Droplet-based Microfluidics: Sensitive Detection of Rare Mutations
- Fig. 1: Principle of the detection method. We want to find a rare mutant sequence (in green) in an ensemble of normal sequences (a). In bulk, a signal coming from the mutant sequence is in the background of the normal sequence which makes it difficult to measure for low dilution. We encapsulate single genes in droplets (~10 microns each) with a fluorescence marker targeted to the sequence of interest (green for a mutant, red for a normal gene) (b) and amplify each sequence in the droplet (c). The whole droplet becomes fluorescent: a green droplet contains a mutant, a red a normal gene and a white none of those. The simple counting procedure of red and green droplets provides a measurement of the initial ratio of mutant to normal genes.
- Fig. 2: Microfluidics is used for the encapsulation of genes in order to obtain droplets of exactly the same size (a,b). The fluorescent droplets obtained after amplification of the genes are counted to determine the initial ratio of mutant to normal gene. (b) Example of droplets of about 50 microns in diameter produced in microfluidics and containing either fluorescein (green), resorufin (red) or are empty (black).
- Dr. Valérie Taly, Universite Paris Descartes, INSERM UMR-S775, Paris Cedex and ISIS-CNRS-Université de Strasbourg, Strasbourg, France, Dr. Jean-Christophe Baret, Max Planck Institute for Dynamics and Self-Organization, Göttingen, Germany
In oncology, the success of treatment against a specific cancer is linked to the ability to detect tumor cells as early as possible: early diagnostics are essential part in the fight against cancer. In that respect, there is a need for sensitive methods to find traces of cancer biomarkers in samples coming from patients. A vision for the future of cancer diagnostics would be based on the analysis of bodily fluids to find these biomarkers. A high sensitivity is there a strong pre-requisite to be able to distinguish genetic alterations at the level of traces in such samples.
Complex networks of genetic alterations have been described for most cancers: deletions, amplifications, point mutations and chromosomal rearrangements are all playing a major role in the development and progression of cancers [1, 2]. These somatic mutations are present in tumor cells but not in normal cells and can serve as highly specific biomarkers . Tumor-specific genetic changes constitute great tools for early cancer detection and can have a profound impact on clinical decision-making and outcome [4, 5]. Plain use of such biomarkers for clinical oncology requires their detection in a large excess of non-mutated DNA from normal cells. From the many possible biomarkers KRAS constitutes a good model. Indeed, this oncogene has been found mutated in adenocarcinomas of the pancreas (70-90 % incidence), colon (50 %) or lung (25-50 %)  and the presence of KRAS mutations is associated with an absence of response to colorectal cancer targeted therapies employing anti-EGFR antibodies .
The techniques that are classically used in clinics, such as dual probe TaqMan assays and pyrosequencing, while quantitative, cannot detect less than ~1 % mutant genes in a background of non-mutated DNA from normal cells . Indeed, majority of genetic tests which aim to identify variations of DNA sequences incorporate a step of PCR and this method has a limited sensitivity when amplifying complex mixtures of DNA (like DNA extracted from tumors, plasma or feces).
More precise and sensitive quantification of mutated DNA is possible using digital techniques, which allow the analysis of many individual DNA molecules in parallel rather than the analysis of a pool of different DNAs.
One of the most important techniques is digital PCR, which is based on the compartmentalization and amplification of single DNA molecules . Digital PCR allows the discrete counting of the mutant and wild-type alleles present in a sample. Its sensitivity is only limited by the number of molecules that can be analyzed and the false positive rate of the mutation detection assay.
Digital assays are particularly well suited for the analysis of clinical samples like stool or plasma where tumor derived DNA fragments represent only a small fraction of the total DNA  and digital PCR has been used for detection of KRAS mutations in various clinical samples [9-12]. The original microtitre plate-based digital PCR techniques were, however, both slow and expensive. To address these issues, several miniaturized methods allowing millions of single-molecule PCR reactions to be performed in a single assay have been developed. The method we have developed , is based on this idea.
In order to perform the assay, we need to manipulate and count single genes, some of them being mutants, the rest being wild-type. For this, we use fluorescent markers which attach to specific sequences: a red marker for a wild-type gene, a green for a mutant. If we would measure the fluorescence of the sample in the red and green color, we could have access to the ratio of genes. But as discussed above, such global strategy would not provide sufficient sensitivity if the mutant is too diluted in the sample. We need to count the red and green genes at the single molecule level to increase the sensitivity. However single gene analysis requires complex optical systems and are slow which does not provide a large number of analysis. To circumvent these problems, we isolate each gene in single reactors and amplify them to enhance the fluorescent signal. We do not measure directly the fluorescence of the single gene but of millions of genes: the whole microreactor becomes fluorescent and easy to measure. To do so, we use droplet-based microfluidic systems that allow the creation of highly monodisperse droplets of a few picoliter (<1.5 % polydispersity) and the precise manipulation and optical analysis of the droplets at the kHz frequencies. The analysis of millions of genes (or droplets) can then be performed in less than 1 hour. The count of green and red droplets simply relates to the initial dilution of genes in a quantitative and sensitive manner. It enabled the precise determination of mutant allelic specific imbalance (MASI) in several cancer cell lines and the precise quantification of a single mutated KRAS gene in a background of 200.000 non-mutated KRAS genes. Furthermore, it was also possible to screen the six common mutations in KRAS codon 12 in parallel in a single experiment. Such parallelization is extremely powerful because it does not only provide a readout of the presence of a tumor but also pinpoints which mutation is involved. This information can then be used to guide the oncologist to the best therapeutic strategy.
Thanks to the sensitivity and quantitativity of the methods, it will perhaps be possible, in the near future, to detect cancer by a simple blood or urine test. We have successfully applied our method to the KRAS oncogene. The DNA bearing this gene was derived from laboratory cell lines. This new analytical method now needs to be tested in a therapeutic context. A clinical study is already scheduled. If it is a success, physicians will have an efficient diagnostic tools, not just for detecting the presence of tumors but also for proposing treatments. The aggressiveness of the cancer, its responsiveness to existing treatments and its risk of recurrence following local treatment: all this information is partly contained in the tumoral DNA. By deciphering it with the microdroplet technology, oncologists could benefit from a powerful diagnostic tool to help predict the evolution of the disease and determine a therapeutic strategy.
 Vogelstein B. et al.: Nat Med, vol. 10, pp. 789-799 (2004)
 Stratton M. R et al.: Nature, vol. 458, pp. 719-724 (2009)
 Diehl F. et al.: Nat Med, vol. 14, pp. 985-990, (2008)
 Negm R. S. et al.: Trends Mol Med, vol. 8, pp. 288-293 (2002)
 Sawyers C. L.: Nature, vol. 452, pp. 548-552 (2008)
 Johnson L. et al.: Nature, vol. 410, pp. 1111-1116 (2001)
 Lievre A. et al.: Cancer Res, vol. 66, pp. 3992-3995 (2006)
 Tsiatis A. C. et al.: J Mol Diagn, vol. 12, pp. 425-432 (2010)
 Vogelstein B. et al.: Proc Natl Acad Sci U S A, vol. 96, pp. 9236-9241 (1999)
 Dong, S. M. et al.: J Natl Cancer Inst, vol. 93, pp. 858-865 (2001)
Please ask the authors for further references.
Additional information can be found in ref 13:
Pekin et al.: Lab. Chip (2011)