myeloid leukaemia7,8,9. The kinase-switching phenomenon reported by Engelman et al. and Stommel et al. requires a broader view that incorporates drugs that target relevant bystander RTKs. So, clearly, the cocktail menu will continue to grow, but the cocktails should be mixed with appropriate molecular guidance.


  1. 1

    Sawyers, C. L. Genes Dev. 17, 2998–3010 (2003).

    CAS  Article  Google Scholar 

  2. 2

    Engelman, J. A. et al. Science 316, 1039–1043 (2007).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Stommel, J. M. et al. Science 318, 287–290 (2007).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Pao, W. et al. PLoS Med. 2, e73 (2005).

    Article  Google Scholar 

  5. 5

    Michor, F. et al. Nature 435, 1267–1270 (2005).

    ADS  CAS  Article  Google Scholar 

  6. 6

    Skaggs, B. J. et al. Proc. Natl Acad. Sci. USA 103, 19466–19471 (2006).

    ADS  CAS  Article  Google Scholar 

  7. 7

    Shah, N. P. et al. Cancer Cell 2, 117–125 (2002).

    CAS  Article  Google Scholar 

  8. 8

    Shah, N. P. et al. J. Clin. Invest. 117, 2562–2569 (2007).

    CAS  Article  Google Scholar 

  9. 9

    Talpaz, M. et al. N. Engl. J. Med. 354, 2531–2541 (2006).

    CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Sawyers, C. Mixing cocktails. Nature 449, 993–995 (2007).

Download citation

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.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

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