# Trustworthiness of IC50 values

My two questions: Is replication of recording $\text{IC}_{50}$ values prior to (primary?) publication less common than performing a reaction several times to determine a more reliable record? Provided the very small quantities used in these tests (vide infra), should a synthetic chemist rather stay skeptic if a newly isolated material is described as highly potent?

In a literature seminar, the story/history of maoecrystal V, a highly congested polycyclic natural product was presented:

(source first part of an overview)

Initially, both the molecular architecture as well as the the $\text{IC}_{50}$ value ($\pu{0.02 ug/mL}$ at a molar mass of $\pu{330 g/mol}$, $\pu{60 nmol}$) in comparison to cis-platin ($\pu{0.99 ug/mL}$ at a similar molar mass of $\pu{300.01 g/mol}$) against cervical HeLa cancer cells reported by Li et al. in Org. Lett. (doi 10.1021/ol0481535) in 2004 triggered the interest of several chemists to synthesize this material.

About a year ago, the Baran group published again efforts towards this compound (doi 10.1021/jacs.6b06623, ACS AuthorChoice / open access), displaying again tremendous efforts in the lab. With larger quantities of the material in hand, however, the initially claimed activity against cancer cells (not only against HeLa) could not be confirmed.

The supporting information of the last paper sheds some light on the frugal scale the $\text{IC}_{50}$ values were recorded:

Synthetic (–)-maoecrystal V (1) was dissolved in 100% DMSO at a concentration of 1 mM. [...] A day after cell seeding, 10 μL of prediluted compound was added to the cells (1:16 dilution).

Which would refer to low umol quantities of the material deployed per test. Still, the question arises why the initially reported activity differs so greatly from the one eventually determined. (Speculation: perhaps fewer groups would have joined the challenge to prepare the material if interested merely in its architecture.)

For the interaction of Maoecrystal V and other, already known, agents for example facing HeLa cells, page S52 of the SI of the 2016 publication shows this plot:

Note that Actinomycin D is considerably heavier ($\pu{1255 g/mol}$) than Maoecyrstal V ($\pu{330 g/mol}$).

• I wish I had time to answer this right now, but generally in natural products chemistry they're awfully unreliable- there just isn't enough material to do the job properly with multiple runs for verification. All too often two groups isolate/make the same thing and report wildly different values, not to mention the huge variety of different assays that groups choose for various reasons which can't be compared anyway. Industry/medchem values, on the other hand are ran many many times over many weeks and generally hold true. – NotEvans. Jul 25 '17 at 23:03
• While I have no experience in recording IC50's, and do not know what level of dexterity is necessary to handle amounts safely, precisely, and accurately (perhaps partially assisted by a robot?) at this small scale, the difference between the curve representing the interaction of Maoecrystal V (e.g. above, in blue; just added) or other agents against HeLa cells is more than expected. However again, this is my perception, as someone who is not in the field; and not aware (beside the note in the 2016 paper) of other instances that an initial determination of IC50 was rectified as here. – Buttonwood Jul 26 '17 at 18:19

I have worked in natural product isolation a little. Although I have never witnessed tests with HeLa cells, I can provide some insight into the general screening procedure.

In the lab I worked in which dealt with isolation of myxobacterial natural products, a novel strain was first tested as a crude extract in a preliminary screening via serial dilution assay. Tests were performed against a set of standard screening organisms including S. aureus, Candida albicans, E. coli and a few others I forgot.

If an isolated showed enough inhibitory activity at this stage, a semipreparative HPLC was performed to locate the peak(s) which caused said activity. The HPLC fractions were collected in a 96-well plate, the solvent removed by nitrogen flow and then the well plate was infected with the organism of interest. Hopefully, one or two wells would remain with out bacterial or fungal growth and hopefully these would not be in the fatty acid realm at the end of the 96 wells.

After that, a slightly larger separation was performed with the goal of gaining a few milligrams of the active compound (after confirming via HR-MS and dereplication that it was not already known) to perform another serial dilution assay to identify the minimal inhibitory concentration (MIC) — the isolationists’ equivalent to an $\mathrm{LD_{50}}$. This was typically performed around three times.

After that, the paper was drafted and submitted to peer review.

(Okay, there are a few steps I left out, but they contain other, unrelated stuff like making sure the myxobacterium was properly classified, elucidating the structure, hoping for crystals and maybe even performing some low-scale fermentation of the relevant myxobacteria to gain enough sample for proper structure elucidation. Ths last step could also preceed the MIC value determination.)

• Interesting! I never realised your group worked in isolation as well as synthesis (jealous...). I would caution your mixing of MIC and LD50 though- both are commonly measured during isolation with reference to different kinds of assays (MIC for things like inhibition of fungal growth, LD/IC50 with reference to the actual killing of cells such as those involved in cancer. Basically inhibition of replication vs cytotoxicity). – NotEvans. Oct 8 '17 at 16:50
• @NotEvans. Yeah, afaik $\mathrm{LD_{50}}$ values were never measured only ever MIC’s (also not $\mathrm{IC_{50}}$, just MIC’s). And it wasn’t exactly my group although we did previously have two people working on isolation. It was the neighbouring group in the same research institute and because of the graduate school basically everyone went there at least once. They also never touched human cells afaik, only ever bacteria and fungi for testing. – Jan Oct 8 '17 at 17:02