Chapter 48: Clinical Biochemistry Applications

Welcome to Last Minute Lecture.

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement not replaced the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

If you've ever used one of those simple at -home tests or had a really detailed blood panel done at a hospital, you've interacted with this essential bridge between biochemistry theory and just practical, everyday medicine.

That's so true.

We tend to just look at the numbers, but behind every single one is the whole field of clinical biochemistry.

Our source today, Chapter 48 from Harper's, really digs into what these tests measure, how we even decide what normal is, and how we prove they're valid in the first place.

So we're diving deep into the logic of lab science.

The mission today is to really get how these tests are designed and used to track our health.

They're just so critical for everything from screening to confirming a diagnosis.

Absolutely.

And we're mostly talking about blood and urine samples, but sometimes it could be cerebrospinal fluid or even a tissue biopsy.

And the technology has come so far.

I mean, so many tests are automated now, done right at the bedside.

Right.

Point -of -care testing has been a revolution.

But for the really specialized stuff, like screening for rare metabolic diseases or complex drug tests, you still need those licensed specialist senders to get it right.

OK, so let's start with the big picture.

You get a lab report and a result is flagged as abnormal.

What's actually happening biochemically to cause that flag?

Well, it usually boils down to one of three things.

The first one is tissue injury or damage to the cell membrane.

The cells are just breaking open and leaking.

Exactly.

They become permeable and just dump their internal contents into the blood.

The classic example here is creatine kinase MB.

When that shows up in the blood, it's a huge signal that heart muscle tissue has been damaged, you know, like in a heart attack.

OK, so leakage is number one.

What's the second cause?

The second is altered synthesis.

This just means the body is either making way too much or way too little of something.

A great example is C -reactive protein or CRP.

The inflammation marker.

Right.

When there's systemic inflammation, the liver just starts pumping out CRP.

So a high CRP level is a biochemical signal that something's wrong.

And the third one feels pretty intuitive, just straight up organ failure.

Precisely.

If an organ can't do its job, the stuff it's supposed to process just builds up.

If the kidneys fail, creatinine accumulates because it can't be cleared.

If the liver fails, bilirubin builds up.

There are immediate chemical flags that an organ isn't working.

That really sets the stage.

But before a doctor can even call something abnormal, the lab has to define normal.

And this is where the statistics get a little tricky.

How do they set that baseline, the reference range?

It's a purely statistical exercise.

And it has to be, because our values vary so much depending on age, gender, even the time of day.

So they take results from a huge healthy population.

And just find the average.

They find the average or the mean, and then they define the range as that mean, plus or minus two standard deviations.

So plus or minus two SD.

What does that actually capture statistically?

It's set up to include 95 percent of that healthy population.

But this leads to a massive clinical headache.

You're talking about the five percent problem.

Exactly.

Just by pure chance, five percent of perfectly healthy people will have a result that falls outside that normal range.

And this became a huge issue in the 1970s when labs got multichannel analyzers.

The machines that run like 20 tests at once.

Yes.

So if you run 20 different tests on one healthy person, the probability that at least one of them is going to be randomly outside that 95 percent range is suddenly very, very high.

You've just generated an abnormal result that might be totally meaningless.

That must cause so much anxiety and unnecessary follow up tests, all from statistical noise.

Oh, absolutely.

It's the constant struggle separating the signal from the noise.

That's why labs will often report something called a Z score.

And what's that?

It's just the number of standard deviations the result is from the mean.

It gives a doctor context.

A Z score of, say, two point one is just barely outside the range.

A Z score of eight point zero is a genuine five alarm fire.

It tells you how abnormal the result is.

OK, that makes sense.

So once we've got the statistical range sorted, we have to trust the test itself.

When a lab brings in a new method, there are four big questions they have to answer.

Let's start with the first two, precision and accuracy.

Right.

So precision is all about reproducibility.

If you test the same exact sample 50 times, how tightly clustered are the results?

High precision means very little scatter, a low standard deviation.

It's like throwing darts.

Precision means all your darts hit the same spot, even if that spot isn't the bullseye.

That is the perfect analogy.

And accuracy is the other half of that.

It's how close your result is to the true value.

So getting your tightly clustered darts to actually hit the bullseye.

Exactly.

You can be very precise, but not accurate at all.

All your results are grouped together, but they're way off the real number.

But how does one lab know what the true value even is?

How do they find the bullseye?

They use external quality control schemes.

Basically, a central agency sends out identical pooled samples to hundreds of labs.

The average result from all those trusted labs becomes the accepted true value.

Then each individual lab can check its own accuracy against that national standard.

OK, so that covers precision and accuracy.

What are the other two criteria for the method?

They are analytical sensitivity and analytical specificity.

Sensitivity is pretty simple.

What's the smallest amount of a substance that the test can reliably detect?

This is a huge deal for things in trace amounts like narcotics or performance enhancing drugs.

And specificity is about making sure you're measuring only the thing you want to measure.

The book has a really great example for this with glucose testing.

It's a perfect illustration.

The old obsolete methods used alkaline copper reduction.

But the problem was other things in your blood besides glucose can reduce copper like vitamin C or other sugars.

So you'd get falsely high glucose readings because the test wasn't specific.

Exactly.

The modern method uses an enzyme, glucose oxidase, and that enzyme is incredibly specific.

It really only reacts with glucose.

So the result is much cleaner.

But even that super specific enzyme test can be fooled, right?

There's a catch.

There is.

And it's a fantastic teaching point.

The glucose oxidase reaction produces hydrogen peroxide.

A second reaction then uses another enzyme, peroxidase, to make a colored dye that we can measure.

But if a patient is taking massive doses of vitamin C, it interferes with the second step.

The vitamin C acts as a reducing agent and can actually turn the colored dye back to colorless.

So you end up with a falsely low or false negative result.

Wow.

So even with a specific enzyme, a different chemical can mess up the measurement part of the test.

It's huge.

It is.

Now, once the method is validated, we have to switch gears and think about how the result is used clinically.

And this means we have to completely redefine sensitivity and specificity.

Right.

We're moving from analytical quality to clinical usefulness.

So let's define these new terms.

OK.

Clinical sensitivity is the percentage of positive results you get in patients who actually have the disease.

It's a measure of true positives.

And clinical specificity.

That's the flip side.

It's the percentage of negative results in people who truly do not have the disease, the true negatives.

And some tests are amazing at this.

The newborn screening for PKU, for example, is almost perfect.

It's like 100 percent sensitive and 99 .9 percent specific.

It finds everyone who has it and almost never gives a false alarm.

But most tests aren't like that.

Take the CEA test for colon cancer.

Its sensitivity in early disease is really low, only about 20 percent.

And its specificity is all over the place.

It's 97 percent in nonsmokers, but drops to 80 percent in smokers.

So one in five healthy smokers will get a false positive.

That's that's a lot.

It is.

And this brings us to this unavoidable tradeoff.

The clinician has to decide where to set the cutoff for a positive result.

If you set that cutoff really high, you'll be very specific.

Very few false positives.

But you'll miss a lot of people who actually have the disease.

You lose your sensitivity.

You lose your sensitivity.

But if you set the cutoff very low, you'll catch almost everyone.

High sensitivity.

But you'll swamp the system with false positives.

Your specificity goes way down.

So you're always balancing the cost of a false negative against the cost of a false positive.

And this choice really impacts something called the test's predictive value.

Yes.

The positive predictive value, the chance that your positive result is a true positive depends hugely on one thing.

The prevalence of the disease in the group you're testing.

Why is prevalence so important?

Think about it.

If you use a test for a very rare disease on the general population, even a good test will probably generate more false alarms than real cases.

But now take that same exact test, let's say for creatinine, and use it in a urology ward where kidney disease is common.

The prevalence is high.

The prevalence is high.

So now a positive result is much, much more likely to be a true positive.

The context completely changes the meaning of the number.

Let's pivot to some technical stuff, starting with urine samples.

It's interesting that you can't just report the concentration per milliliter.

No, because your fluid intake changes all the time.

So the volume of your urine is constantly changing.

To compare samples over time, you have to normalize it.

We report the amount of the substance per mole of creatinine.

Because creatinine excretion is pretty constant for one person.

Right.

It's a breakdown product from muscle.

So based on your muscle mass, you excrete a fairly steady amount each day.

It's a perfect internal standard to correct for hydration.

And in blood, quickly, what's the difference between serum and plasma?

So they're both the liquid part of blood.

To get serum, you let the blood clot first, then you spin it down.

So serum has no clotting factors left in it.

To get plasma, you add an anticoagulant before it clots.

So plasma still has all the clotting factors.

OK, so what tools are labs actually using to get these measurements?

The workhorse is absorption spectrophotometry.

You run a reaction that produces a colored product and then you shine a light through it.

The amount of light it absorbs is proportional to the concentration.

It's simple and reliable.

And for things that need more sensitivity.

Then you use fluorimetry.

Instead of measuring absorbed light, you excite the sample with light of one color and you measure the different color of light that it emits.

That fluorescence is incredibly sensitive.

This brings us to a really cool part of the chapter.

The three different ways enzymes are used in the lab.

Let's start with rule one.

Using the enzyme as a regent to measure something else.

Right, like using glucose oxidase to measure glucose.

The key principle here is that the enzyme and all the other regions must be in huge excess.

The thing you're trying to measure the analyte has to be the only limiting factor in the reaction.

And the kinetics are really specific here, right?

In relation to the enzymes, Michaelis constant, the gum oliglory.

Exactly.

You want to run the assay where the analyte concentration is well below the enzymes cumulator.

If you look at the reaction curve, that's the steep part at the beginning.

A tiny change in the analyte causes a huge measurable change in the reaction rate.

That gives you maximum sensitivity.

So that's rule one.

Enzyme is in excess, analyte is the bottleneck.

What about rule two, where the enzyme itself is the analyte you're measuring?

This is for measuring tissue damage, like we talked about with creatine kinase.

Here, the logic completely flips.

We want to measure how much enzyme has leaked out.

So the enzyme itself must be the limiting factor.

So you have to flood the system with the substrate.

You flood it.

You add the substrate in massive excess, something like 20 times the Guller -Allers.

This forces the enzyme to work at its absolute maximum speed.

It's V max dollars.

At that point, the reaction rate depends only on how much enzyme is there.

That's such an elegant switch.

It's a beautiful piece of biochemical logic.

OK, finally, rule three, enzyme activation assays.

This is for measuring vitamin status, right?

Yes.

And it's a measure of functional status.

Many enzymes need a coenzyme that's derived from a vitamin.

So what you do is you take a patient's red blood cells and measure the enzyme's activity as is.

Then you measure it again after adding extra coenzyme to the text tube.

And if the activity jumps up?

If the activity increases, it means the enzyme protein, the opal enzyme was there, but it wasn't fully active because the patient didn't have enough coenzyme.

It's a direct measure of a functional deficiency.

And the result is that enzyme activation coefficient.

Right.

It's a ratio.

We use it for things like transketolase to check thiamine or B1 status.

It's so much better than just measuring the vitamin level in the blood, which can change hour to hour.

So for more complex things like hormones, we use amino assays.

Yes.

These use a binding protein, usually a very specific antibody, to grab on to the target molecule, which we call the ligand.

The classic method was radioimmunoassay, but now we mostly use fluorescent labels instead of radioactivity.

And the most common type is the ELISA or the sandwich assay.

Can you walk us through that visually?

Sure.

Imagine the molecule you want to measure is the filling in a sandwich.

The first antibody is like the bottom slice of bread.

It's stuck to the testing plate and it captures your molecule.

Then you wash everything else away.

So you've isolated it.

You've isolated it.

Then you add a second antibody, the top slice of bread, which is labeled with an enzyme or a fluorescent tag.

It binds to a different spot on your molecule, completing the sandwich.

That tag is what you measure, and it's incredibly specific.

That's a great explanation.

And on the complete opposite end of the complexity spectrum, you have dry chemistry.

That's the stuff in your pocket.

It's the technology behind a glucometer strip, which has all the reagents like glucose oxidase just dried onto it or a urine dipstick, which can test for 10 different things at once.

It's all about making testing simple, fast and accessible.

Let's finish up by seeing how this all comes together in some common organ function tests.

We can start with the kidneys.

For kidneys, the routine markers are serum urea and creatinine.

They both go up when kidney function goes down.

But creatinine is definitely the better marker.

Its level isn't affected as much by things like your diet or how hydrated you are and for detecting early damage, especially in diabetes.

You look for protein in the urine, specifically microalbuminuria.

That's the excretion of just a small amount of albumin.

And it's a critical early warning sign that the kidneys filters are starting to get damaged.

And we can estimate the overall filtration rate, the GFR, using clearance, right?

We can.

We calculate the creatinine clearance.

It's a good estimate, but it actually slightly overestimates the true GFR.

That's because a little bit of creatinine is actively secreted into the urine by the kidney tubules, not just filtered.

The true gold standard is inulin clearance, but that's not practical for routine clinical use.

OK, moving to the liver function tests.

What are the key things to look at there?

You look at bilirubin.

If you see mainly conjugated bilirubin, that points towards a physical blockage in the bile ducts.

If you see both conjugated and unconjugated, it suggests more widespread damage to the liver cells themselves, like in hepatitis.

And what are the liver enzymes?

ALT and AST shoot up in acute hepatitis.

We usually consider ALT to be a bit more specific to the liver, since AST can also come from damaged heart or skeletal muscle.

And then there's alkaline phosphatase, or ALP, which really spikes in obstructive jaundice.

Finally, let's just quickly touch on some markers for the endocrine and cardiovascular systems.

Sure.

For the thyroid, it's best to measure the free T4 and T3 hormones, along with TSH.

Measuring the total T4 can be misleading because it depends on binding proteins.

For the adrenal gland in something like crushing syndrome, you look for a loss of cortisol's normal daily rhythm and you use the dexamethasone suppression test.

And the really critical ones for the heart.

For a heart attack, a myocardial infarction, you need a confirmation from cardiac specific proteins.

That means cardiac troponin or the creatine kinase MB isoenzyme, CKMB.

They only leak out when heart muscle itself has died.

So we've gone from a single drop of blood all the way to diagnosing organ failure, unpacking the statistics, the kinetics and the analytical logic behind it all.

I think you, the listener, should now have a really strong grasp on the two different meanings of sensitivity and specificity.

That statistical tightrope of the reference range and the three ingenious ways we use enzymes in the lab.

And what's really amazing to me is how the reliability of modern medicine really rests on our ability to synthesize these incredibly complex biochemical and statistical ideas.

Interpreting a clinical picture, is it a heart attack, a vitamin deficiency or just statistical noise?

It's all made possible because we can quantify chemistry down to a single molecule.

That five percent error rate, though.

It's like the ghost in the machine of diagnosis.

It absolutely is.

And the entire process is about navigating that statistical certainty that five percent of healthy people will have an abnormal result.

And that brings up a final thought.

What other fields outside of medicine define normal using these kinds of arbitrary statistical cutoffs?

And what does it mean when a really important outcome like a job performance review or a grade falls just outside that threshold?

It's something to think about the next time you look at a lab report.