Antidepressants and diabetes

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

Attending1985

Full Member
7+ Year Member
Joined
Apr 1, 2014
Messages
676
Reaction score
656
What are everyone’s thoughts on the evidence showing increased risk of incident cases of dm2 with antidepressant use? Do you discuss it with patients?

Members don't see this ad.
 
What's the RR increase with antidepressants? I'd have to imagine it's somewhat small. Being a fat, lazy American is a risk factor that probably overshadows the antidepressant risk.
Around 1.27-2 is what I’m seeing
 
Last edited:
Members don't see this ad :)
well isn't it difficult to actually get that data considering a. depression is a risk factor for diabetes and b. diabetes (can go long time without being diagnosed) itself can render someone depressed?.
 
  • Like
Reactions: 1 user
well isn't it difficult to actually get that data considering a. depression is a risk factor for diabetes and b. diabetes (can go long time without being diagnosed) itself can render someone depressed?.
If you look at the data it’s an independent risk factor, there is some evidence in animals that antidepressants induce Beta cell dysfunction
 
  • Like
Reactions: 1 user
What's the RR increase with antidepressants? I'd have to imagine it's somewhat small. Being a fat, lazy American is a risk factor that probably overshadows the antidepressant risk.

And a fat, lazy, DEPRESSED American at that! Its already a cardiac risk factor...
 
Reading the evidence it’s pretty convincing I’m wondering why this isn’t more publicized
 
Do you have links to some of the studies you are referencing? And do you know if the increased risk for diabetes is driven by weight gain, or does it seem to be independent of that? By that same logic, would antidepressant use worsen existing diabetes?

At any rate, if the evidence is convincing for a meaningful increase in diabetes independent of weight gain, that sounds like it's worth mentioning in some populations (family history of diabetes, already have risk factors for diabetes for instance).
 
Reading the evidence it’s pretty convincing I’m wondering why this isn’t more publicized

I was curious, so I poked around some of the research. While I think there is probably something there, the studies are somewhat weak, mostly out of necessity. So, there may be something there, but I'm not convinced that the RRs that I'm seeing are free from confounds.

That being said, why not just frame the whole thing in the context of good lifestyle decisions? Minimizing negative side effects of both a patients comorbid medical/psychiatric conditions with good diet and exercise. Yes, 90%+ will still continue to make poor to disastrous lifestyle choices, but at least you're doing you're part of the informed consent.
 
  • Like
Reactions: 1 user
Do you have links to some of the studies you are referencing? And do you know if the increased risk for diabetes is driven by weight gain, or does it seem to be independent of that? By that same logic, would antidepressant use worsen existing diabetes?

At any rate, if the evidence is convincing for a meaningful increase in diabetes independent of weight gain, that sounds like it's worth mentioning in some populations (family history of diabetes, already have risk factors for diabetes for instance).
I’m on my phone so I can’t send links but I can do it later when I get home. Just google diabetes and antidepressants some good studies will result. These studies combined with animal studies showing beta cell dysfunction with SSRIs have me convinced.
 
Anyone planning on glucose monitoring for your patients in long term AD therapy based on this?
 
I’m on my phone so I can’t send links but I can do it later when I get home. Just google diabetes and antidepressants some good studies will result. These studies combined with animal studies showing beta cell dysfunction with SSRIs have me convinced.

I don't think it's reasonable at this point to rule out weight gain (as a side effect of medication) as a driver of the effect. Even the authors surmise as much. Also, it's interesting that the study with the highest effect size came out of the Diabetes Prevention Program, and only in a subset of patients who received placebo rather than metformin. When you get a chance, look up that paper and peek at those big confidence intervals in the antidepressant exposure group.

I have to wonder whether there is any old data set that contains information pertinent to this question, maybe from a randomized trial of depression treatment (eg, psychotherapy v. antidepressant). Replicating the findings in that type of design would be much more compelling.
 
  • Like
Reactions: 1 users
Members don't see this ad :)
I'd be really skeptical about their ability to draw conclusion that the antidepressant is the culprit from these studies. There are lots of potential unmeasured and unknown confounders. There's lots of research on inflammatory and hormonal differences in depressed patients that isn't exactly clear what to do with the information but could be of very pertinent relevance to this finding. If there was an identified biologic mechanism or animal data, that could help.
 
  • Like
Reactions: 1 user
Anyone planning on glucose monitoring for your patients in long term AD therapy based on this?

Whats your definition of "glucose monitoring"? Quarterly?

Because all patients in our clinic get annual HbA1c/CMP...so does that count as 'glucose monitoring'?
 
Whats your definition of "glucose monitoring"? Quarterly?

Because all patients in our clinic get annual HbA1c/CMP...so does that count as 'glucose monitoring'?
That's so interesting. In 21 years of receiving psych meds, I've never had a blood test (saliva tested for genetic response to drugs 3 separate times; duplications because the next test was allegedly better than the last; each time results never discussed).

Anyhow, CBC can be a good idea with A1C to make sure that the A1C is generally reliable. You can also prescribe a 14 day sensor to examine glycemic excursions in patients, which might be more useful than A1C data which can be a bit misleading (for example, I maintain an A1C of 5.4 with no meds after being told I'd be type 2 for life but if I don't eat a low-glycemic diet I can have a glycemic excursion to 180 from a potato for example). It's basically exactly the same as a regular Freestyle CGM sensor, except the doctor keeps the reader, so the patient doesn't see the results as they wear it. I use the consumer version, and it's quite accurate and is painless (the consumer version is applied every 10 days in the US, and I have never once felt any pain applying it): FreeStyle Libre System Providers | Personal and Professional CGM

I'd basically describe it as a Holter monitor for glucose values.
 
A problem is the specific antidepressant may be the factor and not the entire class of meds.
A similar problem is with antipsychotics. Yes as a class in general they are not good for metabolism but some of them have worse effects while Ziprasidone doesn't affect weight or metabolism and the data on this is solid.
So then is the metabolic risk with antipsychotics apply to Ziprasidone? Seems it doesn't but the APA/ADA guidelines say to do metabolic tests with all antipsychotics even Ziprasidone.

Bupropion causes weight loss, so correspondingly my hypothesis is it would lower risk of diabetes. Paroxetine raises risk of weight gain.

Add to the complexity that some people gain weight (others lose weight) when depressed and lose it (or gain) when they get appropriately treated. So then would the antidepressant, even if it's of type that causes weight gain, have more of a benefit or less vs the weight gain caused by depression?
 
Last edited:
I'd be really skeptical about their ability to draw conclusion that the antidepressant is the culprit from these studies. There are lots of potential unmeasured and unknown confounders. There's lots of research on inflammatory and hormonal differences in depressed patients that isn't exactly clear what to do with the information but could be of very pertinent relevance to this finding. If there was an identified biologic mechanism or animal data, that could help.
Fluoxetine-induced pancreatic beta cell dysfunction: New insight into the benefits of folic acid in the treatment of depression. - PubMed - NCBI

Selective Serotonin Reuptake Inhibitors (SSRIs) Inhibit Insulin Secretion and Action in Pancreatic β Cells
 
A problem is the specific antidepressant may be the factor and not the entire class of meds.
A similar problem is with antipsychotics. Yes as a class in general they are not good for metabolism but some of them have worse effects while Ziprasidone doesn't affect weight or metabolism and the data on this is solid.
So then is the metabolic risk with antipsychotics apply to Ziprasidone? Seems it doesn't but the APA/ADA guidelines say to do metabolic tests with all antipsychotics even Ziprasidone.

Bupropion causes weight loss, so correspondingly my hypothesis is it would lower risk of diabetes. Paroxetine raises risk of weight gain.

Add to the complexity that some people gain weight (others lose weight) when depressed and lose it (or gain) when they get appropriately treated. So then would the antidepressant, even if it's of type that causes weight gain, have more of a benefit or less vs the weight gain caused by depression?

Absolutely, after Paxil basically single-handedly caused the BBW for ADs in folks under 25, this type of "class=all the same" garbage is a terrible way to think about pharmacology. Would not surprise me at all if Paxil was dragging the class down here too.
 
This is the type of study where before it even started I would've told the researchers to approach it different and base it on specific meds and not the class as a whole. If they wanted to be more broad than 1 med than to do a few meds. E.g. 1 SSRI, 1 SNRI, Wellbutrin, etc.

I didn't look up all the individual authors but this reeks of people outside the field not really knowing enough about it trying to quantify something based on a flawed premise.

Poor Wellbutrin. To be lumped as a metabolic-worsening med when it does the opposite. Wellbutrin please file a complaint with the ACLU for being discriminated!
 

Both of these papers are looking at in vitro beta cells. There is just a yawning, gaping chasm between in vitro studies (one of which is solely in rat preparations!) of isolated cell lines that are not participating in the complex homeostatic interactions typical of biological systems on the one hand and actual human responses with any clinical significance on the other.

What this tells you is what probably happens to someone's pancreas if you cut it out of them and soak it in an ssri solution, not what is going to happen to your patients in clinic.
 
  • Like
Reactions: 2 users
Both of these papers are looking at in vitro beta cells. There is just a yawning, gaping chasm between in vitro studies (one of which is solely in rat preparations!) of isolated cell lines that are not participating in the complex homeostatic interactions typical of biological systems on the one hand and actual human responses with any clinical significance on the other.

What this tells you is what probably happens to someone's pancreas if you cut it out of them and soak it in an ssri solution, not what is going to happen to your patients in clinic.
I realize that but I think the combination of this with the aforementioned studies is compelling enough to consider that there is something there. Psychiatry is notoriously bad at recognizing the side effects of psychiatric medications.
 
I realize that but I think the combination of this with the aforementioned studies is compelling enough to consider that there is something there. Psychiatry is notoriously bad at recognizing the side effects of psychiatric medications.
I call it dilution of liability.

Short clinical trials. 10 years for side effects to coalesce clearly. 20 years for regulators to notice. 30 years for prescribing practices to change. (Optimistic.) People die of diverse proximate causes, hard to trace back to metabolic changes. By this point the originator drug is no longer even sold as a brand-name product or is owned by an entirely different entity than the one that submitted it to the FDA. Patients have seen many doctors over that time period. No one is ever truly responsible. It's the same with benzos. Nobody dies quickly when they're taken as prescribed (they don't even die quickly when they're used in overdose for lethal injections . . . sorry for my morbid quip), but the all-cause mortality is increased, and no one is responsible, not manufacturers, not prescribers. I think that's one of the aspects of prescription drugs, outside of psychiatry as well, that has made people rather cavalier with them. You don't need true expertise to prescribe them because like supplements they generally don't kill people even when the benefits don't outweigh the harms.
 
Are patients gaining significant weight on SSRIs? As in more than 10 pounds? Is this a common thing with SSRIs. Because I always thought it was an antipsychotic drug side effect, not an SSRI one.
 
Are patients gaining significant weight on SSRIs? As in more than 10 pounds? Is this a common thing with SSRIs. Because I always thought it was an antipsychotic drug side effect, not an SSRI one.

The problem IMHO is not the class of meds but the individual meds. As mentioned above Wellbutrin (Bupropion) causes weight loss, about 10% per the last study I read. Most SSRIs aren't associated with weight gain or if so not by much but out of all of them Paxil (Paroxetine) is the biggest offender in large part because it's strongly antihistaminic.

I have seen patients gain weight on pretty much all antidepressants minus Wellbutrin but with the SSRIs with most of them (e.g. Escitalopram, Prozac) it's hardly ever, and if so on the order of only a few pounds, plus when some people get out of depression they naturally want to eat more cause they're enjoying life again so often-times that weight gain is not really from the med but because depression (that robbed them of their appetite and passion for food) is now gone.

Tricyclics on the other hand almost always cause significant weight gain.

Antipsychotics-that too depends on the med. Back in the day pretty much all of them did it, hence the now false paradigm that they're all metabolically bad. Later on Geodon (Ziprasidone) came out and it doesn't cause weight gain at all but because of APA guidelines you still have to treat it like it's a metabolic offender. Since Geodon's been out other much more metabolically-neutral meds have come out but the entire class has been given a bad name.

Geodon, file a complaint with the ACLU for being discriminated against!
 
Antipsychotics-that too depends on the med. Back in the day pretty much all of them did it, hence the now false paradigm that they're all metabolically bad. Later on Geodon (Ziprasidone) came out and it doesn't cause weight gain at all but because of APA guidelines you still have to treat it like it's a metabolic offender. Since Geodon's been out other much more metabolically-neutral meds have come out but the entire class has been given a bad name.

Geodon, file a complaint with the ACLU for being discriminated against!
Pfizer rolled out Geodon when I was a resident. When the drug reps came by they were all armed with tape measures, BMI calculators, Geodon scales for the clinic.
Lilly immediately countered with EKG references, calipers, and LOTS of "educational programs" about QT prolongation in their Zyprexa marketing.
:rolleyes:
 
I think this is an artifact. The following large prospective 18-year study found no association when diabetes was detected by *screen* - versus the clinical diagnosis/medical records approach used in most of the studies that comprised the meta-analysis cited above by resident1985. This suggests that the relevant confounder in studies that suggest otherwise is an increased rate of physician oversight among those on antidepressants, and thus increased likelihood that nascent diabetes will actually be detected.

http://www.biologicalpsychiatryjournal.com/article/S0006-3223(11)00692-5/fulltext



Additionally, there is a fair amount of accumulated evidence that treatment with SSRIs actually improves glucose regulation in confirmed diabetics.

Effects of antidepressants on glucose metabolism and... : Current Opinion in Psychiatry


Meta-analyses are garbage in, garbage out, just like anything else.
 
Last edited:
  • Like
Reactions: 1 user
l
Inhibition of insulin secretion is *antidiabetic*. It's what you *want*. Type II diabetes is a disorder involving a vicious cycle of increasing insulin resistance and *increasing* insulin secretion.

The aforementioned study shows that fluoxetine promotes insulin resistance in peripheral tissues and also inhibits the compensatory action of increased insulin secretion. It caused apoptosis of Beta cells. This would be a pro-diabetic effect not an anti diabetic effect.
 
l


The aforementioned study shows that fluoxetine promotes insulin resistance in peripheral tissues and also inhibits the compensatory action of increased insulin secretion. It caused apoptosis of Beta cells. This would be a pro-diabetic effect not an anti diabetic effect.

Are we reading the same article? De long et al.? I see nothing about peripheral response; all figures discuss effects of fluoxetine incubation on cultured insulinoma cells. Figure 5 shows that glucose treatment stimulates insulin secretion from these cells, an effect that is blocked by fluoxetine preincubation. Again since the issue in Type 2 (at least until the very end) is insulin oversecretion, I fail to see how this promotes the T2D phenotype.

I don't see anything indicating apoptosis either. I see some increase in the production of cellular ROS, which could be an expected byproduct of the increased rates of protein synthesis. Fluoxetine had no effect on cellular survival at any concentration tested (Section 3.1, first sentence of Results).
 
Last edited:
Are we reading the same article? De long et al.? I see nothing about peripheral response; all figures discuss effects of fluoxetine incubation on cultured insulinoma cells. Figure 5 shows that glucose treatment stimulates insulin secretion from these cells, an effect that is blocked by fluoxetine preincubation. Again since the issue in Type 2 (at least until the very end) is insulin oversecretion, I fail to see how this promotes the T2D phenotype.

I don't see anything indicating apoptosis either. I see some increase in the production of cellular ROS, which could be an expected byproduct of the increased rates of protein synthesis. Fluoxetine had no effect on cellular survival at any concentration tested (Section 3.1, first sentence of Results).
I was referencing the second article but I still disagree with your reasoning. In a normal subject inhibiting insulin secretion would cause hyperglycemia. In a person with insulin resistance it would inhibit the compensatory hyperinsulinemia causing hyperglycemia. There is some thinking that an antidiabetic agent inhibiting insulin secretion would let the beta cells “rest” and prevent burnout in those with metabolic syndrome. This is not what this article is demonstrating fluoxetine does.
 
1) The second article (Isaac et al) uses supraclinical concentrations of SSRIs. De Long et al. reference a previous study showing the clinically effective range of fluoxetine to be 68 nm to 2 uM. This is the range they used, and found no cell death.

Isaac et al. used 50 uM fluoxetine to achieve increased JNK phosphorylation (no effect at 30 uM, see their Fig 3). Similarly, all their sertraline
effects are demonstrated at 10-30 uM, which is still least 5x over max of clinical range (30-200 nM).

I don't think this is meaningful clinically.

2) I'm not wedded to the SSRIs-improve-glucose-tolerance theory (I think a lot of it may be behavioral) but my point is that the finding that SSRIs inhibit glucose-stimulated insulin secretion is not prima facie evidence that they increase diabetes risk. It's an interesting finding but how it relates to human effects at clinical doses is an open question. Overall the clinical evidence that SSRIs improve diabetes control is good, and the evidence that they increase diabetes risk is shoddy.
 
Last edited:
Top