Chapter 28 — Key Takeaways

Natriuretic Peptides: The Heart Failure Biomarkers That Became Treatments

The one sentence

Three interventions were made on one peptide system — measure it, supply it, preserve it — and the answers were yes, no, and yes, which means no single verdict on "natriuretic peptides" can be right about more than one of them.

The biology

• The heart is a sensor as well as a pump. Cardiac muscle releases peptides in response to stretch, from volume overload or pressure overload. De Bold's 1981 experiment — atrial extract injected into rats produced rapid natriuresis, ventricular extract did not — established it, and it was delayed for two decades because "endocrine organ" was treated as a biological category rather than an administrative one.

• Three family members, different jobs. ANP comes predominantly from the atria, stored preformed and released fast. BNP comes predominantly from the ventricles, made on demand and therefore slower and steadier. CNP is mainly endothelial and chondrocytic, signals through a different receptor, and does very little natriuresis despite the family name.

• The receptor is an enzyme. NPR-A and NPR-B have guanylyl cyclase built into their intracellular domain; binding the peptide makes cGMP directly. NPR-C binds all three and does not signal — it is a clearance receptor.

• The effects oppose RAAS at every point: sodium and water excretion, vasodilation, suppression of renin and aldosterone, and restraint of hypertrophy and fibrosis.

• The paradox to hold on to: in heart failure, natriuretic peptide levels are high and the effect is blunted — natriuretic peptide resistance, from receptor downregulation, increased clearance, and impaired precursor processing. What the assay detects and what is biologically active are not quite the same population of molecules.

The measurement

• proBNP is cleaved into active BNP and inactive NT-proBNP. Both are measurable. NT-proBNP is more stable — in the blood and in the specimen — which makes the useless half of the molecule the more useful measurement.

• The two numbers are not interchangeable. Reference ranges differ by roughly an order of magnitude. A result means nothing until you know which assay produced it.

• Because levels rise in proportion to cardiac wall stress, measuring them helps distinguish breathlessness caused by heart failure from breathlessness caused by lung disease — a distinction that previously required considerably more guesswork. It is one of the most useful blood tests in cardiology.

• It is a rule-out test. High sensitivity, moderate specificity: a low value is informative, a high value keeps the question open.

• Four confounders, each with a known direction. Age ↑. Kidney impairment ↑ (more so for NT-proBNP). Obesity ↓ — the dangerous one, because it is the only one that undermines the test's rule-out strength. Atrial fibrillation ↑, sometimes markedly.

• An aid to judgment, not a verdict. That formulation applies to every biomarker in this book.

The central argument

The same peptide system supports different claims, each requiring a different study design, each with a different answer:

• 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.

• Rule 3 exists for this chapter. Never upgrade a rating with mechanism. And note the sharper version: a better mechanism should make you more careful, not less, because an elegant mechanism makes a negative trial feel like an error rather than an answer.

The therapies

• Nesiritide — recombinant human BNP, approved for acute decompensated heart failure on hemodynamic and symptomatic grounds. A subsequent randomized outcome trial of roughly 7,100 patients found no meaningful effect on death or rehospitalization. Use declined sharply. This is a Chapter 5 §5.6 surrogate story: approval on physiological and symptomatic measures, a hard-endpoint trial that did not confirm. Note carefully: absence of benefit, not presence of harm. A second natriuretic peptide, tested independently, failed the same way.

• Neprilysin degrades natriuretic peptides — and bradykinin, adrenomedullin, substance P, and angiotensin II. Inhibiting it raises endogenous peptides rather than supplying more from outside.

• The ARB is mechanistically required, not a marketing convenience. Neprilysin degrades angiotensin II, so inhibiting neprilysin alone would raise it and could worsen the disease. And it must be an ARB rather than an ACE inhibitor: both neprilysin and ACE degrade bradykinin, and an earlier combined neprilysin–ACE inhibitor produced angioedema often enough that it was never approved. Hence the contraindication with ACE inhibitors and the mandatory washout.

• Sacubitril/valsartan in PARADIGM-HF: 8,442 patients, compared against enalapril rather than placebo, stopped early, primary composite 21.8 vs 26.5 percent, hazard ratio 0.80 (95% CI 0.73–0.87), all-cause mortality also reduced. Guidelines changed. The HFpEF claim is separate and did not meet its primary endpoint — it does not inherit the ✅.

• The most elegant fact in the chapter: neprilysin degrades BNP but not NT-proBNP, so on this drug BNP rises while the patient improves and NT-proBNP behaves normally. A therapy aimed at the peptide corrupts a biomarker derived from the same precursor.

• Vericiguat stimulates soluble guanylate cyclase — a different enzyme from the receptor-linked cyclases, reaching the same second messenger. Approved for HFrEF after a recent worsening event; HR 0.90 (95% CI 0.82–0.98). It is a small molecule, not a peptide, and it is here to show that a signaling system can be addressed at several levels.

• CNP found its use outside cardiology. A CNP analog is approved for achondroplasia, where CNP signaling opposes the over-active FGFR3 pathway at the growth plate. The endpoint is annualized growth velocity — a surrogate. Final adult height and functional outcomes are a separate, less settled claim.

The generalizable lesson

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

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.

And the therapy that worked is not a peptide. Sacubitril and valsartan are both small molecules. The peptide was the target, not the drug — a useful corrective to any framing that treats "peptide therapeutics" as a field with a shared destiny.

Ratings issued in this chapter

Six ratings across one physiological system. Four molecules. Two of the ✅ and one of the ❌ attach to the same gene product. That is rule 6 — one molecule, many ratings — demonstrated as thoroughly as this book can demonstrate it.

Dossier discipline added here

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

Ask of each entry: is this ever used as a measurement rather than a treatment? What is the population? Is the endpoint a surrogate or something a patient would notice? What design would settle it, and does that design exist? And the question that does the work: if your entry has one rating, what did that single rating hide?