Cell health check: How often should I karyotype?

21 September 2026

It is good practice to karyotype a cell line periodically. For many labs, this happens when the line arrives, or when it is first banked, the report comes back clean and goes into a folder. Several years and a few hundred passages later, the same line is still in use and nobody has looked at its chromosomes since.

Everyone accepts that lines drift, but the guidance that does exist is spread across banking standards, journal policies and supplier FAQs, most of it written for pluripotent lines even though immortalised and edited lines have the same problem.

Here we define the points where testing is obviously worth it, the interval to use in between, and the things that should send you to the cytogenetics laboratory whatever the passage count says.

Why Once Is Not Enough

Selection is what breaks the single-baseline approach. The abnormalities that recur in culture recur because they help the cells that carry them, usually by improving survival through passaging, so the abnormal subclone is growing faster than the normal population it sits in from the moment it appears.

Abnormalities therefore do not creep up on a culture. They stay below the detection limit of every available assay for a while, then take the population over in a handful of passages. Nothing looks wrong at the bench during the transition.

The 20q11.21 gain in human pluripotent lines is the usual illustration. It carries BCL2L1, which reduces apoptosis on passaging, and it is small enough that G-banding can miss it, which is partly why the field talks about recurrent abnormalities as a monitoring problem rather than a testing problem.

So if the last karyotype was at passage 20 and the culture is now at passage 45, that report is not telling you much about what is in the flask.

The Fixed Decision Points

Some points in a workflow are worth testing at whatever the line has been doing, because that is where the damage from a wrong assumption concentrates.

Decision points for cytogenetic testing
Point in the workflow	Why here	Cost of skipping it
On receipt of a new line	Establishes your own baseline rather than inheriting someone else's, and identifies constitutional rearrangements present from derivation	Any later abnormality cannot be attributed to your culture or theirs
After reprogramming or derivation	Reprogramming is a bottleneck that both generates and selects abnormalities	An entire clone selection exercise built on an abnormal founder
After gene editing or single-cell cloning	Nuclease activity and the clonal bottleneck both raise the rate of structural change, including events no dosage panel targets	A phenotype attributed to the edit that belongs to an unrelated rearrangement
At master cell bank creation	The bank is the reference point everything downstream returns to; it needs a formal description	Every vial in the bank inherits an undocumented problem
At working cell bank creation	Confirms the expansion from master to working bank did not select something	A working bank that no longer matches its master
Before a long or expensive experiment	Differentiation campaigns, animal work and screening runs commit months of effort to whatever is in the flask	Months of data that cannot be interpreted or published
Before transferring a line to a collaborator	A current report travels with the line and settles later disputes about where a problem originated	An unresolvable disagreement between two laboratories
Before submission or publication	Increasingly requested by journals and reviewers, and confirms integrity was retained across the work	A revision request that cannot be answered because the cells are gone
At regulatory or GMP milestones	Required as part of the characterisation package for clinical-grade lines	A gap in the dossier found late, when remediation is most expensive

Collaborator transfer and the pre-publication check are the two that get skipped most, and both are inexpensive next to what they cover. A line arriving at a partner institution with a current karyotype attached cannot later be blamed for whatever happens there, and a report generated before submission heads off a reviewer question that is becoming standard.

The Routine Interval

In between those points it becomes a question of frequency. For human pluripotent lines the figure most often quoted is every ten passages, and several groups argue for five to ten on the grounds of how fast a recurrent variant can sweep a culture. The ISSCR standards put it in terms of covering the timespan of the experiments themselves, so that a change is caught while the work is still running.

Ten passages is a default rather than a rule, and several things pull it in one direction or the other.

Pluripotent lines are the least stable in ordinary culture and deserve the tightest interval. Immortalised and cancer lines are often aneuploid to begin with and go on evolving, so the useful question there is whether the line still matches its own documented karyotype, not whether it is normal. Primary cultures are a different case again: they run out of passages before drift becomes the dominant risk, and testing usually attaches to the experimental window instead.

Culture stress shortens the interval. Enzymatic single-cell passaging, marginal media, repeated freeze-thaw, any stretch of poor growth. A line that has had a rough month is worth checking early rather than waiting for the count to come round.

So do the stakes. A line feeding a screening campaign or a clinical programme carries a larger cost when a problem surfaces late, and that alone justifies testing it more often than a line being used to explore an idea.

Worth considering first: the cheapest way to test less often is to passage less far. A working bank that is actually returned to, rather than a single culture carried forward month after month, holds the line inside a narrow passage window and makes most of this scheduling question go away. It is not always practical, particularly for lines that recover badly from freezing, but a laboratory karyotyping constantly is sometimes paying to compensate for a banking strategy it never set up.

Unscheduled Triggers

A schedule only covers what you can anticipate. The observations below are worth acting on whenever they turn up, since each of them is a late symptom of something that happened several passages ago.

Growth rate has increased and the culture now needs passaging sooner than it used to. Counterintuitively, cultures that suddenly look healthier are a classic warning sign, because the abnormalities that take over are the ones conferring a growth advantage.
Differentiation efficiency has dropped, or the line has become resistant to a protocol that previously worked.
Morphology has shifted: altered colony edges, changed packing density, a different appearance at confluence.
The line has become noticeably easier to passage or more tolerant of dissociation than it was.
Results have become variable between operators or between vials with no procedural explanation.
A screening assay or copy number panel has flagged anything, which tells you something has changed but not what it is.
The line has recovered from an incident: contamination, an incubator failure, a mislabelled flask, a period of neglect.
An experiment has produced a result that contradicts established behaviour of that line.

The first of those deserves repeating, because everything about it feels wrong. A stem cell culture that has suddenly started growing beautifully is rarely a reward for better technique.

Building Your Schedule

The tool below builds a rough schedule from line type and intended use. Treat the output as a framework to argue with rather than a specification.

1. What kind of line is it?

2. What is it being used for?

Select a line type and an intended use to build a suggested schedule.

Guidance is indicative and drawn from published banking and research standards. Intervals should be adapted to the behaviour of the individual line and the requirements of the work.

Testing at the Right Depth

How thoroughly to test is a different decision from when. Twenty metaphases is the standard count, and it rules out mosaicism above roughly 14%, which does for routine monitoring most of the time.

Ask for more cells when a small abnormal population would be expensive to miss: banking a line headed for clinical work, committing to a long differentiation run, or following up a screening flag. Thirty cells takes the figure to around 9.5% and fifty to around 5.8%. Sensitivity you can buy outright is unusual in a laboratory, and this is one case where the money does something.

Depth of method follows the same reasoning. For edited lines, and for anything heading towards the clinic, array CGH run alongside banding covers the imbalances that sit under the resolution of G-banding. Which method answers which question is the subject of our article on PCR versus karyotype.

What a Schedule Looks Like in Practice

Indicative testing schedules by line type
Line type	Routine interval	Additional points
Human iPSC and ESC	Every 5 to 10 passages	Post-reprogramming, both bank stages, before differentiation campaigns, before publication
Mouse ES cells	Every 5 to 10 passages	Every clone before microinjection; the cost of a failed germline transmission round dominates the testing cost
Edited or single-cell cloned lines	Every 5 to 10 passages	Every clone taken forward, before and after the edit, since the bottleneck matters as much as the nuclease
Immortalised and cancer lines	Every 15 to 20 passages, or annually	On receipt, at banking, and whenever behaviour diverges from the documented characteristics of the line
Primary and finite lifespan cultures	Tied to the experimental window	At derivation and at the end of the window; metaphase yield can be a practical constraint
Any line on a clinical or GMP path	Tightest applicable interval	Full characterisation package with extended counts and array CGH; every bank stage documented

What to Do With a Result

Two habits make the difference between a folder of reports and something you can act on, and neither costs anything.

Write down the passage number, the date, and which bank and vial the culture came from, next to the report reference. An abnormal finding can then be pinned to a window of passages instead of to a vague stretch of last year. A karyotype floating free of a passage number is close to useless when you need it.

Then keep vials from each point you tested. The reason to monitor at all is to be able to go back to a vial from before the trouble started, which only works if the vials and the reports point at each other. Plenty of laboratories monitor conscientiously, bank casually, and find out at the worst moment that they have nowhere to retreat to.

Summary

The short version is that most laboratories should be testing more often, and at particular moments rather than when it occurs to somebody. A baseline on receipt or derivation, a test at each bank stage, one after reprogramming or editing or single-cell cloning, and one before anything long, costly, leaving the building or going to a journal. Between those, a routine interval: tight for pluripotent and edited lines, looser for immortalised ones that have been stable, and attached to the experimental window rather than a passage count for primary cultures.

Then there are the triggers that ignore the schedule entirely. The one to remember is that a culture which has started growing unusually well is asking to have its chromosomes looked at.

None of this settles how often is often enough for any particular line, and the guidance available will not settle it either. What it does is put the decision somewhere you can defend it, which is usually the most that can be asked of a monitoring plan.