It’s not unheard-of to use a cell’s shape to inform a diagnosis. The classical example is sickle cell disease, where the red blood cells have an unusual, crescent-like shape, but other such disorders exist. What if we could use cell shape to not only detect disease, but also predict its course? That’s the question Elaheh Alizadeh, Ashok Prasad and their colleagues took on in their recent paper on the shapes of cancer cells (1).
We already know that the disease process of cancer involves misregulation of the cytoskeleton, which leads to cell deformity. What Alizadeh and Prasad hypothesized is that, before such changes become visible to a pathologist at the microscope, there may be much smaller, subtler changes detectable only with the aid of a computer. To decipher these potential changes, they selected a set of 256 individual cell shapes and characterized each one – and the differences between them – mathematically. Next, they used four known osteosarcoma cell lines with varying degrees of invasiveness to teach a computer to distinguish between different degrees of aggression based on shape. Only once the computer was fully trained did they expose it to a hypothetical fifth cell line – and discovered that it accurately predicted the invasiveness of the new cells. What’s next? A lot more work. The scientists need to examine better ways of preparing cells for imaging so that their shape is not impacted, as well as exploring new methods of evaluating and quantifying cell shape. Once the method is refined to the point where it can reliably be used to link shape with potential prognosis, might it give pathologists another perspective on disease.
Newsletters
Receive the latest pathology news, personalities, education, and career development – weekly to your inbox.
References
- E Alizadeh et al., “Measuring systematic changes in invasive cancer cell shape using Zernike moments”, Integr Biol (Camb), 8, 1183–1193 (2016). PMID: 27735002.