Semaglutide & Its Friends

Beware the composition effect of weight loss

Scott Alexander has a good write-up on Semaglutide and some of the more novel GLP-1 agonists now burgeoning in popularity here and a good follow-up with curated user comments here. The latter is more useful in my view (hive mind, etc.). He focuses mostly on the economics of its use, which is an interesting topic unto itself. It deals with differing trade names and price points based on indication (for use in type 2 diabetes vs in obesity treatment). At present, the eligibility criteria for obtaining a Semaglutide prescription for weight loss is fairly onerous, and the price is prohibitive (on the order of $1000/month without prescription). Licensed generics may hit market by the end of 2023 but will likely still be fairly cost prohibitive for use in weight loss without a prescription or a type 2 diabetes diagnosis. As with anything in the prescription drug market, there is considerable opacity and price discrimination.

Either way, these compounds are only going to continue growing in popularity. Elon Musk even recently indicated that he is taking Wegovy (Novo Nordisk tradename for 2.4 mg Semaglutide for weight loss).

5 year Google trend data on “Semaglutide”

5 year Google trend data on “Ozempic” (Novo Nordisk tradename for Semaglutide for treatment of type 2 diabetes).

GLP-1 agonists have been around for awhile but only gained FDA approval last year. In fact, something called the ‘incretin effect’ has been known since ~1932. Per wikipedia:

The incretin effect describes the phenomenon whereby oral glucose intake elicits a higher insulin response compared to intravenously introduced glucose that produces the same levels of serum glucose levels.[4]

In effect, these are gastric peptides that signal feeding to the body to elicit faster serum glucose clearance (mostly via modulation of pancreatic enzyme output) and slower digestion to facilitate nutrient uptake in the gut. It makes sense that that the stomach would have preliminary, first-pass nutrient and energy sensors in addition to those in the liver and pancreas. Eating is how we get our energy after all.

An interesting corollary to this is that this is a primary mechanism by which parenteral nutrition (food by IV) given to cancer patients can circumvent the emesis/vomiting response brought about by cytotoxic chemo and radiation treatments. It effectively avoids incretin release in the stomach.

Endogenous incretins like GLP-1 have a short terminal half-life, so the modern incretin-like analogs (dulaglutide (Trulicity), exenatide (Byetta), liraglutide (Victoza), semaglutide (Ozempic, Wegovy and Rybelsus) and exenatide extended-release (Bydureon)) have been designed to be long-lasting with half-lives on the order of one week. They have poor oral bioavailability (though some have been designed to be taken orally), so most are injected subcutaneously once per week, typically in the abdomen. Wegovy and Ozempic come in pre-loaded injection pens with ease of functionality.

The upshot is that GLP-1 agonists tend to be insulinotropic (they increase insulin release from the beta cells of the pancreas) and decrease glucagon release from the alpha cells. This increases serum glucose clearance, lowers blood sugar, and slows gastric emptying. The last mechanism is how these molecules induce weight-loss: by increasing satiety and attenuating the desire to eat. At a crude pass, this is a bit of a vindication for the isodynamic calorie-in-calorie-out model (CICO) that is so anathema to some despite being broadly (and fairly intuitively) true.

In some ways it’s like a pharmacologically initiated gastric bypass surgery, likely the most effective weight loss intervention in the armamentarium, as assessed by sustained weight reduction and all-cause mortality rates.

According to one trial, 2.4 mg weekly semaglutide works for between ~60-80% of patients and reduces voluntary energy intake by 24% over a 12 week trial (n=30) with an attendant average weight loss of ~10 lbs. Other studies show a year long intervention to result in an average of ~15-20% reduction in body weight over a year of use. Most studies report that a majority of the weight loss is fat/adipose tissue, but I want to remonstrate with that below.

There is a lot more to be said mechanistically about these drugs, but this goes beyond the scope of this article. Peter Attia has a good but paywalled precis on this drug class and some ancillary concerns dating from late last year.

Don’t Cowboy GLP-1 Agonists

Now, this wouldn’t be a contrarian blog with a contrarian take: Semaglutide isn’t a great drug, and I think unfavorable side effect profiles will emerge with sustained use among large patient populations. There are complex biochemical reasons for this assertion which I’ll attempt to bolster below.

But first, an anecdote (with the requisite qualifier that the plural of anecdote isn’t data, bla bla)… I got interested in this drug last year and took it for 6 weeks at the follow-up ramp-up intervals: 2 weeks of 1 mg/weekly, 2 weeks, of 1.5 mg/weekly, and 2 weeks 2 mg/weekly. Although I’m 11% body fat (as measured by not-entirely-accurate bioelectrical impedance analysis (BIA)) and muscular, my mom started taking it at the same time for weight loss, and I was curious about the subjective effects. It basically did everything I expected: reduced the desire to eat and fomented some nausea and lethargy. I went from 86 kg to 82 kg, but the composition of my weight loss was not good. I was 13% body fat (again as measured by BIA) at the end of my n=1 trial. Effectively, I gained ~1.3 kg of fat during these 6 weeks (which is about 2.86 lbs, or 10,010 additional kcal of fat tissue). In other words, I lost 5.3 kg (~11.7 lbs) of fat free mass (which isn’t all entirely muscle but is predominately so) for a net loss of 4 kg (~8.8 lbs) when factoring in the 1.3 kg of adipose gain. On paper this looks like a positive weight loss outcome; in practice it’s anything but. Further, my compensatory hunger response on cessation was voracious.

Now, this could be down to my specific physiology and some other confounders like 1.) BIA measurement error, 2.) scale error, 3.) reduced weightlifting intensity due to lethargy, 4.) no segregation between visceral and subcutaneous fat stores, etc. I would have liked to do a DEXA scan before and after the trial to mitigate some of these concerns, but I live in Ireland, and they are hard to come by here. In practice, the ceteris might not have been precisely paribus. But, this is how these compounds are used in clinical practice, with patients adjusting lifestyle to accommodate for energy levels.

Here’s what I think is happening with GLP-1 agonists and why I think some people will run into problems with continuous use. The obvious qualifier is that for the morbidly obese or for those with uncontrolled type 2 diabetes, they could be beneficial on net as a drug of last resort.

The basic inputs to my stylized model:

• Type 1 diabetes: typically a congenital or acquired autoimmune problem where the beta cells of the pancreas don’t work properly. Blood sugar is uncontrolled is the presence of high glycemic food because insufficient insulin is produced to shunt it into tissues. Exogenous insulin is needed and a low carb diet is warranted.

• Type 2 diabetes: due to chronic overconsumption (typically of high glycemic foods) and concomitant weight gain, insulin receptors on tissues like muscle and fat are compensatorily desensitized or otherwise down-regulated so that the body is inundated with insulin and glucose. This causes vascular damage and all the known sequela associated with this condition.

• Insulin: a pancreatic peptide hormone that is broadly pro-growth/anabolic. Facilitates nutrition uptake into peripheral tissues.

• Glucagon: a pancreatic peptide hormone that is broadly catabolic. Stimulates glycogenolysis to free stored glucose in the liver and also fatty acids to maintain blood glucose and energy homeostasis.

• Leptin: a hormone mostly produced by fat tissue (but also by the gut) to inhibit hunger and limit feeding. It acts on hypothalamus to signal satiation and fullness and to regulate fat stores.

• Ghrelin: “the hunger hormone”, produced by the gut & also acts on the hypothalamus and the pituitary.

The model:

1. GLP-1 agonists limit ad libitum food consumption to aid in weight loss due to slowed digestion. Per the CICO model, less energy is available and the patient is now in a caloric deficit. Fat stores and muscle stores are mobilized to compensate. The relative balance of energy coming from muscle vs fat stores is dependent on pre-existing energy stores. For the very obese, this favors fat loss and some skeletal muscle wasting. For me, this resulted in a relative loss of muscle and even some fat gain as I’m comparatively very muscular.

2. The suppression of glucagon (remember: broadly catabolic) by the incretin effect/GLP-1 agonists limits some of the weight loss potential but is protective against some muscle atrophy. The increase in basal insulin (remember: broadly anabolic) output from the pancreas, on the converse, can contribute to increased glucose uptake in both fat and muscle cells.

3. For type 2 diabetics, this might manifest as better blood sugar control in the near term as now more insulin in present to signal to peripheral tissues to clear glucose. In the longer term, higher basal insulin will further foment insulin resistance as these receptors are increasingly de-sensitized to the hormone.

4. Leptin (remember: signals fullness) will be higher in the near term as the patient feels fuller with smaller quantities of food. This reduces the desire to eat as it signals to the hypothalamus that enough fat is accumulated. In the longer term, leptin receptors on the hypothalamus will also be desensitized and more and more leptin will be needed to signal satiety. (An interesting corollary is that according to five GWAS analyses, an SNP associated with lower circulating leptin [the AT and AA risk variants at rs9939609 of the FTO gene] is the most commonly associated gene variant with obesity. This variant has no impact on energy expenditure but rather has a direct impact on food intake, with those having the risk variant consuming on average 125-280 more kcal per day than those with the protective TT variant. I bear the AA risk variant.)

5. GLP-1 agonists will exacerbate insulin resistance in the longer term (though the risks of this might be offset by the benefits of weight loss and improved glycemic control for those most at risk) and will foment leptin resistance, especially with those having a diathesis towards leptin receptor desensitization and/or lower basal circulating levels of leptin.

6. These compounds will need to be taken in perpetuity to maintain weight loss, and as with anything, homeostatic tolerance will develop. Withdrawal effects might even induce a net weight gain.

In sum, nothing is a free lunch, and caution is warranted with these novel compounds. Our body is full of complex feedback and homeostatic mechanisms that demand a level of respect and perspicuity. I’ll write more on what are in my view better ways to maintain appetite suppression and to engender sustained fat loss in coming posts.

Let me know what you think or any contrary evidence you might have!

---