|dc.description.abstract||This study aims for mathematical modeling, experimental validation, and systematic optimization of Expression Microdissection (xMD). xMD selectively heats targeted cancer cells via a photothermal effect to enable their procurement from tumor tissue samples to provide highly pure cell populations for molecular analysis so that misdiagnoses caused by non-neoplastic cell contamination can be reduced. Several theoretical models have been validated for the photothermal effect in tissues. However, these models are not generally applicable to the physics behind the process of xMD. In this study, we proposed a mathematical model that analyzes the spatial and temporal temperature distribution and heat melt track in an xMD sample that is composed of a thermoplastic film and a tumor tissue section sandwiched by two glass slides. To experimentally validate the model, we designed and built a continuous-wave laser raster system from scratch to implement xMD on the sample in which the tissue was replaced by a tissue-mimicking phantom fabricated by spin-coating. The phantom is used to imitate the physical properties of an immunohistchemically-stained tissue, such as thickness, light absorption, and scattering, etc. Moreover, we proposed an indirect method that uses absorbance spectral slope of the xMD-treated film as temperature indicator for the sample in order to overcome the challenge of temperature measurement on a multilayered micro-scaled medium and experimentally validate the model. The result shows that the experimentally measured temperature of the phantom and melt track width on the film were in good consistency with those predicted by the model.
Furthermore, based on the validated model, we systematically optimized the xMD process under realistic tissue variations for commercially-available laser raster systems, featuring two laser types (pulse v.s. continuous-wave) and two system configurations (top-down v.s. bottom-up). Specifically, we analyzed the temperature distribution of the xMD sample under three cases of the variations: (1) size of the stained cancer cell, (2) tissue section thickness, and (3) tissue stain intensity to find the optimal xMD operation space (i.e., laser intensity, scan speed, and pulse-on time) for the systems. In the optimization results, the optimal laser intensity and pulse time of the pulse xMD system for the sample variations range from 6 x 10^7 W/m^2 to 13 x 10^7 W/m^2 and from 2 ms to 10 ms, respectively. However, over-melting problem may occur when dealing with thicker tissue samples. The result suggests the pulse time of less than 0.8 ms. Similarly, for the continuous-wave xMD system, the optimal range of the intensity and speed are from 7 x 10^7 W/m^2 to 1 x 10^8 W/m^2 and from 60 mm/s to 100 mm/s, respectively. These ranges are overlapped by the specifications of the continuous-wave systems, indicating they are capable of processing the samples in real clinical practice. Furthermore, no obvious difference of the optimal range can be seen between the laser systems of the two configurations when extracting cancer cells from the thin tissues (5 um). When processing the think tissues (15 um), our simulations however show that the bottom-up pulse xMD system has better heating efficiency in tissue than the other xMD systems do, indicating it has smaller optimal operation window. Additionally, the top-down xMD system induces higher temperature on the film/tissue interface. Such result points out that the top-down xMD system can provide better xMD performance than the bottom-up system does.
Our model demonstrated its validity to describe the xMD mechanism. The optimization results revealed the optimal xMD range for the varying realistic tissue samples. We anticipate the xMD model and parametric simulations enable researchers to facilitate the cell retrieval process and maximum the xMD performance without contaminating subsequent molecular profiling of cancer and other diseases so that cancer patients can receive molecular medical treatments in a timely manner.||en_US