NILICS: Next Generation of Filtering

Sorting of Circulating Tumor Cells in Lab-on-chip Systems

  • Fig. 1: The soft lithography process for manufacturing micro-channels.Fig. 1: The soft lithography process for manufacturing micro-channels.
  • Fig. 1: The soft lithography process for manufacturing micro-channels.
  • Fig. 2: NILICS experimental setup [8].
  • Fig. 3: Micrograph images of the sorting operation at various locations in the channel and sorting efficiency as a function of flow rate Q, and hematocrit [8].
  • Table 1: Sorting efficiency of the NILICS setup for MV3 cells for various focusing flow rates and hematocrit values

Sorting of cells according to their characteristics is of fundamental importance for applications in research and medicine. The emerging field of microfluidics is opening up new possibilities for mastering this challenge. With non-inertial lift induced cell sorting (NILICS) it is possible to efficiently separate circulating tumor cells from the surrounding blood cells on a chip measuring only a few millimetres.

Necessity for Improved Diagnosis
With around 14 milion new cases and over 8 million deaths throughout the world in 2012 reported by the World Health Organization (WHO) in their global cancer report for 2014, the extent of which will become even worse over the next 20 years, the dimension of the burden is clearly apparent [1]. An important factor which increases patient‘s chances of survival is the early diagnosis of the disease. In addition, this can also result in enormous savings in treatment costs. The problem is, that over 60 % of patients live in Africa, Asia, Central or South America, i.e. countries in which widespread medical care cannot be provided.

In order to provide adequate medical care in these underdeveloped regions of the world it is necessary to promote the development of cheap and effective diagnostic methods.

A very promising approach is provided by so-called lab-on-a-chip systems (LOC). In a similar way to the development of microprocessors in the field of electronics, laboratory processes are miniaturized and integrated into a chip with only a few millimeters size. Such tiny chips are cheap to produce and are easy to transport and store, as well as requiring only a very small quantity of samples and reagents. Additionally they are easy to operate in parallel and to automate [2]. LOC systems not only reduce costs for personnel and equipment, they are considerably more sensitive too. Thus the results achieved are more precise and the analysis time is dramatically shorter than with conventional methods. This has enormous effects on patient treatment because the laboratory results are available almost immediately after the examination and samples do not have to be sent to special laboratories for assessment.

Miniaturization through Microfluidics

The facilitation of such small laboratories requires miniaturized processes and methods.

However, this changes the physical laws which are applicable and results in new technical challenges. The discipline of physics which deals with the behavior of fluids and particles, e. g. cells and bacteria, with sizes in the order of micrometers, is known as microfluidics [3]. To obtain a measurement for the physical phenomena in microfluidics which have to be considered in the particular system dimensionless numbers are used [4]. The most well known of these is the Reynolds number Re.

It describes the relationship between inertial effects and effects which are caused by the viscous properties of the medium due to the density of the liquid ρ, its viscosity µ and the characteristic speed ν, as well as the characteristic size of the channel. The behavior of fluids at high Re is more familiar to us, as it can be experienced in daily life. This means that e. g. in rivers, for example, the inertia of the water causes turbulence which results in eddies. In contrast, at low Re, inertia is insignificant, which results in turbulence-free, so-called laminar flows. In such flows, eddies are not caused and layers of liquid move next to each other without intermixing. In closed channels, a parabolic velocity profile develops - in the layer of liquid next to the wall of the channel the velocity is considerably reduced by friction. The next layer is affected to a lesser extent and flows more rapidly. In the middle of the channel the the flow rate is highest. The measurement of the change in flow rate, perpendicular to the streamlines ( z direction) is the shear rate γ⋅ = dv/dz. It is highest near to the wall and reduces to zero at the middle of the channel.

Manufacture with Soft Lithography
While formerly very small channels could only be produced by heating and drawing out glass capillaries, the technique of soft lithography was developed at the turn of the millennium [5]. This enables the quick and cheap development of microchannels with a wide range of geometries (Rapid Prototyping). The manufacturing process illustrated in Figure 1 starts on a computer with the design of the microchannel structure using a vector graphic software. The mask is printed on a special film and then used to selectively expose a silicon wafer which has a photosensitive coating. After development, only the exposed structures remain and form the „channel die". The die is coated with liquid polydimethyl siloxane (PDMS), which solidifies into a rubber-like material, which can be pulled off. After the inlets for the tubes have been punched and the PDMS and the surfaces of the object carrier have been activated with oxygen plasma, the channel can be firmly bonded to the glass object carrier.

NILICS: Separation in Flow
As early as the middle of the 19th century it was observed that in small capillaries, red blood cells (RBC) migrate to the middle of the channel. The physical reason for the migration of deformable objects in a laminar flow with Re < 1 is a repulsive cell-wall interaction known as „non-inertial lift" [6]. The Non-inertial lift occurs due to a discontinuity of symmetry in the laminar flow. This is caused by the disturbance to the flow field by the object and its interaction with the wall. At Re < 1, the necessary discontinuity of symmetry is caused by the deformability of the objec and rigid spherical particles do not experience any lifting force.

The magnitude of the effect on objects such as cells, can be derived from a theoretical consideration of the lift velocity [7]. According to this, a cell migrates with a lift velocity νl

transversely away from the wall with the effective cell radius R, the distance of the cell from the wall z and the dimensionless drift velocity U(λ,s1,s2), which depends on the viscosity contrast λ = ηinout between the cell and the liquid and the cell shaped parameters s1 and s2. The formula shows that an object migrates faster, the larger and more deformable it is and more slowly, the greater the distance from the wall and the lower the shear rate.

This lift effect can be used to sort cells (non-inertial lift induced cell sorting, NILICS) [8]. In a relatively simple microfluidic setup (Fig. 2) cells are added in various concentrations (haematocrit, Hct) to a solution consisting of phosphate-buffered saline (PBS) and Dextran. The sample solution is injected with a syringe into a channel (cross section: 60 x 60 µm, length: 20 mm). A heath flow is driven by a second syringe pump, which initially focusses the cells against the wall of the channel (see x1 in Fig. 2). While they are flowing through the channel, the cells migrate away from the channel wall. According to their properties, the cells experience a different lift force and occupy separate height levels at position x2. The subsequent widening of the channel amplifies distances of the cells from each other (x3) and facilitates their sorting into separate height reservoirs.

Sorting of CTCs
After a detailed characterization of the system, it was able to be used for sorting RBCs and blood platelets [9]. Further investigations were carried out on the basis of circulating tumor cells (CTC) and RBC solutions [8]. CTC are of special interest in medical research, as they offer a possibility for diagnosis without a biopsy of the primary tumor. However, only 1-100 CTC occur in approx. 6 million physiological blood cells, so that these first have to be sorted out for further examinations. To model such conditions for the experiments whole blood samples were diluted (Hct = 1 %, 4 % and 9 %) and mixed with MV3 melanoma cells from a cell culture. The sample solution was injected into the NILICS channel and the separation was monitored with a high speed camera. As they are larger than RBCs, the CTCs are correspondingly further away from the wall (see Fig. 3) at the end of the channel and can therefore be separated out from the RBC solution. With low hematocrit values, a greater sorting efficiency (number of correctly sorted CTCs / total number of CTCs) and greater purity (number of CTCs / number of RBCs in the collected sample) were achieved. As can be seen in Figure 3 and Table 1, these factors became somewhat poorer with a constant flow rate and an increase in hematocrit. This is due to cell-cell interactions, which are more pronounced due to the larger number of cells. The higher number of RBCs results in an increased number of collisions between the CTCs and the RBCs, which hinders the migration of the CTCs. Thus, the CTCs cannot rise as high and some of the RBCs are deflected upwards. However, this could be counteracted by an increase in the sheath flow. The stronger focusing of the cells at the channel inlet results in most of the RBCs being below the CTCs, so that the trajectories of the cells cross each other less often. This is confirmed by the consistently good values for sorting efficiency at higher flow rates and the excellent sorting efficiency of 100 % with Hct = 9%. The cells which are sorted out remain undamaged and can be returned to the culture.

The method presented is attractive due to the elimination of the cell labelling which was previously necessary and the high sorting efficiency in comparison with conventional methods. Due to the simple setup, the possibility of cheap mass production and the complete automation of the test, NILICS could be an approach for routine checks by GPs. The next steps must be the further improvement of the method with regard to the useable hematocrit and direct tests with patient samples. Then this promising method could be investigated with regard to its suitability for everyday clinical use.
The research was supported by the Deutsche Forschungsgemeinschaft DFG 1253 and the Federal Ministry BMBF via FROPT.

Title image by courtesy of Christoph Hohmann (Nanosystems Initiative Munich, NIM).

[1] WHO/IARC, "IARC Press Release No. 224," 2014. [Online]. Available: [Accessed: 14-Feb-2014]
[2] Franke T. and Wixforth A.: Chemphyschem, vol. 9, no. 15, pp. 2140-56, Oct. (2008)
[3] Stone H. A. and Kim, S.: AIChE J., vol. 47, no. 6, pp. 1250-1254, (2001)
[4] Squires T. and Quake S.: Microfluidics: Rev. Mod. Phys., vol. 77, no. 3, pp. 977-1026, Oct. (2005)
[5] Xia Y. und Whitesides G. M.: Annu. Rev. Mater. Sci., vol. 28, no. 1, pp. 153-184, Aug. (1998)
[6] Geislinger T. M. Adv. Colloid Interface Sci., accepted.
[7] Olla P.: Journale Phys. II, vol. 10, pp. 1533-1540 (1997)
[8] Geislinger T. M. and Franke T.: Biomicrofluidics, vol. 7, no. 4, p. 44120, Jan. (2013)
[9] Geislinger T. M. et al.: Appl. Phys. Lett., vol. 100, no. 18, p. 183701 (2012) |




University of Glasgow
School of Physics and Astronomy
G12 8QQ Glasgow
United Kingdom

Register now!

The latest information directly via newsletter.

To prevent automated spam submissions leave this field empty.