Breaking Bad... Controls. How to do it correctly.

Why “home-made” IHC controls are a bad idea — and how standardized materials save diagnostics.

In immunohistochemistry, every error can have real clinical consequences. No surprise there: IHC determines treatment, eligibility for targeted therapies, tumor diagnostics, and even risk assessment. And yet, in many laboratories, in-house controls still reign — cut from patient blocks, assembled into tissue arrays, or “put together” from whatever happens to be available. Nobody pays attention to the fact that 50 micrometers deeper, the immunophenotype is no longer the same, or the level of expression is completely different. Few people notice that controls should also be used for CISH and FISH. And even if they are used, it becomes very difficult to interpret such a non-standardized control in a test of such clinical importance.

Sounds familiar?

Home-made control production may look like a clever and cheap way to achieve quick results… until everything starts to fall apart.

1. In-house controls — why are they a controlled disaster?

1.1 Sample heterogenity

What is the biggest problem?The fact that every slice from the same block can look different — literally. We cannot be sure that just a few cuts deeper, the tissue will be identical in terms of immunophenotype.

Cut 20–50 µm deeper and you may find:

• the expression is different,

• the intensity drops,

• the antigen is no longer present,

• or you end up in a completely different cell population.

This absolutely eliminates the possibility of reproducibility, comparing runs, using the material for quantitative markers, or applying it in AI and digital pathology — all of which are steadily becoming the gold standard in diagnostics.

1.2 Lack of standardized expression

Tissue material from a patient is never “factory-standardized.” NEVER.

Expression is influenced by factors such as:

• the underlying disease,

• the condition of the tissue,

• fixation,

• storage,

• and most importantly — very often, pure coincidence.

  
  

Effect? You simply cannot treat this as a reliable control. How are you supposed to use such a control in FISH or CISH? How do you even know whether the test worked or not?

Very often in FISH there is no positive or negative control at all, and when some “control” is used, it has no real value, because the control section on each subsequent slide will yield a completely different result. So is any control better than none? As always — it depends.

For standard markers, where the tonsil or appendix is an ideal control material, it is indeed better to use some control than a fully standardized one. However, for tests like HER2, PD-L1, melanoma markers, MMR, or estrogen and progesterone receptors — in other words, everywhere the result directly determines further treatment — using just “any” control is not really better than using none at all.

It is therefore surprising that companies providing IHC solutions, for example, deliver 10 or 20 control slides for 1,000 tests. Is this due to “savings,” or is there a fear that it might turn out the system is not as perfect as advertised? More on that in a moment.

1.3 The cost everyone underestimates (or ignores)

If you think making your own control block in the lab is much cheaper, you are seriously mistaken. Why?

The cost includes the work of the administrative staff who must locate suitable cases, the technician who retrieves the blocks, the pathologist who reviews the slides and marks the areas to be used, the technician performing trial stains (if needed), and the technician creating the tissue array — punching cores, assembling the paraffin block, embedding, and finally cutting it. Quite a lot of people and quite a lot of work.

The biggest surprise, however, is that many facilities or hospitals focus heavily on keeping operating costs low, while at the same time increasing them by allowing homemade controls and assigning highly qualified personnel to tasks that are not diagnostic work.

Aside from personnel issues, there is the risk that TMA cores fall out, that embedding errors occur and the order gets mixed up, that one of the cores is exhausted after a few sections, or that you produce 70 slides only to discover the block is unusable — and you must start all over again.

When you calculate the real labor cost, in-house control production turns out to be one of the most expensive and least efficient forms of quality control in IHC (and this excludes the simplest case of taking a random tonsil fragment and placing it on a slide). Realistically, when everything goes very smoothly, the cost is around 300$ per block, assuming average hospital salaries. That is 3$ added to every IHC stain.

But that’s if everything goes smoothly. If the selected block turns out to be unsuitable for a TMA, or there is very little material left, everyone must spend additional time. Twice the work means twice the cost.

So can a pathology department afford to be “frugal” in this way? Not really. First, nearly all institutions face shortages of both technical and specialist staff. Second, not everyone even has access to tissue that can legally and ethically be used as control blocks. And according to guidelines, controls must be used.

1.4 Variable quality of tissue sections

Grooves, folds, and tears — each of these problems forces additional cutting. Home-made arrays have a much higher failure rate than professional materials. They are unsuitable for digital pathology, particularly if you intend to use them for calibrating digital systems. AI systems require uniformity, reproducibility, and stability. In-house material meets none of these criteria.

On the other hand, not everyone invests in digitization. But that does not mean variable quality doesn’t affect such laboratories — quite the opposite. Changing tissue structure disrupts our perception of staining standardization. When we constantly view slides with the same expected staining intensity and suddenly something deviates, we can immediately detect the problem — something that is impossible when one control is strong, another weak, and another completely negative.

How are we supposed to know that a patient-derived control tissue still has the same expression, or whether it has any expression at all?

With standardized positive control tissues, the result will be the same every single time.

2. Why do standardized, commercial controls solve all these problems?

I wanted to use an example about building a car, but a home renovation is probably more relatable. It seems cheaper when we decide to paint the walls ourselves, right? We buy paint, a roller, and off we go. We paint the first coat. When applying the second one, it turns out the previous layer sticks to the roller. We didn’t prime the walls, so now we have to scrape off the peeling paint and go to the store to buy primer. We prime and paint again. After the first coat dries, we realize we ran out of paint. We rush to the nearest store because we’re short on time. We apply the second coat and suddenly the shade is wrong — different batch, or maybe it wasn’t exactly the same product and instead of “Creamy Morning” we accidentally bought “Pleasant Sunrise” (paint names never have anything to do with their actual color). So we go to the store where we bought the first can, and of course that store is across town. Traffic jams, delays — the whole trip takes half a day.

That’s what work looks like when we’re not fully knowledgeable about it. Now imagine we pay a few thousand for a specialist. They arrive, put down protective foil — something we forgot, so our entire floor now needs scraping, just like the windows and anything else nearby — they tape the edges, mix primer with paint, use a spray gun, and in five minutes the entire wall is primed and evenly painted white. In this situation EVERYONE will say we paid a professional — otherwise we’d have kept those few thousand in our pocket. But think about this: during that same time, a pathologist or cytologist could have been reviewing slides instead of spending a week with a roller. We never pay attention to time, even though time is the most valuable resource. Yes, we could have earned money during that time — but we could also have gone for a walk in the forest or taken a vacation.

This is what happens in every laboratory where a specialist ends up doing tasks outside their actual expertise. And now, returning to our QC...

2.1 Homogenity

Professional controls are based on stable, reproducible cell lines or standardized reference materials. This means that every section — from the first to the last — looks identical. Something that is simply impossible to achieve with a hand-made tissue array.

2.2 Defined, reproducible antigen expression

These controls are designed to have:

• a known level of expression,

• stability over time,

• reproducibility across batches.

That is why they are suitable for tests such as HER2, PD-L1, MMR, Ki-67, digital pathology, and validation of quantitative protocols.

From a single block you can produce up to 450 identical sections (depending on the manufacturer).With in-house controls, this is absolutely impossible.

2.3 Minimization of technical risk

Professionally prepared blocks are:

• resistant to core loss,

• perfectly embedded,

• easy to section,

• predictable.

All these factors ensure that a control block will have the highest possible performance: it will require far less trimming, sections will not tear, and—most importantly—all cellular regions will have the same thickness and will last for the same number of tests.

Given all these arguments, one essential question remains: who validates the IHC systems themselves? In many, if not most laboratories, IHC control is either nonexistent or performed in a way that makes standardization impossible. Everyone focuses on checking whether the instrument stains at all. But hardly anyone checks whether it stains properly or whether the instrument itself is also standardized.

Quality control should be performed not only for the assays, but also for the devices used to perform them. Does that make no sense?Well then— let's put the cat among the pigeons.

3. Problem różnic między stanowiskami w instrumentach automatycznych

Studies have shown that in some IHC systems (e.g., multi-position automated stainers), staining intensity can vary by several dozen percent depending on the slide’s position in the instrument. In one of the most commonly used systems, the differences reached up to 40% within the same device.

This means that depending on where the slide is placed, a patient may or may not qualify for treatment. It is worth asking ourselves whether we would want to be in that patient’s position.

What does it mean?

Without standardized control:

• HER2 3+ can appear as 1+,

• PD-L1 50% can appear as 10%,

• MMR may look lost when it isn’t,

• a patient may receive the wrong treatment — or be denied the treatment they need.

This is not a technical detail. It is a serious clinical problem that should be verified every single time.

3.1. Why are differences between slide positions truly a problem?

Visiopharm and hard data instead of “looks OK.”

For years, many laboratories have relied on visual assessment in IHC — the human eye, the pathologist’s experience, and a “common-sense” comparison.

The problem is that the human eye cannot detect subtle differences in staining intensity, and quantitative staining (HER2, PD-L1, Ki-67, MMR and others) does not tolerate even a 10% deviation, let alone 30–40%.

This is precisely why it is worth referring to systems such as Qualitopix — an analytical platform that uses:

• computer-based image analysis algorithms,

• AI-driven cellular segmentation,

• photometry-based staining intensity measurements,

• validated models for quantitative marker assessment.

This is not “eyeballing it.”It is digital laboratory photometry, measuring every pixel, every nucleus, every bit of DAB intensity.

In one comparative test at UMC Utrecht, which evaluated the performance of various slide positions in a commonly used IHC stainer, Visiopharm detected differences in staining intensity even when using the same control slide, stained in the same run, on the same platform. It is worth noting that the IHC system in question had been in use for years, and throughout those years no one noticed that something might be wrong.But what if standardized controls had been used?What if AI had been monitoring quality?We would have no doubts whether a result issued five years ago on system X might have excluded a patient from a life-saving therapy.

In short, an in-house control, being inherently variable, has no chance of detecting instrument-related inconsistencies.Only a standardized, homogeneous control allows us to see whether the instrument is decreasing, increasing, or otherwise altering staining intensity.

3.2. What does this mean in practice?

Without a stable control:

• the algorithm has no reference point,

• positional differences go undetected,

• laboratory deviations can appear as “normal staining,”

• quantitative results become incomparable,

• and the risk of incorrect patient classification increases — significantly and realistically.

Qualitopix has demonstrated, in hard numbers, what the naked eye cannot see. But to prove anything, you need a reference point — and that role can only be fulfilled by a control material with stable, known, and homogeneous expression.

4. New panels that will ignite the future of diagnostics

Professional control systems increasingly include:

• MMR panels (MLH1, MSH2, MSH6, PMS2) — essential for colorectal cancer and MSI diagnostics, and

• melanocytic panels — valuable in melanoma and challenging lesion diagnostics.

This is the direction in which global diagnostics is heading.

Summary: It’s time to stop “cooking up” controls in the basement

In 2025, IHC requires stability, reproducibility, quantitative comparability, consistency between runs, and resistance to instrument-related variation. In-house controls do not meet these requirements. They introduce risk — organizational, technical, clinical, and very likely legal, no matter how you look at it.

Modern, standardized controls and quality-assurance systems solve all these problems in one move. They provide certainty of results, patient safety, and the level of reproducibility that is now a necessity rather than a luxury. I encourage you to get in touch and discuss this further.