Question 20 of 20 Advanced Interpretation
When is a cell line the wrong model, and what would you use instead?
Short answer
A cell line is wrong when the question depends on something the line has lost: tissue architecture, a mix of cell types, normal genetics, or the patient's own biology. Primary cells, organoids, co-culture and in vivo models each restore part of that, at the cost of throughput, control and reproducibility.
Why
Immortalized lines are convenient because they are uniform, cheap and endlessly available. Those same properties come from selection in plastic dishes over many years, which changed them. Ben-David and colleagues documented genetic and transcriptional divergence between strains of the same line, with drug responses that often did not agree.
Lines also lack the structure and neighbors of a tissue, and three-dimensional work shows how much that matters, with oxygen and drug gradients inside spheroids changing which compounds work. Matching the model to the question is the discipline. A biochemical mechanism may be fine in a line.
A question about polarity, barrier function or gradients needs a three-dimensional model. A question about immune interaction needs the immune cells present. A question about patient variability needs patient-derived material, where organoids and primary cells preserve more of the donor's biology.
Each alternative costs something. Primary cells vary between donors and senesce. Organoids are slower, more variable and need matrix that binds compounds.
Animal models add whole-body pharmacology but differ in species biology. The strongest designs usually use a line for mechanism and throughput and confirm the key claim in a model that restores the missing feature.
A strong answer covers
- Lines are selected products of long culture
- Architecture, mixed cell types, genetics and donor variation
- Match the model to the feature the question depends on
- Each alternative trades throughput for realism
- Use lines for mechanism and confirm elsewhere
Common mistakes
- Treating a patient-derived organoid as automatically more predictive without validating it
- Using one cell line as a stand-in for a whole tumor type
- Choosing a complex model when the question is purely biochemical
The interviewer might ask next
How would you decide which single result from a cell line study is worth repeating in a patient-derived model?