Pinprick diagnostics

Rare tumour cells with mutations that confer drug resistance can go undetected by standard testing procedures, according to two studies, which show that such mutations can be detected in patients' blood. See Letters p.532 and p.537

Some tumours carry mutations that confer resistance to specific drugs, precluding the use of these drugs in patients with these mutations. However, drug resistance can arise even when a tumour does not seem to meet these criteria. In this issue, Misale et al.1 (page 532) and Diaz et al.2 (page 537) report that a commonly acquired drug resistance in colorectal cancer can be explained by rare tumour cells that have mutations in a particular gene, KRAS, and that outgrow susceptible cancer cells during drug exposure. Excitingly, the two research groups demonstrate that they can detect this mutation in the tumour DNA that circulates freely in a patient's bloodstream. Their findings highlight the potential for blood samples to be used in the diagnosis and monitoring of cancers, thereby reducing the need for more-invasive procedures such as tumour biopsies.

Patients with colorectal cancers are often treated with drugs that target epidermal growth factor receptor (EGFR), a signalling protein that activates cellular proliferation. A key predictor of a person's suitability for anti-EGFR therapy is whether their cancerous cells contain a mutated form of KRAS; the KRAS protein functions downstream of EGFR, such that cells with overactive KRAS will not respond to anti-EGFR therapies. The use of KRAS mutations for predicting the effectiveness of anti-EGFR treatment is an example of successful, although imperfect, biomarker development in oncology, and also demonstrates how an understanding of cancer biology should inform drug development at an early stage of the process.

However, the majority of patients who are selected for anti-EGFR therapy develop secondary resistance to the drug within 5–11 months. One of the main challenges in colorectal cancer research is to uncover the molecular mechanisms involved in this acquired resistance. An essential question is whether the resistance mechanisms derive from new mutation events or from existing clones — rare (and undetected) mutation-carrying cells that are present in the tumour at the start of treatment.

Results obtained by Misale et al. and Diaz et al. using in vitro experiments and mathematical models suggest that secondary resistance to anti-EGFR therapy is a fait accompli, because of the presence of KRAS-mutant clones in tumours that initially seem to express only wild-type KRAS. Both papers also present data from colorectal cancer patients whose tumours were classified, by biopsy, as KRAS wild-type at the initiation of therapy, but in whom the authors detected KRAS-mutant clones following treatment with either of two anti-EGFR drugs, cetuximab or panitumumab. These clinical studies were made possible by the use of 'liquid biopsies' in which circulating tumour DNA (ctDNA) was extracted from the serum of the patients' blood. The ctDNA was then studied using an ultrasensitive mutation-detection method called the BEAMing assay3. In both studies, the mutant clones were identified before drug resistance could be detected using standard tumour-imaging techniques.

These findings suggest that the same biological mechanism (in this case, KRAS mutation) can drive both primary and secondary drug resistance, and that secondary resistance can arise when tumours are genetically heterogeneous, such that some cells can acquire a growth advantage during treatment. However, this idea contrasts with another mechanism for secondary resistance that has been reported recently, involving a mutation in EGFR that impairs cetuximab binding but that seems to be absent in pre-existing tumour-cell subpopulations4. Another of Misale and colleagues' observations is also at odds with the existing-clone hypothesis. In their experimental models, Misale et al. report examples of resistance arising from mutations in codon 146 of the KRAS gene (a codon is a string of three DNA bases that codes for an amino acid). However, the typical sites of KRAS mutations in tumour samples are codons 12 and 13, and codon-146 mutations have been described in less than 5% of colorectal tumours5. These unusual mutations are more suggestive of resistance arising from an adaptive response of tumour cells, in which novel mutations accumulate over time, rather than from clonal selection linked to tumour heterogeneity.

It is becoming clear that the existing strategies used to select therapies on the basis of a tumour's genetic status are suboptimal. The existence of heterogeneous tumours and the evolution of diverse cancer-cell populations over the course of the disease, especially under the selection pressure of local and systemic treatment, need to be taken into account in both individual treatment strategies and drug development6. We anticipate that tumour heterogeneity will be a recurring theme in future oncology research aimed at improving therapeutics.

The schematic representation of tumour heterogeneity as a branched tree, an idea borrowed from the field of ecology, is a realistic approximation of the evolution of a cancer7. Following this theme, therapeutic approaches could be likened to a gardener's toolkit — by targeting selected cellular pathways, certain branches of the tree will be 'pruned'. This pruning will provide selection pressures that favour the growth of cells that are present in very low abundance, which will shift the behaviour of the tumour from that determined by the mutations present in the pruned branches to that specified by the biology of the now-dominant cell populations (Fig. 1). Nonetheless, greater understanding of the proliferative dynamics of cell clones, and of the interactions between them, is needed for such strategies to be successful. The complexity of these interactions is exemplified by the fact that although only 0.4–17% of the mutated alleles detected by Misale et al. in patients with anti-EGFR-resistant tumours involved the KRAS gene, the whole tumour became resistant to therapy.

Figure 1: Monitoring for resistance.

Colorectal cancer patients are often treated with drugs that target a protein called EGFR, but many patients develop resistance to these drugs. Misale et al.1 and Diaz et al.2 propose that this resistance could be caused by the proliferation of rare tumour cells with mutations in the KRAS gene that allow them to survive anti-EGFR therapy. The authors show that these mutated cells can be detected and quantified (orange line) by testing for the presence of KRAS mutations in circulating tumour DNA (ctDNA) derived from a patient's blood. This ctDNA screening could allow the onset of drug resistance to be detected at an earlier stage of disease progression than can be achieved using standard techniques, which rely on imaging to look for enlarged tumour size and/or metastasis. Earlier detection would allow treatment with a drug targeting an alternative pathway to be commenced sooner. However, it is likely that the tumour will also contain rare cells that harbour mutations conferring resistance to this second drug, and that these cells will, in turn, become more prevalent (green line)

Major challenges for the accurate assessment of tumour heterogeneity include gaining access to tissue and the question of appropriate sampling. Currently, biomarker-driven therapeutic decisions rely mostly on testing biopsied samples of primary tumours that have been stored since the initial diagnosis, or biopsies from new metastatic sites. However, single biopsies are inadequate. Primary tumours and their metastases are connected by a mixture of different tissues, including cancer cells and supporting structures such as stroma and blood vessels. Moreover, as has been reported8 in renal-cell cancer, common alterations in cancer-causing genes may differ in different regions of the tumour. Therefore, a single cancerous site and a single tumour biopsy will probably not provide a representative picture of the tumour landscape.

As has previously been suggested9, these sampling problems could be eliminated by the use of blood biopsies to study genomic alterations in ctDNA. The virtues of this approach include its low invasiveness, the ease of obtaining samples at different time points, and the lack of spatial sampling bias. However, we still need a better understanding of the contribution of each dynamic cell population in a heterogeneous tumour to ctDNA. Furthermore, assays are required that can interrogate mutations in several genes at the same time, which represents a technical challenge because the total amount of ctDNA and its quality are both relatively low. In one recent attempt10, researchers used next-generation DNA-sequencing technology, instead of BEAMing, to simultaneously screen ctDNA from patients with ovarian cancer for the presence of mutations in seven cancer-related genes. Such emerging mutation-detection platforms will raise standards for interrogating heterogeneous cancer-cell populations. Applications of this technique are likely to expand from the early diagnosis of secondary mutations that could lead to drug resistance, as demonstrated by Misale et al. and Diaz et al., to include metastasis monitoring, diagnosis of relapse after surgery, and the early detection of individuals at risk of developing cancer.


  1. 1

    Misale, S. et al. Nature 486, 532–536 (2012).

  2. 2

    Diaz, L. A. Jr et al. Nature 486, 537–540 (2012).

  3. 3

    Diehl, F. et al. Nature Methods 3, 551–559 (2006).

  4. 4

    Montagut, C. et al. Nature Med. 18, 221–223 (2012).

  5. 5

    Edkins, S. et al. Cancer Biol. Ther. 5, 928–932 (2006).

  6. 6

    Marusyk, A., Almendro, V. & Polyak, K. Nature Rev. Cancer 12, 323–334, (2012).

  7. 7

    Yap, T. A., Gerlinger, M., Futreal, P. A., Pusztai, L. & Swanton, C. Sci. Transl. Med. 4, 127ps10 (2012).

  8. 8

    Gerlinger, M. et al. N. Engl. J. Med. 366, 883–892 (2012).

  9. 9

    Diehl, F. et al. Nature Med. 14, 985–990 (2008).

  10. 10

    Forshew, T. et al. Sci. Transl. Med. 4, 136ra68 (2012).

Download references

Author information

Correspondence to Eduardo Vilar or Josep Tabernero.

Rights and permissions

Reprints and Permissions

About this article

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.