# Three Evidence Bases, Priced Side by Side

> The value ledger: what each evidence base cost — A side-by-side value ledger for Research Peptide Fundamentals — semaglutide, NAD+ and GHK-Cu compared on sample size, duration, comparator and endpoint type, the four things that set the price of a finding.

**THE LEDGER**

Not which compound is best. Which compound's claims are covered by what was actually paid for them — graded on sample size, duration, comparator and what was measured at the end.

## The short version

This page puts the three compounds next to each other on one measure: how much their evidence cost to produce. That is deliberately not the same question as which one is most useful, and it is definitely not a question about products or where to obtain them.

The reasoning is simple. Every claim in this field arrives with an implied confidence, and confidence should track what was spent buying it. A trial that randomly assigned seventeen thousand people to a placebo and then counted strokes for years spent an enormous amount. A gene-expression table generated from a public database spent almost nothing. Both can be true. They cannot be believed to the same degree.

The ledger below grades on four things only: how many people, how long, against what comparison, and what was measured at the end. Those four set the price of any human finding, and the differences between these three compounds fall out of them almost entirely.

## The value ledger

| | Semaglutide | NAD+ | GHK-Cu |
|---|---|---|---|
| **What it is** | Engineered 31-residue GLP-1 analogue | Endogenous redox coenzyme, taken as precursors | Three-residue peptide chelated to copper(II) |
| **Regulatory standing** | Approved prescription medicine, several indications | Not an approved drug; sold as a supplement | No approved indication; topical cosmetic ingredient |
| **Largest controlled human study cited** | 17,604 adults, cardiovascular outcomes [3] | 45 to a few hundred adults per precursor trial [7][10] | 45 men, six months, combination formula [15] |
| **Longest cited follow-up** | 72 weeks [1] | 60 days for the multicentre precursor trial [7] | Six months [15] |
| **Randomised placebo comparator** | Throughout, in every headline result | Present in the key precursor trials [7][10] | Present in the hair trial and some small topical studies [15] |
| **What was measured** | Clinical events: heart attack, stroke, kidney failure, death [2][3] | Blood NAD+ level, plus one insulin-sensitivity clamp study [8][10] | Cosmetic appearance, hair count, collagen response, gene expression |
| **Independent replication** | Extensive, across multiple trial programmes | Two independent precursors give a consistent marker result | Limited; much of the foundational literature is one group's |
| **What the evidence licenses** | Outcome-level statements, in the populations studied | "Measurably changes a human biomarker" and little beyond it | "Plausible topical cosmetic effect"; systemic claims unbought |
| **Where it thins out** | Narrowly, at the edges of the enrolled populations | At the step from marker to benefit [6] | At the step from laboratory mechanism to human outcome |

## The exchange rate between markers and outcomes

The single most useful idea for reading this field is that surrogate markers and clinical outcomes are different currencies, and the exchange rate between them is unknown until somebody buys it.

NAD+ makes the point cleanly. Nicotinamide riboside raised whole-blood NAD+ by 22%, 51% and 142% across ascending doses over eight weeks [10], and oral NMN raised it dose-dependently over 60 days in a multicentre double-blind design [7]. Both are solid, and both purchase the same thing: proof that the marker moves. What they do not purchase is the inference that moving it improves health, which is exactly what the 2025 human synthesis declines to grant, describing efficacy as limited and tissue-level data as sparse [6].

Semaglutide is the counter-example, and the reason it anchors this ledger. Its headline results were not markers. They were counts of events in people — a hazard ratio of 0.80 for major adverse cardiovascular events in 17,604 adults [3], and 0.76 for major kidney events in 3,533 [2]. Buying an outcome directly removes the exchange-rate problem, and it is expensive precisely because it cannot be shortcut.

GHK-Cu sits at the far end. Its most-quoted evidence is a gene-expression shift, reported at 31.2% of human genes at a 50%-or-greater change threshold [14]. Gene expression is a currency two steps removed from an outcome, and the source itself notes the popular round figure of roughly 4,000 genes overstates its own table [14].

## Why sample size is not a technicality

Small studies do not simply produce less certain versions of the same answer. They produce a systematically different distribution of answers, because a result has to be larger to reach significance in a small sample — so the small studies that do get published tend to report bigger effects than the truth.

That asymmetry is why a six-month, 45-man trial [15] and a 72-week, 751-participant head-to-head [1] cannot be read with the same posture, even though both are randomised. Both are legitimate. Only one is well-powered enough that the reported effect size can be taken close to its face value.

It is also why the trial-versus-marketing gap widens as evidence gets cheaper. Where the evidence base is a handful of small studies, the space around it fills with mechanism, extrapolation and confident summary — because there is room. Where the evidence base is a set of large event-driven trials, the confident summaries have to stay inside what the trials found, because someone will check.

## What the ledger deliberately does not say

It does not rank the compounds by usefulness. A cheap evidence base does not mean a compound does nothing, and an expensive one does not make a compound appropriate for any given person. Topical Copper Tripeptide-1 has a long cosmetic safety record and modest supporting trials; that combination is perfectly coherent, and it is not a failure.

It does not rank them by risk. Cost of evidence and margin of safety are independent axes. Semaglutide has the best-characterised evidence in this hub *and* the most fully documented adverse-effect profile, including gastrointestinal intolerance as the leading cause of discontinuation and an increased risk of biliary disease [5] — precisely because it was studied enough for those things to be counted.

And it does not price products. There are no prices on this desk, no vendors, no sources and no comparisons of where anything might be obtained. The only currency here is evidence, and the only question is whether a given claim has been paid for.

---

An independent literature digest that prices findings rather than products: the only currency weighed on this desk is evidence, nothing here is sold, and no dose is recommended to anyone.