Making Research human-relevant

Lately I’ve been thinking about how to make cell culture more human-relevant: What if all our drug discovery pipelines used physiological glucose levels instead of the 25 mM you find in standard DMEM?

I suspect it would slow things down in the short-term; slower proliferation and poor adaptation of some lines would mean longer experimental timelines.

But I imagine the long-term upside would be pretty massive, because right now, we’re actually doing the opposite. Standard DMEM has 25 mM glucose, while human blood sits around 5 mM. That’s a five-fold difference, and enough to push cells into metabolic states they’d never see in vivo.

At these supraphysiological levels, cells often lean heavily on glycolysis even when oxygen is plentiful. That can mimic a “Warburg-like” phenotype, but one that may say more about the media than the cancer.

The consequences are easy to imagine:

- Drug sensitivity profiles that don’t translate to patients
- Biomarkers that are actually cell culture artifacts
- Potential therapeutic targets discovered under hyperglycemic stress that turn out to be irrelevant in vivo

The result is years of expensive R&D on a series of targets that doesn’t efficiently lead to new therapeutics.

And here’s the kicker: cells long adapted to 25 mM glucose often struggle when you dial back to physiological 5 mM. That struggle itself tells us something important. We’ve been optimizing our models for convenience and fast proliferation, at the expense of human relevance.

Of course, it’s not as simple as “5 mM glucose = human-relevant biology.” Tumors often experience lower glucose than blood, and glucose is just one piece of the puzzle.

Still, I can’t help but wonder if we had standardized around human-like conditions decades ago, how many “failed” drugs might have never made it past the cell culture dish and how many more real opportunities might have been identified sooner?

Maybe the future includes testing across a panel of media which includes physiological plasma-like media as well as tumor-mimicking low-nutrient media.

It would be slower and messier…but possibly much more efficient for patching up our leaking drug development pipelines, which would ultimately make it more cost-effective.

Anyway, I realise it’s a possible future rather than an easily solved problem today, but I’m curious to hear any thoughts, especially from folks involved with these sorts of large scale drug discovery screens.