Chapter 28: Natriuretic Peptides — The Heart Failure Biomarkers That Became Treatments

"The heart of animals is the foundation of their life, the sovereign of everything within them, the sun of their microcosm, that upon which all growth depends, from which all power proceeds." — William Harvey, De Motu Cordis (1628)

Overview

Harvey was arguing that the heart is a pump. He was right, and for three and a half centuries that was understood to be the complete answer. The heart moves blood. Other organs make hormones. The division of labor was tidy, it was in every textbook, and it was wrong.

In 1981 a physiologist named Adolfo de Bold ground up rat atrial tissue and injected the extract into other rats. Within minutes the recipients were producing startling volumes of sodium-rich urine. Extract from ventricular tissue did nothing of the kind. Something in the atria — something stored in granules that electron microscopists had been photographing without explanation for decades — was a hormone. The heart is not only a pump. The heart is a sensor that reports on its own working conditions, in a chemical language the kidney and the blood vessels can read.

That discovery produced three separate lines of medicine. The same peptide system was turned into a diagnostic test, into an infused drug, and into a target for a drug that is not a peptide at all. The test reorganized how breathlessness is evaluated in emergency departments worldwide. The infused drug was approved, doubted, tested properly, and abandoned. The target produced one of the most consequential heart failure therapies of the last twenty years.

Three interventions. One system. Two successes and one failure, and you cannot predict which was which from the biology alone.

That is why Chapter 28 sits where it does. Rule 6 of our rating system says one molecule, many ratings, and every chapter so far has illustrated it with a molecule that works for one thing and not another. This chapter goes further: the same molecule earns different ratings in different roles — measured versus administered versus preserved — and each role required a different kind of study to settle. Any source that hands you a single verdict on "natriuretic peptides" has compressed away everything you actually need.

In this chapter, you will learn to:

• Explain what the heart secretes, what triggers it, and why that was a genuine surprise
• Tell ANP, BNP, and CNP apart by source, receptor, and job
• Read a BNP or NT-proBNP result the way a cardiologist does — including what makes it unreliable
• Separate a biomarker claim from a therapy claim and match each to the study design that settles it
• Reconstruct the nesiritide story as a textbook Chapter 5 surrogate failure
• Explain why raising a peptide by blocking its breakdown beat supplying it from outside
• Say why the neprilysin inhibitor had to be paired with an ARB, and why that is chemistry rather than marketing
• Generalize the lesson to any peptide system you encounter for the first time

Learning Paths

This chapter is the book's cleanest demonstration of the ratings system. Everyone should read §28.4, whatever else you skip.

💊 GLP-1 — §28.6 is your section. Neprilysin inhibition is the same "preserve what the body makes" strategy as the DPP-4 inhibitors in Chapter 7 §7.6, and the comparison explains a lot about why your drug class looks the way it does. 🏋️ Performance — §28.4 and §28.5. Nesiritide is what it looks like when a compound improves every measurement you can take and does not improve the outcome you care about. You will meet that pattern constantly in Part III. 🔬 Science — read straight through. §28.2 and §28.6 are the receptor and enzyme detail; §28.8's CNP story is the book's best illustration of a peptide system finding its use somewhere nobody was looking. 💄 Cosmetic — §28.4 only, read as method rather than cardiology. The biomarker-versus-therapy distinction is exactly the distinction between an ingredient that can be detected doing something in skin and one that changes how skin looks (Chapter 30). 🏥 Clinical — the whole chapter is your territory, but §28.3's confounder list and §28.7's safety callout are where the general reader most often gets it wrong.

28.1 The heart as an endocrine organ

Start with why this was surprising, because the surprise is the point.

By the middle of the twentieth century the cardiovascular system had a settled description. The heart pumps. The kidneys manage salt and water. The adrenal glands and the kidney between them run a hormonal system — renin, angiotensin, aldosterone — that raises blood pressure and retains sodium when volume falls. Everything had a job and the jobs did not overlap.

There were two loose threads. Electron microscopists looking at atrial muscle cells kept seeing dense, membrane-bound granules clustered near the nucleus. In any other tissue, granules that look like that mean one thing: something is being packaged for secretion. In heart muscle they meant nothing anyone could explain, so they sat in the literature for roughly two decades. And distending the atria of an experimental animal produced a diuresis whose size the available neural explanation never quite accounted for.

De Bold's 1981 experiment closed both threads at once. Homogenized atrial tissue, injected into recipient rats, produced a rapid and large increase in sodium and water excretion together with a fall in blood pressure. Ventricular tissue did not. The granules were secretory granules. The diuresis of atrial distension was hormonal, not just neural. The heart had been making a hormone the whole time, and the reason nobody found it earlier is that nobody was looking, because the heart was categorized as a pump.

The generalizable lesson is about categories, not cardiology. Organs do not know what chapter of the textbook they are in. Fat tissue turned out to be endocrine. The gut turned out to be the largest endocrine organ in the body, which is the entire foundation of Part II of this book. Bone secretes signals that act on the pancreas. The category "endocrine organ" was administrative, and when it was treated as biological it delayed a discovery by twenty years.

What "endocrine organ" actually means here

An endocrine organ makes a substance, releases it into the bloodstream, and that substance acts on distant tissue through a receptor. The heart qualifies fully: cardiac muscle cells transcribe the genes, make the precursor peptides, release them, and the peptides travel to the kidney and the vasculature and bind specific receptors there.

What makes the cardiac case distinctive is the trigger. Most endocrine glands respond to a chemical input — a rising glucose concentration, a falling calcium level, a stimulating hormone from the pituitary, which is the whole architecture of the axis you met in Chapter 27. Cardiac muscle responds to a mechanical input. It responds to being stretched.

A muscle cell measures the tension in its own wall and converts that measurement into a hormone concentration in the blood. It is a strain gauge wired to a chemical transmitter. Whatever else you take from this chapter, take that: the heart reports its own loading conditions to the rest of the body, continuously, and we can now read the report with a blood test.

28.2 ANP, BNP, CNP, and what stretch is saying

Three related peptides make up the family. They share a structural signature — a ring closed by a disulfide bond between two cysteines, the staple you met in Chapter 1 §1.2 — and they differ in where they are made, what triggers them, which receptor they prefer, and what they are for.

THE NATRIURETIC PEPTIDE FAMILY

  PEPTIDE   MADE MAINLY BY      RELEASED WHEN          RECEPTOR   ROLE
  ───────   ────────────────    ──────────────────     ────────   ──────────────────────────
  ANP       atrial muscle       atria are stretched    NPR-A      fast, circulating hormone;
            (stored, ready)     by extra volume                   salt and water off-loading

  BNP       ventricular muscle  ventricles are under   NPR-A      slower, sustained; the
            (made on demand)    pressure or volume                signal of a struggling
                                load                              ventricle

  CNP       endothelium,        local signals, not     NPR-B      paracrine; vascular tone,
            cartilage           chamber stretch                   and bone growth at the
                                                                  growth plate

  All three signal through a receptor that IS an enzyme: binding switches on
  guanylyl cyclase activity inside the cell, which makes cGMP, which activates
  protein kinase G. One second messenger, three peptides, different tissues.

  A fourth receptor, NPR-C, binds all three and does not signal. It internalizes
  and destroys them. It is a garbage disposal, and it matters in §28.6.

Two things in that table deserve emphasis before we go on.

The receptor is the enzyme. In Chapter 2 you met the standard peptide receptor architecture: a G-protein-coupled receptor that binds a peptide on the outside and recruits machinery on the inside. The natriuretic peptide receptors are built differently. NPR-A and NPR-B are single-pass membrane proteins whose intracellular portion is a guanylyl cyclase. Binding the peptide activates the enzyme directly. The second messenger is cyclic GMP, not cyclic AMP. Remember cGMP — it is the common currency that lets a drug in §28.8 join this chapter without being a peptide.

Only two of the three are natriuretic in any meaningful sense. CNP carries the family name and does very little natriuresis. It is a local signal in blood vessel walls and, as it turns out, in cartilage. The naming is historical rather than functional, and it will matter in §28.8 when CNP's therapeutic use arrives from an entirely unexpected direction.

What the signal does

When ANP or BNP reaches its receptor, four things happen, and each one is precisely the opposite of what the renin-angiotensin-aldosterone system does:

Natriuresis and diuresis. The kidney excretes more sodium and more water — filtration pressure rises, and sodium reabsorption in the collecting duct is inhibited. Volume comes down.

Vasodilation. Arterial and venous smooth muscle relaxes. Blood pressure falls, and — importantly for a failing heart — the pressure the ventricle has to pump against falls with it.

Suppression of RAAS. Renin release is inhibited and aldosterone secretion is inhibited. The system that retains sodium is turned down at two points.

Restraint of remodeling. Over longer timescales, natriuretic peptide signaling opposes cardiac hypertrophy and fibrosis — the structural changes by which a chronically overloaded heart makes itself worse.

The logic is clean. When the heart is overfilled, it says so, and the response is to shed volume, lower pressure, and stop the compensatory system that is making things worse. It is a negative feedback loop with the heart as the sensor.

🧬 The Molecule — three peptides, one ring, and one badly chosen name

ANP is 28 amino acids, cut from a 126-residue precursor stored preformed in those atrial granules. Because it is stored ready to go, release is fast — a change in atrial filling produces a change in circulating ANP within minutes. Its half-life in blood is very short, on the order of a few minutes, which makes it a good acute signal and a nearly impossible drug (Chapter 4's problem, exactly).

BNP is 32 amino acids. It is not stored in any quantity; ventricular muscle responds to load by transcribing more of the gene. That makes BNP slower to rise and slower to fall than ANP, which is a disadvantage for physiological signaling and an enormous advantage for measurement. A number that integrates over hours is a more stable readout than one that swings minute to minute.

The "B" originally stood for brain, because the peptide was first isolated from porcine brain tissue in the 1980s. It was subsequently found to be overwhelmingly a cardiac ventricular product, and the name was quietly retconned to B-type. You will still see "brain natriuretic peptide" in circulation, including in patient-facing material, and it causes real confusion — patients ask why their heart test is measuring a brain hormone. It is not. The name is an accident of which tissue happened to be ground up first.

CNP is 22 amino acids, made mainly by endothelial cells and by chondrocytes in the growth plate, and it signals through a different receptor, NPR-B. It circulates at low concentrations and acts mostly where it is made.

All three share a 17-residue ring closed by a disulfide bond, and that ring is the part the receptor recognizes. Open the ring and the activity is gone. This is Chapter 1 §1.4 in miniature: sequence is what you can control, shape is what the receptor sees.

The paradox that should bother you

Here is a fact that complicates everything that follows, and the honest thing to do is put it up front rather than in a footnote.

In heart failure, natriuretic peptide levels are high — and the effect is blunted.

Patients with advanced heart failure are retaining sodium and water while carrying circulating natriuretic peptide concentrations many times normal. The hormone that exists to make you shed volume is present in abundance, and volume is not being shed. The phenomenon has a name, natriuretic peptide resistance, and several contributing explanations: receptors downregulate under sustained exposure; clearance through NPR-C and through enzymatic degradation is increased; and the processing of the precursor becomes less efficient, so more of the circulating material is unprocessed or partially processed precursor with reduced biological activity.

That last point has an uncomfortable implication for the biomarker. What the assay detects and what is biologically active are not perfectly the same population of molecules. The test is still extremely useful, as §28.3 will show. But "high BNP means high BNP activity" is not safe to assume.

Notice also that receptor downregulation under sustained exposure is precisely the Chapter 3 mechanism — the difference between a signal that arrives in pulses on the body's schedule and one that arrives flat and continuously. Heart failure creates, endogenously, the exact condition Chapter 3 warns about in drugs. Hold that thought until §28.5, because it is the best available explanation for why supplying more of this hormone from outside did not do what everyone expected.

28.3 The blood test that reorganized a diagnosis

Now the most useful thing this system has given medicine, and it is not a drug.

How the measurement works

The BNP gene produces a precursor, proBNP, 108 amino acids long. Enzymes cleave it into two fragments, and both end up in the blood:

PROBNP PROCESSING — one precursor, two measurable fragments

              proBNP (108 amino acids)
     ┌────────────────────┴────────────────────┐
     │                                         │
  NT-proBNP (76 aa)                        BNP (32 aa)
  N-terminal fragment                      C-terminal fragment

  BIOLOGICALLY INACTIVE                    BIOLOGICALLY ACTIVE
  does not bind NPR-A                      binds NPR-A, makes cGMP
  half-life over an hour                   half-life around 20 minutes
  cleared largely by the kidney            cleared by NPR-C AND by the
                                           enzyme neprilysin (remember this)
  more stable in the sample tube           degrades faster in the tube

  BOTH are measured clinically. Both rise with cardiac wall stress. They are
  NOT interchangeable numbers — the reference ranges differ by roughly an order
  of magnitude, and a result means nothing until you know which test produced it.

The inactive fragment is, for measurement purposes, the better molecule. NT-proBNP is more stable in the blood and more stable in the specimen, and it is produced in the same molar quantity as the active peptide, so it reports the same underlying signal with less noise. This is a small, delightful irony: the useless half of the molecule is the more useful measurement.

Both are used. Which one your laboratory runs is a matter of which analyzer it bought, and the two numbers are not comparable to each other. If you take only one practical fact from this section, take that one.

The clinical problem it solved

Consider what "shortness of breath" means to someone receiving a patient at three in the morning. It is one of the most common presenting complaints in emergency medicine and one of the least specific. The causes that matter include heart failure, pneumonia, an exacerbation of chronic obstructive pulmonary disease, asthma, pulmonary embolism, anemia, kidney failure, and anxiety. Several can be present at once, which is the part that makes it genuinely hard — a lifelong smoker with a bad heart can be breathless from either organ, or both, and the initial treatments differ substantially.

The tools available before natriuretic peptide testing were history, physical examination, chest radiography, and — if it was available at that hour, which frequently it was not — echocardiography. Each is useful and each is fallible. Physical signs of heart failure are notoriously insensitive, and chest films are ambiguous in exactly the patients in whom the question is hardest.

Because BNP and NT-proBNP rise in proportion to cardiac wall stress, measuring them adds an independent piece of information about whether the heart is under load, which is close to the question the clinician is actually asking. The pivotal work was done in emergency departments in the early 2000s — most prominently a large multinational study of patients presenting with acute dyspnea, in which BNP measured at presentation was compared against a blinded expert adjudication of the final diagnosis, and a companion study a few years later that did the same for NT-proBNP.

The results were consistent and they were substantial. The commonly cited figures for BNP at the conventional threshold of 100 pg/mL are a sensitivity approaching 90 percent with a specificity in the mid-70s — which tells you exactly what the test is good for. A high sensitivity with a lower specificity is a rule-out test. A low value makes acute heart failure unlikely and is genuinely informative. A high value is compatible with heart failure and with a list of other things, and requires the clinician to keep thinking.

💊 In the Clinic — what the number actually changes

It is easy to overstate what a test does. A natriuretic peptide result does not diagnose heart failure. It shifts a probability, and it shifts it most usefully at the low end.

The realistic picture: a patient arrives breathless, the history is muddy, the examination is equivocal, and the chest film could be read either way. An NT-proBNP that comes back low is a strong argument that the breathlessness is not from acute heart failure, and it redirects the work toward the lungs. That redirection happens in minutes rather than after an overnight wait for a morning echocardiogram, and it changes which treatment is started first.

A markedly elevated result in the same patient does not close the question, but it raises the probability enough to justify treating for heart failure while the rest of the workup proceeds.

Guidelines across major cardiology bodies now incorporate natriuretic peptide measurement into the diagnostic pathway for both acute and chronic heart failure, and it is used again for prognosis: higher values, in a patient with established heart failure, are associated with worse outcomes, consistently, across many cohorts.

What it is not. It is not a screening test for the general population. It is not a substitute for an echocardiogram, which answers a different question — what is the structure and function of this heart — that no blood test answers. And per §28.4, it is not a treatment target. Any decision about your own care belongs with a clinician who can put the number next to the rest of you (Chapter 39).

What limits it — and which direction each confounder pushes

An honest account of a good test includes the ways it misleads. Four confounders matter most, and what makes them tractable is that each pushes in a known direction.

Age raises it. Levels rise with age in people without heart failure. A value clearly abnormal in a forty-year-old may be unremarkable at eighty, which is why age-stratified thresholds exist for NT-proBNP rather than a single cutoff.

Kidney impairment raises it. Reduced clearance raises measured levels, and the effect is larger for NT-proBNP, which is cleared more dependently on renal function. In advanced kidney disease the test loses much of its discriminating power in exactly the population that most often presents breathless.

Obesity lowers it. Counterintuitive and clinically important. People with higher body mass index have lower natriuretic peptide levels for the same degree of cardiac stress; proposed mechanisms include increased clearance receptor expression in adipose tissue and reduced production. So a reassuringly low value in a person with severe obesity is less reassuring than the same value in a lean person — the one situation where the rule-out strength of the test is weakest.

Atrial fibrillation raises it, sometimes dramatically. Fibrillating atria produce natriuretic peptides abundantly, so a patient in atrial fibrillation can carry a level in the heart failure range without heart failure, and specificity in that group falls substantially.

Others exist — pulmonary embolism and pulmonary hypertension through right ventricular strain, and acute coronary syndrome, sepsis, and anemia. And one confounder is created by a drug in this very chapter, which §28.7 gets to and which is the single most elegant fact in it.

The correct posture is that this is an aid to judgment and not a verdict. That formulation is worth memorizing, because it applies to every biomarker you will meet in this book, and because the failure mode is always the same: a number that was validated as one input to a decision gets treated as the decision.

📊 Evidence Rating

Claim: Measuring BNP or NT-proBNP aids the diagnosis of heart failure — particularly its exclusion — in adults presenting with undifferentiated breathlessness. Rating: ✅ Strong clinical evidence (as of 2026). Why: Multiple large prospective diagnostic-accuracy studies in emergency populations, with blinded outcome adjudication, showed that the test substantially improves discrimination between cardiac and non-cardiac dyspnea; it is embedded in the diagnostic algorithms of every major heart failure guideline and is among the most-ordered tests in cardiology. What would change it: the accuracy itself is not seriously in doubt. What would move the practical value is evidence that the confounders — age, renal function, obesity, atrial fibrillation — cause materially more misclassification in routine use than in study populations, or the arrival of a test with comparable sensitivity and better specificity.

28.4 Why a biomarker rating and a therapy rating are different ratings

This is the section the chapter exists for.

You now know that the same peptide system can be measured and can be administered. Those are not two aspects of one question. They are two questions, they are answered by different study designs, and in this system they have different answers.

Here are the three claims, stated precisely, with the design that settles each:

ONE SYSTEM, THREE CLAIMS, THREE STUDY DESIGNS, THREE ANSWERS

 (a) DOES MEASURING IT HELP DIAGNOSE HEART FAILURE?
     Study design that settles it .... prospective diagnostic accuracy study:
                                       enroll consecutive patients with the
                                       symptom, measure the marker, compare
                                       against a blinded reference diagnosis
     What it can show ............... sensitivity, specificity, predictive values
     Answer ......................... YES.  ✅

 (b) DOES INFUSING A SYNTHETIC VERSION IMPROVE OUTCOMES IN ACUTE HEART FAILURE?
     Study design that settles it .... randomized controlled trial with a hard
                                       clinical endpoint: death, rehospitalization
     What it can show ............... whether patients do better, not whether
                                       measurements improve
     Answer ......................... NO.  ❌

 (c) DOES PREVENTING ITS BREAKDOWN IMPROVE OUTCOMES IN CHRONIC HEART FAILURE?
     Study design that settles it .... a DIFFERENT randomized controlled trial,
                                       different population, different duration,
                                       different comparator
     What it can show ............... same category of answer, different question
     Answer ......................... YES.  ✅

 ANY SINGLE VERDICT ON "NATRIURETIC PEPTIDES" IS WRONG ABOUT AT LEAST TWO OF THESE.

Read that last line again, because it is the whole thesis of the book compressed into one system. "Natriuretic peptides: do they work?" is not a question with an answer. It is three questions wearing a trench coat, and the answers are yes, no, and yes.

Why the designs cannot substitute for each other

This is the part people skip, and it is where most bad reasoning about peptides lives.

A diagnostic accuracy study tells you how well a measurement tracks a condition. It enrolls people with a symptom, measures the marker, and compares against the best available reference standard. It answers: if I know this number, how much better do I predict the diagnosis? It says nothing whatsoever about whether acting on the number helps anyone. It cannot. There is no intervention in it. A randomized outcome trial answers the other question — if I do this, do patients do better? — and is a poor and expensive way to learn about a measurement.

Notice that the currency is different in each. Accuracy studies deal in sensitivity and specificity. Outcome trials deal in event rates and hazard ratios. A claim expressed in one currency cannot be paid in the other, and when you see someone attempt it — "the biomarker correlates so strongly with outcomes that improving it must improve outcomes" — you are watching a category error.

The fourth claim, which fails

To make the point unmissable, consider a claim that sits between (a) and (c): does using the measurement to guide treatment improve outcomes? Not diagnose. Guide. Adjust the medications until the number comes down.

This sounds almost self-evidently reasonable. The number tracks severity. Lower is better. Titrate to the number.

It was tested — randomized, in high-risk patients with reduced ejection fraction, biomarker-guided titration against usual guideline-directed care — and the trial was stopped for futility. There was no benefit to titrating toward the target. The likely explanation is unglamorous and instructive: the treatments the biomarker would push you toward are the treatments that good guideline-directed care already prescribes, so the number added no decision that was not already being made.

Same molecule. Same assay. Same disease. A fourth claim, and a different answer again.

📊 Evidence Rating

Claim: Titrating heart failure therapy toward an NT-proBNP target improves clinical outcomes in chronic HFrEF, compared with usual guideline-directed care. Rating: ❌ Hype outpaces evidence (as of 2026). Why: A randomized trial of biomarker-guided titration versus usual care in high-risk patients was stopped for futility; the strategy did not improve outcomes over care that was already following guidelines. What would change it: a trial in a setting where usual care is substantially less optimized, or a strategy in which the biomarker triggers a decision that guidelines do not already dictate. Note carefully that this ❌ does not touch the ✅ in §28.3 — different claim, different design.

⚠️ Hype Check — "the body makes it to fix this exact problem, so more of it must help"

The claim, in its usual form:

"Your heart releases natriuretic peptides precisely when it is overloaded. That is the body's own solution to heart failure. Giving more of it is just helping the body do what it is already trying to do."

What's true in it. All of it, as physiology. The peptides really are released in response to overload. They really do promote exactly the changes a failing heart needs — less volume, lower pressure, less RAAS activation. The reasoning is not stupid; it is the reasoning that motivated a genuine drug development program at a serious pharmaceutical company, cleared a regulatory review, and put a product on the market.

Where it fails. It failed empirically, and that is the only way this kind of argument can fail. Section 28.5 is the record. A recombinant human BNP was infused into patients with acute heart failure in a trial of roughly seven thousand people, and they did not do better.

Why the mechanism was not enough. At least three reasons are on the table, and none of them was visible from the physiology. First, §28.2's paradox: in heart failure the receptors are already downregulated and the system is already resistant, so adding ligand to a resistant system is not the same as restoring a deficient one. Second, Chapter 3's lesson: a continuous infusion is not the signal the body produces, which is graded, local, and terminated on cue. Third, and most mundane, the patients were already receiving treatments that relieve the same physiology, so there was less room to improve than the mechanism suggested.

Rule 3 of the rating system exists for this. Never upgrade a rating with mechanism. A mechanism tells you a claim is worth testing. It never tells you the test will pass, and the better and more elegant the mechanism, the more careful you should be — because a beautiful mechanism makes a negative trial feel like an error rather than an answer.

🔍 Check Your Understanding

1. A colleague says "NT-proBNP is well validated, so lowering it should be the goal of therapy." Name the two different claims being conflated, and state which study design settles each.
2. Why can a diagnostic accuracy study never demonstrate that a treatment works, no matter how large it is?
3. Nesiritide's ❌ and sacubitril/valsartan's ✅ apply to the same physiological system. Explain to someone who finds that contradictory why it is not.

28.5 Nesiritide: approval on a surrogate, and the trial that answered it

Nesiritide is recombinant human BNP — the natural 32-amino-acid peptide, made in bacteria, identical in sequence to what a ventricle secretes. As a piece of molecular biology it is unimpeachable. It is exactly the molecule, and there is no analog trickery of the kind you met in Chapter 33 to argue about.

It was approved in the United States in 2001 for acute decompensated heart failure with dyspnea at rest — the situation in which a patient is admitted acutely breathless from fluid overload.

What the approval rested on

The pivotal evidence was hemodynamic and symptomatic. Infused nesiritide lowered pulmonary capillary wedge pressure — a direct measurement of filling pressure on the left side of the heart, obtained with a catheter, and about as objective a physiological measurement as cardiology has. It also improved patient-reported dyspnea at a few hours compared with placebo.

Both of those are real findings. The pressure came down. The patients said they could breathe more easily. Nobody faked anything and nobody was careless.

But look at what the evidence consisted of, in the vocabulary of Chapter 5 §5.6: a surrogate endpoint and a short-term symptom score. Filling pressure is a surrogate. It is a measurement that correlates with the disease and is expected to track benefit. Dyspnea at three hours is a real patient experience, but it is not survival and it is not rehospitalization.

The question that had not been answered was the one that matters: do people who get this drug live longer or come back to hospital less often?

The doubt

Within a few years of approval, published re-analyses pooling the randomized data raised two concerns: a possible increase in worsening renal function, and a possible increase in short-term mortality. Those analyses were pooled and retrospective, which is a genuine limitation and was argued about at length. They did not establish harm. What they did was make it untenable to leave the question open — and the mechanism by which that happened was not regulatory action but other researchers reading the submitted data and publishing an analysis the sponsor had not commissioned.

The manufacturer then commissioned a definitive trial. That is the correct response, it was expensive, and it is worth stating plainly that they did it.

🔬 Read the Study — the trial that settled nesiritide

Design. A randomized, double-blind, placebo-controlled trial in patients hospitalized with acute decompensated heart failure. Roughly 7,100 patients across many countries. Nesiritide or placebo was added on top of standard care — which is the ethically necessary design and also the honest one, since the practical question was never "nesiritide or nothing" but "does adding this help?"

Endpoints. Co-primary: change in self-reported dyspnea at 6 and 24 hours, and the composite of rehospitalization for heart failure or death at 30 days. Note the design decision — they kept the symptom endpoint that supported the approval and added the hard endpoint that had never been tested. That lets the trial answer both the old question and the new one.

Results. Dyspnea improved slightly more with nesiritide, and the difference did not meet the prespecified threshold for significance. The 30-day composite of death or rehospitalization was essentially the same in both groups. There was more hypotension with nesiritide. The trial did not confirm the increases in mortality or renal dysfunction that the earlier pooled analyses had raised — which is worth saying clearly, because the story is often told as though harm was proven. It was not. What was demonstrated was absence of benefit.

What it is good for. This is close to a model trial for the question asked: large enough to detect a clinically meaningful difference, randomized, blinded, placebo-controlled, hard endpoint, conducted by the party with the most to lose from the answer.

What it cannot tell you. It studied a specific population — patients acutely hospitalized, already receiving effective standard treatment — over a short window. It does not exclude benefit in some different population, at a different point in the illness, in a different setting. That caveat is real. It is also the caveat every negative trial has, and it should not be used to keep a claim alive indefinitely.

Use of nesiritide fell sharply after the trial reported. It is no longer a meaningful part of practice.

And this pattern is not a one-off. A different natriuretic peptide, an analog of a related renal peptide, was later tested in acute heart failure in its own large randomized trial. It, too, improved hemodynamic measures. It, too, failed to improve cardiovascular mortality. Two independent attempts to supply natriuretic peptide from outside in acute heart failure, two improvements in physiology, two failures on outcomes. A single negative trial is a result. Two independent negative trials on the same strategy is a finding about the strategy.

📊 Evidence Rating

Claim: Infused nesiritide (recombinant human BNP) improves clinical outcomes — death or rehospitalization — in adults hospitalized with acute decompensated heart failure. Rating: ❌ Hype outpaces evidence (as of 2026). Why: Approved on hemodynamic surrogate and short-term symptom measures; a subsequent randomized, placebo-controlled outcome trial of roughly 7,100 patients found no meaningful effect on death or rehospitalization at 30 days and only a small, non-significant dyspnea difference, with more hypotension. Evidence is present, adequate, and negative — a different situation from the ❌ compounds in Part III, where evidence is largely absent. What would change it: this claim has been tested at scale in the population it was approved for, and answered. Reopening it would require a trial in a materially different population or illness phase, with a prespecified hard endpoint and a mechanistic reason to expect a different result — not a re-analysis of subgroups.

Read this as a Chapter 5 §5.6 story, because that is what it is. A drug was approved on measurements that correlate with the disease. The hard-endpoint trial came later. The hard-endpoint trial did not confirm. Everything in that sequence is legal, was done in good faith, and is exactly why §5.6 tells you to ask what endpoint was this approved on before you ask anything else.

28.6 Neprilysin, and the strategy of not throwing things away

If supplying the peptide from outside does not work, there is another way to raise it: stop destroying the peptide the body already makes.

The enzyme

Neprilysin is a membrane-bound zinc-dependent enzyme, abundant on the brush border of the kidney tubule and present in lung, vasculature, and elsewhere. It is a peptidase — one of the enzymes Chapter 1 §1.3 told you are everywhere and are the central problem of peptide pharmacology.

Its substrate list is long and, crucially, it is not selective:

NEPRILYSIN — what it degrades, and why the list matters

  SUBSTRATE            EFFECT OF INHIBITING ITS BREAKDOWN
  ─────────────────    ─────────────────────────────────────────────────────
  ANP                  ↑ more natriuresis, vasodilation      ← the point
  CNP                  ↑ vascular effects
  BNP                  ↑ (BNP is a relatively POOR substrate — see below)
  adrenomedullin       ↑ vasodilation                        ← probably helpful
  bradykinin           ↑ vasodilation, and ANGIOEDEMA RISK   ← the problem
  substance P          ↑                                     ← also angioedema
  angiotensin II       ↑ vasoconstriction, sodium retention  ← THE problem
  amyloid-beta         ↑ (a theoretical long-term concern)

  An enzyme with eight substrates has eight consequences when you block it.
  You do not get to choose which ones.

Three consequences follow, and each shaped the drug that eventually worked.

First, and this is a nuance worth having: BNP is a relatively poor neprilysin substrate compared with ANP and CNP. So a neprilysin inhibitor probably delivers most of its natriuretic peptide effect through ANP rather than through BNP, and probably delivers additional effect through adrenomedullin and other substrates. The drug is not a "BNP booster," whatever the shorthand suggests. It raises several peptides at once, and attributing the clinical benefit to any one of them is not something the trial evidence can do.

Second, bradykinin. Neprilysin degrades bradykinin, so inhibiting it raises bradykinin. Elevated bradykinin is the mechanism behind the cough and the angioedema associated with ACE inhibitors, because ACE degrades bradykinin too. Block both enzymes and bradykinin accumulation is compounded. This is not theoretical: an earlier drug that combined neprilysin inhibition with ACE inhibition in a single molecule was tested extensively and produced angioedema substantially more often than an ACE inhibitor comparator. It was not approved. That failure is why the successful drug is built the way it is, and §28.7's safety callout is its direct descendant.

Third, angiotensin II. This is the one that makes the combination mechanically necessary rather than commercially convenient. Neprilysin degrades angiotensin II — the peptide at the business end of the RAAS, the one that constricts vessels and drives aldosterone and sodium retention. Inhibit neprilysin on its own and you raise the good peptides and the bad one. You would be lifting the brake and the accelerator simultaneously, in a disease where the accelerator is already the problem.

The solution is to block the angiotensin II receptor at the same time. Then neprilysin inhibition raises natriuretic peptides and adrenomedullin, and the angiotensin II it also raises has nowhere to act.

Say this clearly, because it is routinely misunderstood: the combination is not two drugs bundled for convenience. Blocking the angiotensin receptor is what makes neprilysin inhibition usable at all. And it must be an angiotensin receptor blocker rather than an ACE inhibitor, because of the bradykinin problem above.

The pattern you have already seen

Step back from cardiology, because this strategy should look familiar. In Chapter 7 §7.6 you met the DPP-4 inhibitors. DPP-4 is the enzyme that degrades GLP-1; inhibit it and endogenous GLP-1 survives longer, amplifying the incretin effect without injecting any GLP-1 at all. Same logic, same architecture. Preserve rather than supply.

In that case the strategy produced a real but modest effect, considerably smaller than a receptor agonist's. The reason is structural: you can only preserve what the body actually makes. A degradation inhibitor is capped by endogenous production. A receptor agonist is not — you can push a receptor far past any concentration physiology produces, which is precisely why semaglutide does things GLP-1 never did.

So the expectation going into the heart failure story should have been: preserving will work, and it will work less well than supplying.

The opposite happened. Supplying failed. Preserving succeeded. That inversion is the most instructive thing in this chapter, and §28.7 has the trial and §28.9 has the explanation.

28.7 Sacubitril/valsartan and PARADIGM-HF

Sacubitril/valsartan pairs a neprilysin inhibitor prodrug with an angiotensin receptor blocker in a single co-crystallized product. The class acquired its own name: ARNI, for angiotensin receptor–neprilysin inhibitor.

Note, and this is a genuinely important point for a book about peptides: neither component is a peptide. Both are small molecules. The peptides in this story are the endogenous ones — the ones the drug protects. The most successful therapy in this chapter targets a peptide system without containing a peptide, and we will come back to that in §28.9.

🔬 Read the Study — PARADIGM-HF

Design. Randomized, double-blind, active-controlled. 8,442 patients with chronic heart failure with reduced ejection fraction, symptomatic, with elevated natriuretic peptide levels, receiving background therapy including a beta-blocker.

Comparator. Enalapril — not placebo. This is the detail that makes the result impressive. Enalapril is an ACE inhibitor with an established mortality benefit in this population, so the trial asked whether the new drug beat a treatment already known to save lives. Beating placebo is a lower bar than beating the standard of care, and this trial cleared the higher one.

Run-in. Before randomization, patients went through sequential single-blind periods on enalapril and then on the study drug, and those who could not tolerate either were not randomized. This is a real design criticism and should be stated: a run-in enriches the trial for people who tolerate the drugs, so the tolerability observed in the trial is likely better than tolerability in unselected practice. It is less clear that it inflates the efficacy comparison, since both arms were run in.

Result. The trial was stopped early, on the recommendation of its data monitoring committee, after a median follow-up of about 27 months, because the boundary for overwhelming benefit on cardiovascular mortality had been crossed. The primary composite of cardiovascular death or heart failure hospitalization occurred in 21.8 percent on sacubitril/valsartan versus 26.5 percent on enalapril — a hazard ratio of 0.80 (95% CI 0.73–0.87). Cardiovascular death alone and heart failure hospitalization alone each moved in the same direction. All-cause mortality was 17.0 versus 19.8 percent, hazard ratio 0.84.

Adverse effects. More symptomatic hypotension with sacubitril/valsartan. Less cough, less hyperkalemia, and less renal impairment than with enalapril. Angioedema was uncommon in both arms and numerically more frequent with sacubitril/valsartan, without a significant excess of serious cases — a reassuring finding given the history in §28.6, and not a reason to relax about it.

What it cannot tell you. Two things. First, it cannot separate the contributions of neprilysin inhibition and angiotensin receptor blockade, because there was no arm receiving the ARB alone — the comparator was a different RAAS drug. The benefit is attributable to the combination, and the mechanistic story in §28.6 is an inference, not a trial result. Second, trials stopped early for benefit tend, as a class, to overestimate effect size, because stopping rules are triggered by favorable random fluctuation as well as by real effect. The direction of the result is not in doubt. The magnitude is plausibly a little generous.

An absolute difference of a few percentage points over roughly two years, on a composite of cardiovascular death and heart failure hospitalization, in patients already receiving good therapy, is a large result by the standards of modern cardiology — somewhere in the region of twenty patients treated for the trial's duration to prevent one primary event. Guidelines were revised, and the ARNI class moved into the position of preferred renin-angiotensin system inhibitor for HFrEF.

The population matters, and here is the proof

A later trial applied the same drug to heart failure with preserved ejection fraction — a different disease that shares a name. The result did not reach statistical significance. It came close, the direction was favorable, and subsequent analyses suggested benefit was concentrated in patients with ejection fractions toward the lower end of the "preserved" range. Regulators eventually broadened the labeled indication with language acknowledging that benefit is most clearly established below normal ejection fraction.

The ✅ in this chapter belongs to the HFrEF claim. It does not automatically extend to HFpEF, and a source that says "sacubitril/valsartan is proven in heart failure" without specifying which one has flattened a distinction the trials worked hard to establish. Rule 1: a rating attaches to a claim with a population and an endpoint.

🩺 Safety and Risk — the washout, the angioedema, and the biomarker trap

Three practical facts, none of which is a protocol and all of which are the kind of thing patients should understand well enough to ask about.

Do not combine with an ACE inhibitor, and do not switch without a gap. Because both neprilysin and ACE degrade bradykinin, blocking both at once compounds bradykinin accumulation and raises angioedema risk — swelling of the lips, tongue, and airway, which is uncommon and can be life-threatening. Sacubitril/valsartan is contraindicated with concurrent ACE inhibitor use, and the label specifies a mandatory washout interval when switching from one to the other. This requirement exists because of the earlier combined neprilysin–ACE inhibitor drug in §28.6, whose angioedema rate ended its development. History is encoded in the label.

Low blood pressure is the common problem. Symptomatic hypotension was more frequent than with the comparator. In a drug whose entire purpose includes vasodilation, that is expected rather than surprising, and it is the most common reason the medication is not tolerated.

The elegant one: this drug breaks the BNP test but not the NT-proBNP test. Neprilysin degrades BNP. Inhibit neprilysin and measured BNP rises — because you stopped clearing it, not because the heart got worse. NT-proBNP is not a neprilysin substrate, so it continues to fall as the patient improves. On this drug, BNP goes up while the patient gets better, and NT-proBNP behaves normally. A clinician who orders the wrong assay reads a rising number as deterioration and may escalate treatment for a patient who is responding.

This is the chapter in one fact. The same molecular system, in two roles, interacting — a therapy aimed at the peptide corrupts a biomarker derived from the same precursor, and the corruption is specific to the half of the precursor that the enzyme happens to recognize. Two roles, two molecules, one gene.

As always, decisions about any of this belong with a clinician who knows your history and your medication list, not with a book.

📊 Evidence Rating

Claim: Sacubitril/valsartan reduces cardiovascular death and heart failure hospitalization in adults with chronic heart failure with reduced ejection fraction, compared with enalapril. Rating: ✅ Strong clinical evidence (as of 2026). Why: PARADIGM-HF randomized 8,442 patients against an active comparator with a proven mortality benefit, was stopped early for benefit on cardiovascular mortality, and showed a hazard ratio of 0.80 (95% CI 0.73–0.87) for the primary composite with consistent effects on cardiovascular and all-cause death; guidelines changed on the strength of it. What would change it: a failed replication in the same population, or evidence that the early stop materially inflated the effect. Neither has occurred. Note that this rating covers HFrEF only; the preserved-ejection-fraction claim is a separate claim that did not meet its primary endpoint and does not inherit this ✅.

28.8 Downstream, sideways, and the frontier

Two more developments belong here, and each teaches something the main story does not.

Vericiguat: the same second messenger, through a different door

Section 28.2 said to remember cGMP. Here is why.

Natriuretic peptides make cGMP by activating a particulate guanylyl cyclase — the receptor itself, sitting in the membrane. There is a second, entirely separate enzyme that makes the same second messenger: soluble guanylate cyclase, which lives in the cytoplasm and is activated by nitric oxide rather than by a peptide. In heart failure, oxidative stress impairs nitric oxide signaling, and cGMP production through that route falls.

Vericiguat is a small molecule that stimulates soluble guanylate cyclase directly, raising cGMP without needing the nitric oxide pathway to be intact. It was tested in a randomized, placebo-controlled trial of roughly 5,050 patients with reduced ejection fraction who had recently had a worsening event — a hospitalization or an intravenous diuretic course — which is a deliberately high-risk population. The primary composite of cardiovascular death or first heart failure hospitalization gave a hazard ratio of 0.90 (95% CI 0.82–0.98). It was approved on that basis.

Two honest notes.

It is not a peptide, and it is not literally downstream of the natriuretic peptide receptor. It acts on a different cyclase. The connection is that both routes converge on the same second messenger. Calling it "the same pathway" is a convenient shorthand that is slightly wrong, and this book would rather be slightly awkward than slightly wrong. It is in this chapter because it shows that a signaling system can be addressed at several levels — the ligand, the enzyme that destroys the ligand, the receptor, or the intracellular messenger — and each level is a different drug with its own evidence.

The effect is modest and the population is narrow. A hazard ratio of 0.90 with an upper confidence bound of 0.98 is a real but small relative effect. It matters clinically because the absolute event rate in that population is very high, and a small relative reduction on a large risk is still a meaningful number of events. That is a general point worth carrying: relative and absolute effects answer different questions, and a modest relative effect in a high-risk group can outperform a large relative effect in a low-risk one.

CNP, and a use nobody was looking for

The third family member spent decades as the one with no obvious therapeutic angle. It barely circulates, it does little natriuresis, and its cardiovascular effects are local.

Then consider where else CNP acts: the growth plate. CNP signals through NPR-B on chondrocytes, and that signaling opposes the MAPK pathway downstream of the fibroblast growth factor receptor FGFR3.

Achondroplasia — the most common form of disproportionate short stature — is caused by a gain-of-function mutation in FGFR3. The receptor signals too much, chondrocyte proliferation at the growth plate is over-suppressed, and the long bones grow less than they otherwise would. If CNP signaling opposes that pathway, then a CNP analog should partially counteract the mutation's effect.

That reasoning produced vosoritide, a CNP analog engineered for a usable half-life — the natural peptide's is far too short. (If you did the Chapter 1 spaced review, you have already met this name and correctly guessed from the "-tide" stem that it was a peptide.) A randomized, placebo-controlled trial in children with achondroplasia measured annualized growth velocity over one year and found roughly 1.5 centimeters per year more growth in the treated group. It was approved on that basis in 2021.

Two things to be careful about, and they are the reason this example earns a place in a book about evaluating evidence.

The endpoint is a surrogate. Growth velocity over one year is not final adult height, and it is certainly not any measure of health, function, or quality of life. Growth velocity can increase without adult height increasing, if growth simply happens earlier. Longer-term follow-up accumulates, and the honest position as of this writing is that the velocity effect is established and the adult-height effect is being determined. Chapter 5 §5.6 taught you to ask that question; ask it here, of a drug this book rates positively.

Whether achondroplasia should be treated at all is genuinely contested, and not by cranks. Many people with achondroplasia and many advocacy organizations describe it as a form of human variation rather than a disease, and regard a growth-increasing drug given to children who cannot consent as raising serious ethical questions. Others, including many parents and some adults with the condition, point to real medical complications — spinal stenosis, foramen magnum compression, sleep apnea — and to a reasonable hope that a treatment might reduce them. This book takes no position on that question. It notes only that a positive evidence rating on a growth-velocity endpoint answers none of it, and that mistaking an efficacy rating for an endorsement is its own kind of category error.

📊 Evidence Rating — two claims at the edge

Claim: Vericiguat reduces cardiovascular death or heart failure hospitalization in adults with reduced ejection fraction and a recent worsening heart failure event. Rating: ✅ Strong clinical evidence (as of 2026), for that population and that endpoint. Why: A randomized placebo-controlled trial of about 5,050 patients gave a hazard ratio of 0.90 (95% CI 0.82–0.98) for the primary composite; approval followed. What would change it: a neutral replication would move this to ⚠️. Any use outside the studied high-risk window is a separate, currently unrated claim.

Claim: A CNP analog increases annualized growth velocity in children with achondroplasia over one year of treatment. Rating: ✅ Strong clinical evidence (as of 2026), for that endpoint only. Why: A randomized, placebo-controlled trial in children demonstrated roughly 1.5 cm per year greater growth velocity than placebo, and the drug was approved on that result. What would change it: nothing likely for the velocity claim. The separate claim that it increases final adult height, or improves any health or functional outcome, is ⚠️ — it depends on long-term follow-up that is still accumulating. Two claims about one molecule, two ratings; you should be expecting this by now.

Genuinely frontier

Three things remain open, so you can recognize them when you meet them.

Engineered chimeric natriuretic peptides combine features of different family members, aiming to keep the renal effects while limiting the blood-pressure drop that constrains the natural peptides. Investigational. 🔬

Natriuretic peptide-guided prevention — using the biomarker to find people at risk of developing heart failure and intervening early — has some encouraging randomized evidence and nothing like the weight behind the diagnostic claim. ⚠️ at best, and note that it is a fifth distinct claim about the same measurement.

Recombinant ANP is approved in some countries for acute heart failure on an evidence base that has never included the kind of large randomized outcome trial that answered the nesiritide question. Given §28.5, the appropriate posture is skeptical and the appropriate action is a trial.

28.9 What this class teaches: a system is not a strategy

Assemble the record.

ONE PEPTIDE SYSTEM, THREE THINGS DONE TO IT

  MEASURE IT     BNP / NT-proBNP as a diagnostic aid        WORKED     ✅
                 → reorganized the evaluation of breathlessness worldwide

  SUPPLY IT      infused recombinant BNP in acute HF        DID NOT    ❌
                 → improved every measurement, changed no outcome
                 → and a second peptide, tested independently, failed the same way

  PRESERVE IT    neprilysin inhibition + ARB in chronic HF  WORKED     ✅
                 → hazard ratio 0.80, trial stopped early, guidelines rewritten

  Same physiology underneath all three. The biology did not determine the result.
  The STRATEGY did.

Here is the generalizable claim, and it is the sentence to carry out of this chapter:

"Targeting a system" is not a strategy. It is a category containing several strategies with very different odds, and which one you pick matters more than which system you picked.

People — including researchers, including investors, including the people writing the press release — routinely talk as though identifying the right biological system is the hard part and the drug is an implementation detail. This chapter is the counterexample. The system was correctly identified in 1981. Everything since has been about how to intervene on it, and the how produced a resounding success, a resounding failure, and a diagnostic test more valuable than either.

Why did preserving beat supplying?

No one can prove the answer, but four explanations are on the table and each of them generalizes.

Timing and setting. Nesiritide was tested over hours to days in acutely ill patients already receiving treatments that produce the same physiological effects. Sacubitril/valsartan was tested over years in chronic disease, where remodeling — the slow structural deterioration that natriuretic peptides oppose — has time to be affected. A drug that modifies disease progression cannot show its value in 48 hours.

Physiological pattern. This is Chapter 3, and it may be the deepest of the four. A preserved endogenous peptide is still released by the body, on the body's schedule, in the tissues where it is produced, in proportion to actual load. An infusion is flat, systemic, and indifferent to what the heart is doing minute to minute. Preservation amplifies a regulated signal; supply replaces it with an unregulated one. In a system already showing receptor downregulation and resistance (§28.2), that difference is plausibly decisive.

The one-peptide framing may be wrong. Neprilysin inhibition raises several peptides, and attributing the benefit to BNP — a poor neprilysin substrate — was never well founded. Possibly the trial succeeded because it changed a network rather than a molecule.

The comparator and the endpoint. PARADIGM-HF beat an active comparator on a hard endpoint over years; the nesiritide trial added a drug to standard care and measured 30-day events. The difference in answer is partly a difference in what was asked.

Only the second of those four is really about peptides. The rest are about trial design and disease biology, which is why Chapter 5's machinery does more work in this book than any amount of molecular detail.

One more thing, and it is uncomfortable for a book about peptides

The successful therapy in this chapter is not a peptide. Sacubitril and valsartan are both small molecules. Vericiguat is a small molecule. The biomarker is a peptide fragment, and the CNP analog is a peptide, but the drug that changed how heart failure is treated contains no peptide at all.

The peptide was the target, not the drug. That is a genuinely common outcome, and it is a useful corrective to the framing that treats "peptide therapeutics" as a coherent field with a shared destiny. Sometimes the best way to act on a peptide system is with a molecule a tenth its size that can be swallowed. Chapter 1 §1.6 laid out the trade-offs of the size spectrum; this chapter is what they look like when a real program has to choose.

📋 Your Evidence Dossier

This chapter does not add a new field. It adds a rule for filling in the fields you already have, and it is the rule that will most improve your dossier's accuracy.

The rule: one row per claim, not one row per molecule — and every row names the study design that could settle it.

Most people build a reference document with one entry per compound and a verdict at the top. That structure guarantees error, because it has nowhere to put the fact that the same molecule can be excellent for one purpose and useless for another. Rule 6 of the rating system is not a philosophical nicety. It is a data structure requirement.

The worked demonstration — BNP and NT-proBNP in two roles

DOSSIER ENTRY — B-TYPE NATRIURETIC PEPTIDE SYSTEM     [worked demonstration]

  MOLECULE            BNP / NT-proBNP (fragments of one precursor, proBNP)

  ROLE 1 — MEASURED (the molecule as information)
    Claim ..........  Aids diagnosis / exclusion of heart failure in adults
                      presenting with undifferentiated breathlessness
    Population .....  Adults with acute dyspnea; adults with suspected chronic HF
    Endpoint .......  Diagnostic accuracy against a blinded reference diagnosis
    Design needed ..  Prospective diagnostic accuracy study
    Rating .........  ✅
    Caveats ........  Age ↑, renal impairment ↑, obesity ↓, atrial fibrillation ↑

  ROLE 1b — MEASURED, DIFFERENT USE
    Claim ..........  Titrating therapy to an NT-proBNP target improves outcomes
    Design needed ..  Randomized trial of a management STRATEGY
    Rating .........  ❌  (trial stopped for futility)

  ROLE 2 — ADMINISTERED (the molecule as drug)
    Claim ..........  Infused recombinant BNP improves death / rehospitalization
                      in acute decompensated heart failure
    Population .....  Adults hospitalized with acute decompensated HF
    Endpoint .......  Death or HF rehospitalization at 30 days
    Design needed ..  Randomized placebo-controlled outcome trial
    Rating .........  ❌  (~7,100 patients; no benefit demonstrated)

  ROLE 3 — PRESERVED (the system as target)
    Claim ..........  Neprilysin inhibition + ARB reduces CV death and HF
                      hospitalization in HFrEF
    Population .....  Chronic HFrEF, symptomatic, on background therapy
    Endpoint .......  CV death or HF hospitalization
    Design needed ..  Randomized active-controlled outcome trial
    Rating .........  ✅  (HR 0.80; stopped early; guideline-changing)
    Scope note .....  HFrEF only. The HFpEF claim is separate and is NOT ✅.

  ONE GENE. FOUR CLAIMS. TWO ✅ AND TWO ❌.

Do this for your own entries

Take one peptide from your dossier — ideally one you feel confident about — and split it into roles the way the demonstration does. Ask, for each:

1. Is this molecule ever used as a measurement rather than a treatment? What is that claim rated, separately?
2. What exactly is the population, and would the claim survive changing it?
3. What is the endpoint, and is it a surrogate or something a patient would notice?
4. What study design would settle this claim, and does that design exist for it?
5. If your entry currently has one rating, what did that single rating hide?

Question 5 is the one that does the work. When you started this project in Chapter 1, most of your entries probably had a single mental verdict attached — works, does not work, unproven — and by now you have seen enough compounds to know that a verdict is almost always the wrong shape for the evidence.

Conclusion

The heart is a sensor as well as a pump. It measures the tension in its own walls and converts that measurement into peptides — ANP from the atria, BNP from the ventricles, both signaling through a receptor that is itself an enzyme, both telling the kidney and the blood vessels to shed volume, lower pressure, and turn down the system that is retaining salt. CNP, the third family member, does something else entirely, mostly in blood vessel walls and in cartilage.

From that one piece of physiology, medicine built three things.

It built a blood test, by measuring the inactive fragment of the BNP precursor, which turned out to be more stable than the active peptide and to report the same signal. That test changed how breathlessness is evaluated, and it is limited in known, directional ways by age, kidney function, obesity, and atrial fibrillation. ✅

It built a drug — the actual human peptide, infused into acutely ill patients. It lowered filling pressures and improved symptoms and was approved on that basis, and when the hard-endpoint trial was finally done in seven thousand patients, the outcomes did not move. ❌

And it built a strategy for not throwing the peptide away, pairing inhibition of the degrading enzyme with an angiotensin receptor blocker — necessarily, because the same enzyme degrades angiotensin II and because blocking bradykinin's other disposal route causes angioedema. That combination beat a proven active comparator, stopped early, and rewrote guidelines. ✅

No single verdict on "natriuretic peptides" could be right about more than one of those. This is the chapter where ignoring the claim-not-molecule rule produces not a slightly imprecise answer but an answer that is wrong in three directions simultaneously.

Carry two things forward. The first is the biomarker-versus-therapy distinction, which recurs everywhere — in Chapter 30's cosmetic ingredients that can be detected doing something in skin without changing how skin looks, and in every Part III compound whose supporters point to a laboratory measurement and call it evidence of benefit. The second is the sentence from §28.9: targeting a system is a category, not a strategy, and the choice within the category is where the outcomes actually diverge.

Chapter 29 continues in the same territory with a different set of molecules — and by now you should be reading every new peptide the same way: not does it work, but which claim, in which population, on which endpoint, settled by which design.

Key Terms

Natriuretic peptide — a member of a family of cardiac and endothelial peptides that promote sodium excretion, vasodilation, and suppression of the renin-angiotensin-aldosterone system. ANP, BNP, and CNP are the three principal members.

ANP (atrial natriuretic peptide) — a 28-amino-acid peptide stored preformed in atrial muscle granules and released rapidly on atrial stretch. Fast-acting and very short-lived.

BNP (B-type natriuretic peptide) — a 32-amino-acid peptide produced largely by ventricular muscle in response to wall stress, synthesized on demand rather than stored. The "B" originally stood for "brain," where it was first isolated; the name is a historical accident.

CNP (C-type natriuretic peptide) — a 22-amino-acid peptide made mainly by endothelium and by chondrocytes, signaling through NPR-B. Minimal natriuretic activity despite the family name; important in vascular tone and in growth-plate biology.

proBNP — the 108-amino-acid precursor cleaved into active BNP and inactive NT-proBNP.

NT-proBNP — the 76-amino-acid N-terminal fragment of proBNP. Biologically inactive, longer-lived, more stable in the specimen, and consequently the more practical measurement. Not a neprilysin substrate, which matters on ARNI therapy.

Natriuresis — excretion of sodium in the urine. Accompanying water loss is diuresis.

Wall stress — the mechanical tension in the heart muscle wall, rising with chamber pressure and chamber size. The stimulus for natriuretic peptide release.

Endocrine organ — a tissue that secretes a substance into the bloodstream to act on distant tissue via a receptor. The heart qualifies; recognizing that took until 1981.

Guanylyl cyclase / cGMP — the enzyme activity built into NPR-A and NPR-B, and the second messenger it produces. Cyclic GMP activates protein kinase G and is the common currency linking natriuretic peptide signaling to nitric oxide signaling.

NPR-A, NPR-B, NPR-C — the natriuretic peptide receptors. NPR-A binds ANP and BNP; NPR-B binds CNP; both are guanylyl cyclases. NPR-C does not signal — it binds and internalizes all three for degradation.

Natriuretic peptide resistance — the observation that in heart failure, circulating natriuretic peptide levels are high while their physiological effect is blunted, through receptor downregulation, increased clearance, and impaired precursor processing.

RAAS (renin-angiotensin-aldosterone system) — the hormonal system that raises blood pressure and retains sodium. The natriuretic peptide system is its physiological counterweight.

Neprilysin — a membrane-bound zinc peptidase that degrades natriuretic peptides, bradykinin, substance P, adrenomedullin, and angiotensin II, among others. The target of the neprilysin inhibitor component of an ARNI.

ARNI (angiotensin receptor–neprilysin inhibitor) — the drug class combining neprilysin inhibition with angiotensin receptor blockade. The pairing is mechanistically required, not a convenience.

Nesiritide — recombinant human BNP, given by infusion. Approved for acute decompensated heart failure on hemodynamic and symptomatic grounds; a large outcome trial did not demonstrate benefit.

HFrEF / HFpEF — heart failure with reduced ejection fraction, and heart failure with preserved ejection fraction. Different diseases sharing a name; evidence in one does not transfer to the other.

Ejection fraction — the proportion of blood in the left ventricle expelled with each beat. The principal way heart failure populations are subdivided in trials.

Surrogate endpoint / hard endpoint — a surrogate is a measurement expected to predict clinical benefit, standing in for an outcome patients experience; a hard endpoint is the outcome itself. Filling pressure is a surrogate; death is not. See Chapter 5 §5.6.

Hazard ratio — the ratio of event rates between treatment and control over time. A hazard ratio of 0.80 corresponds to roughly a 20 percent relative reduction; it says nothing on its own about the absolute number of events prevented.

Angioedema — swelling of the deeper layers of skin and mucosa, potentially airway-threatening, associated with bradykinin accumulation. The reason ARNIs pair with an ARB rather than an ACE inhibitor, and the reason a washout is required when switching.

Soluble guanylate cyclase — the cytoplasmic, nitric-oxide-responsive enzyme that produces cGMP. A separate enzyme from the receptor-linked cyclases, converging on the same second messenger.

Vericiguat — a small-molecule stimulator of soluble guanylate cyclase, approved for heart failure with reduced ejection fraction after a recent worsening event.

Vosoritide — a CNP analog approved for achondroplasia, on an endpoint of annualized growth velocity.

Achondroplasia — the most common form of disproportionate short stature, caused by a gain-of-function mutation in FGFR3 that over-suppresses growth-plate chondrocyte proliferation.

Spaced Review

1. (Ch 28) A patient with known atrial fibrillation and moderate kidney impairment presents breathless, and the NT-proBNP is elevated. Name each confounder that could be contributing, state the direction it pushes the number, and explain why the test is less useful in this patient than in the emergency department populations where it was validated.

2. (Ch 5 + Ch 28) Nesiritide was approved on a reduction in pulmonary capillary wedge pressure and an improvement in dyspnea at three hours. Classify each of those two endpoints using Chapter 5 §5.6's vocabulary, and state what a regulator gains and loses by accepting them.

3. (Ch 3 + Ch 28) Section 28.2 describes natriuretic peptide resistance in heart failure — receptors downregulated under sustained high levels. Explain how that phenomenon connects to Chapter 3's account of pulsatile versus continuous signaling, and use it to argue why a continuous infusion of BNP might have been unlikely to succeed even before the trial was run. Then state clearly why that argument, however good, is not evidence.

4. (Ch 27 + Ch 28) Chapter 27 described two ways to act on one hormonal axis — a receptor agonist and a receptor antagonist — that produce opposite immediate effects while serving related clinical goals. Chapter 28 describes three ways to act on one peptide system: measure, supply, preserve. Write two sentences explaining what these two chapters together suggest about how you should react the next time someone tells you a company is "targeting" a promising system.

5. (Ch 28, synthesis) Someone hands you an article headlined "Natriuretic Peptides: Miracle or Mirage?" Without reading it, write the three questions you would need answered before the headline could mean anything, and explain why the headline's framing guarantees that at least two of its possible answers are wrong.