The Fragment Problem
Why a Fragment of a Hormone Isn't a Smaller Hormone
Cut a piece out of a hormone and you get a piece of a hormone. The assumption that it keeps working is exactly that — an assumption.
Behind a whole category of peptide products sits a single argument. It's rarely stated, because stating it plainly makes the problem visible.
Here it is:
1. Growth hormone burns fat. 2. Residues 176–191 of growth hormone are involved in that. 3. Therefore the peptide made of residues 176–191 burns fat.
Steps one and two can both be perfectly true. Step three does not follow from them.
That's the whole article. Everything below explains why the third step breaks, why it nonetheless sounds reasonable, and what happened when somebody actually put it to the test.
Why it sounds right
The reasoning has an obvious appeal: isolate the part that does the work and discard the rest. You get the effect without the baggage. It's how we think about machines — remove the component responsible for a function, and you have that function in a smaller package.
Proteins don't work that way, and the reason is shape.
That last line is where the fragment argument fails.
Break one: the residues lose their positions
In a folded protein, a working region is held in a particular arrangement by everything around it. The surrounding structure acts as a scaffold, holding those residues at specific angles and specific distances from one another.
Cut them out and the scaffold is gone.
What you have left is a short, floppy chain. The same amino acids in the same order — and no longer arranged the way they were. A peptide of ten or twenty residues is generally too short to fold into anything stable on its own. It flops around in solution, sampling countless shapes, mostly not the one that mattered.
Think of a key. The teeth do the work in the lock — that's the functional part. But cut off just the teeth and you can't open anything. You've lost the shaft that sets their depth and the bow that lets you apply torque. The working part was never working alone.
Break two: some surfaces can't be cut out at all
This one is more fundamental, and it rules out fragments entirely in many cases.
When a protein binds something, the contact surface is often assembled from residues that are far apart in the sequence and brought together by folding. Residue 12 sits next to residue 47, which sits next to residue 80, and together the three of them form one pocket.
There is no contiguous stretch you can cut out that reproduces that pocket. Take residues 45–50 and you have one wall of a structure that no longer exists. Take all three regions and you have three disconnected pieces with nothing to hold them in relation to each other.
Whether a fragment can work at all depends on whether the relevant region happens to be one continuous stretch that folds independently. Sometimes it is. Frequently it isn't. And you can't tell from the sequence alone — it takes structural work to find out, and testing to confirm.
Break three: it's a different molecule in your body
Even setting shape aside, a fragment isn't pharmacologically equivalent to its parent.
It has two new exposed ends. Every fragment has a fresh start and a fresh finish where the chain was cut. Those are precisely the sites the enzymes described in Article 3 attack first — vulnerabilities the intact protein never had.
It clears differently. Different size, different charge, different distribution through the body, different route out. A fragment can have a half-life measured in minutes where the parent lasted hours.
It may have lost machinery it needed. Proteins often use one region to bind and another entirely for something else — reaching the right tissue, pairing with a partner molecule, holding a cofactor. Keep the binding region and discard the rest and you may have kept the part that acts without the part that gets it where it acts.
The case that was actually tested
Here's what makes the growth hormone fragment worth dwelling on: the hypothesis wasn't just asserted. It was tested properly, in humans, at scale.
The idea originated in animal research decades ago. The region near the tail end of growth hormone appeared to be associated with fat metabolism, and it seemed separable from the growth-promoting and blood-sugar effects that make growth hormone problematic. A pharmaceutical developer took a modified version of that fragment forward as an obesity drug candidate — AOD-9604.
It went into human clinical trials. Real ones — randomized (participants assigned to treatment or comparison by chance, so the groups don't differ systematically), controlled (a comparison group receiving something inert, so the drug's effect can be separated from everything else that happens during a trial), and powered to detect a weight-loss effect.
That last word matters most here. A trial's power is its statistical ability to find an effect of a given size if one is really there — driven mainly by how many people are enrolled. An underpowered trial can miss a real effect, which makes its negative result uninformative. A properly powered one that finds nothing has produced actual evidence of absence.
It did not meet its primary endpoint. The largest study enrolled 536 participants and compared AOD-9604 with placebo over 12 weeks. The fragment did not produce clinically meaningful weight loss in humans. Development for obesity was discontinued. FDA's review of the available evidence likewise concluded that there was insufficient evidence of effectiveness for obesity.
This is unusually clean evidence. Most unapproved compounds sit in a fog where you can't say much beyond "untested." This one isn't untested. The fragment hypothesis was examined with the tools designed to examine it, and the result was negative for the headline claim.
One more distinction worth knowing: AOD-9604 is a modified analog of the native growth-hormone fragment 176–191 — related, but not the same molecule. Products sold as "GH fragment 176–191" and studies of AOD-9604 don't necessarily describe the same thing. Evidence for one doesn't automatically transfer to the other. The same pattern shows up throughout this category: similar isn't necessarily identical.
Products built on that same fragment continue to be sold with that same claim.
The fragment that worked
None of this means fragments can't work. One of the more important osteoporosis drugs is a fragment, and looking at why it succeeded shows exactly what the others are missing.
Parathyroid hormone is 84 amino acids. Teriparatide is the first 34 — PTH(1–34).
The truncation was rational: the front end of the molecule carries the region that activates the receptor, and it turns out to fold and function independently of the rest. That's the favourable case described above — a continuous stretch that stands on its own.
But rational wasn't the point. The point is that they proved it. Teriparatide went through the full development process, including a large randomized trial in postmenopausal women measuring an outcome that actually matters — fractures, not a surrogate marker. It reduced them. It was approved on that basis and has been in clinical use for over two decades.
Same starting logic as the growth hormone fragment. Opposite outcome — and the difference wasn't the reasoning. It was that somebody ran the experiment and the experiment came back positive.
The question to ask
For any fragment product, one question does most of the work:
Was this fragment itself tested for the effect being claimed — or is the claim inherited from the parent molecule?
If the evidence describes what the whole hormone does, then discusses which part is "responsible," you're looking at an assertion. The chain of reasoning may be entirely sound. It's still an assumption until someone tests the fragment.
And where testing has happened, the result is what counts — including when it's negative. A fragment that went to clinical trial and missed its endpoint has more evidence against it than a fragment nobody ever studied. That's worth knowing, and it's rarely mentioned in the places these are discussed.
Next, a different unexamined assumption — this one about dose. There's a class of compound where increasing the amount cannot increase the effect past a certain point, for reasons built into how your body regulates itself. Most protocols using them are written as though that ceiling doesn't exist.
This article is educational and not medical advice. It does not recommend any compound, dose, or protocol. Several compounds mentioned are not approved for human use. Decisions about any therapy belong with a qualified clinician who knows your history.