Talk:Metformin

PCOS symptoms other than infertility

The subsection on PCOS itself was removed and there is only a section on infertility (which is all about infertility from PCOS). Metformin is used to treat other symptoms of PCOS per the office of womens health ref and others. I think we should rename this section to PCOS and discuss other symptoms as well. Thoughts? Jytdog (talk) 21:29, 19 November 2018 (UTC)

Sure. Its use for infertility outside of PCOS is poor. Doc James (talk · contribs · email) 21:55, 19 November 2018 (UTC)

OK by me to have a PCOS section, although that compels other adjustments. Many obese patients with GDM or DM2 but without PCOS are treated with metformin during pregnancy, although this is not supported by the Dodd meta-analysis. How about a "metformin during pregnancy" section and a separate "metformin for enhancing female fertility section?" That would seem to be more intuitive for the reader. No strong feelings on this, though.Sbelknap (talk) 22:35, 19 November 2018 (UTC)

Lazarus Study

I note that Doc James has removed the Lazarus et al citation from the discussion of adverse effects of metformin. This study analyzed data from a ommunity-based cohort of 75 413 patients with diabetes in Geisinger Health System, with time-dependent assessment of eGFR stage from January 2004 until January 2017. Results were replicated in 67 578 new metformin users and 14 439 new sulfonylurea users from 2010 to 2015, source from a separate dataset. There are two problems here. First, Doc James did not acknowledge that he was deleting this citation and it's conclusions. Second, Doc James continues to misapply the wikipedia guidelines for medical articles by removing many high-quality primary sources. This is not what these guidelines state. Primary sources vary in quality, and are worthy of citation, particularly on issues where clinical trials and resulting meta-analyses do not adequately study adverse drug effects. Doc James (and others) are asked to be explicit in their edits when they are removing citations. Also, Doc James is asked to reread and reconsider the intent of the medical wikipedia guidelines w/r/t primary sources.Sbelknap (talk) 13:54, 19 December 2018 (UTC)

References

Per WP:MEDPRI, "If material can be supported by either primary or secondary sources – the secondary sources should be used." Rather than quote a cohort study, why not simply quote the FDA's guidance on the matter? ―Biochemistry🙴❤ 14:00, 21 December 2018 (UTC)

Better to use a secondary source on the topic. Doc James (talk · contribs · email) 07:52, 11 January 2020 (UTC)

Text

"Metformin increases circulating levels of GDF15, which reduces food intake and lowers body weight through a brain stem-restricted receptor.""

Would belong under mechanism of action if anywhere. Doc James (talk · contribs · email) 07:52, 11 January 2020 (UTC)

Note authors think the relevance is to obesity: [1] Ratel (talk) 10:58, 11 January 2020 (UTC)

References

Liver and kidneys

"Lactic acidosis rare but potentially fatal. Increased risk of lactic acidosis in patients with renal impairment and advanced age"

"Generally avoid use in patients with clinical or laboratory evidence of hepatic disease."

"Do not use in patients with severe renal disease or dysfunction"

https://www.drugs.com/monograph/metformin-hydrochloride.html

Doc James (talk · contribs · email) 00:03, 22 April 2020 (UTC)

The text in the lead is not accurate, because there is strong evidence that it is safe to use metformin in patients with renal failure. Please review the text and citation in the main body of the article, which present high-quality evidence that metformin is OK to use in patients with renal failure. The lead, as it stands is wrong. Also, see Sbelknap (talk) 17:08, 22 April 2020 (UTC)

References

The AHFS citation does not advise against use of metformin in patients with renal failure or liver disease. This topic is covered in more detail in the main body of the article. I've corrected the lead to reflect the cited information.Sbelknap (talk) 20:03, 23 April 2020 (UTC)

With respect to the above those are exact quotes. What do you think "Do not use in patients with severe renal disease or dysfunction" means? I guess the kidney issue is controversial. Doc James (talk · contribs · email) 00:08, 24 April 2020 (UTC)

The current statement in the lead states: "High blood lactic acid level is a concern if the medication is used in overly large doses or prescribed inappropriately, such as those with kidney problems." This is overly broad, as most of the population aged > 60 has some sort of kidney problem. The statement in the lead does not reflect the statement in the cited reference, "Do not use in patients with severe renal disease or dysfunction." The statement in the lead does not mention severity and the term "kidney problems" is so vague as to be almost meaningless. It is specifically diminished renal clearance that is a problem, and not other sorts of kidney disease. It is not controversial to use metformin in patients with eGFR>45 ml/min. Please review the FDA-approved FPI. This current statement in the lead is incorrect and has potential to do harm. It is also inconsistent with more detailed information in the main article. Sbelknap (talk) 04:23, 24 April 2020 (UTC)

Pharmacokinetics

The pharmacokinetics section seems not to be up to date with the latest literature. Rather than edit this myself, I would prefer if an author of this article would look over the newer literature and make changes as necessary. I would, in particular, suggest that the metformin tail is asymptotically a power function. There are some drugs (e.g., amiodarone) for which this is known to be important, and the new information is that it is important for metformin as well. For example, with a terminal power function, there is no terminal half-life per sey. This explains why the literature includes wildly divergent half-life values for metformin; for a power function the half-life depends on when the last sample was collected, and the longer one samples, the longer the apparent terminal half-life seems to be. This can be explained as follows: 1) Only monoexponentials have a constant half-life. 2) When a monoexponential is plotted tangent to a negative power function of time at any particular time, its half-life is that of the monoexponential at that particular time. However, 3) negative power functions of time have more shallow curvature than monoexponentials such that the later in time one constructs the tangent, the longer the monoexponential's half-life appears to be. Those are the most basic concepts for metformin pharmacokinetics. For a much more elaborate treatment please see.

For an independent evaluation of that work please see, , and for an executive summary, please read. CarlWesolowski (talk) 21:26, 4 July 2020 (UTC)

Not done, this is too soon. The article hasn't been independently reviewed yet the second source appears to be a collaborator. If you can provide independent reviews in a scientific journal of this work we can add it. We don't cite primary sources so it won't be your article that is cited, therefore if you can find an independent source to add the detail you want there won't be a conflict in you directly editing this article. It doesn't really work like normal science writing so it is necessarily behind in the times sorry I couldn't be of more help to you! PainProf (talk) 23:58, 4 July 2020 (UTC)

First of all, Prof. Tucker wrote the review when he was not a collaborator and was so impressed with the work that he contacted the first author to write an executive summary. As for being too soon, the results are very solid and if you wish to dispute them a lack of argument is not in itself a discussion point. Put in the obverse, the evidence in the paper for not using sums of exponential terms and not assuming exponential conditions is solid. Furthermore, 100 years of repetition of bad models without testing of assumptions is too long, and it is none too soon to question them. Seen from that POV, what is written in the PK section herein is very misleading. I strongly suggest that you read the article in the first reference, which includes such niceties as reviewing the scattered values that have been provided for metformin half-life, their lack of coherence and the alternative explanation for them. There will never be any scientific progress if the evidence is not even considered, and familiarity with models that are not explanatory is not an excuse for ignoring explanatory models that are unfamiliar. CarlWesolowski (talk) 00:26, 5 July 2020 (UTC)

Hi sorry the main point is fact it hasn't been cited. It would be original research to cite it at the moment. It needs to be reviewed as it is a primary source. I must confess to not being up to date on pharmacokinetics. I am wary to include this because; the explanation is unclear and dense, it isn't accepted or debated by the field at large. Either way it won't be included as it is a primary source. Probably the best is to wait a few hears once the field have had time to read it and debate it. Wikipedia isn't the place for the latest and greatest research. It gets updated mainly when there are significant new reviews or synthesis on the topic. If you can point to one I'd be delighted to help you add it. PainProf (talk) 00:50, 5 July 2020 (UTC)

Could you show a testable hypothesis that is best explained by your model?

I was not suggesting, nor did I suggest that the articles I cited should be included in the text. All I did was to point to articles that show how poorly the text is written. In particular the half-life data cited in the article is ridiculous and does not represent any consensus on that topic, and this is reviewed in the articles I cited on this talk page. I do not care what you do but if you disagree with me you should follow protocol on that. Harassment for things for which you have no particular training is not proper. If you disagree with me for example, seeking out variable volume pharmacokinetics, which admittedly could have a dozen more citations and going to war over it is not appropriate behaviour. Appropriate behaviour is as follows:

Follow the normal protocol

When you find a passage in an article that is biased, inaccurate, or unsourced the best practice is to improve it if you can rather than deleting salvageable text. For example, if an article appears biased, add balancing material or make the wording more neutral. Include citations for any material you add. If you do not know how to fix a problem, ask for help on the talk page.

Any more harassment will be reported. CarlWesolowski (talk) 02:52, 5 July 2020 (UTC)

Hi the only reason this was drawn to my attention is that I noticed it already had a Wikipedia link on the publication when I read it. I found that surprising as there are no literature citations. It says that you included the link which may be wp:citespam but also suggests you may not have an entirely neutral point of view. I notice that this theory isn't the accepted model for pharmacology, you claim to have invented it but there are no independent citations as yet. The prose is needlessly dense, honestly I should not find it difficult to understand. None of this is harrassment rather it is due diligence when potential citation spam is found. Please just include a few independent references and I won't propose the article for deletion. PainProf (talk) 03:19, 5 July 2020 (UTC)

I did not invent variable volume PK, I added variable half-life to it. That is not something that is up for debate, the only thing having constant half-life is the monoexponential; all the others have variable half-life. That is elementary mathematics, and the fact that it has been blithely ignored we can speculate on all we want, but it is here to stay. As for citations, there are 4, two mine, two not. There need to be more, and when I have time, if no one else does, I will include them. You really need to read the papers before you jump to conclusions. CarlWesolowski (talk) 03:52, 5 July 2020 (UTC)

Several points: Pure mathematics as a matter of formal proof does not accrue validity, it is axiomatic, is not hypothetical and is unchanged by citation, unless that citation is a counter proof. Second, either one can read the proof or not. Its validity stands irrespective of the readers ability to understand it. Third, to copy a proof verbatim without citation would be plagiarism. Fourth, proofs are not "invented" they are shown. Fifth, the article included review and was itself was both reviewed and cited. The issue at hand is as follows:

"Two time-sample exponential $t_{1/2}$,

$\xi =-\ln(2){\dfrac {t_{i+1}-t_{i}}{\ln(C_{i+1})-\ln(C_{i})}}\qquad (5)$

Instantaneous half-life (definition),

$t_{1/2};f(t)=-\ln(2){\dfrac {f(t)}{f'(t)}}\qquad \qquad (6)$

...Equation (5) is common knowledge for any two time-samples, $\{t_{i},C_{i}\}$ and $\{t_{i+1},C_{i+1}\}$...However, Eq. (5) only applies for monoexponential conditions or in the limit if we substitute $f(t)\approx C_{i}$, and $t_{i+i}=t_{i}+\Delta t$ and let $\Delta t\to 0^{+}$, where $f(t)$ is ${\textit {any}}$ continuously differentiable function. That yields Eq. (6), which is of classical mechanical type, and applies equally well to the evolution of half-life of satellite orbital decay as to variable half-life of concentration of any drug. Half-lives can be both positive and negative. For an $f(t)$ that increases in time, the $t_{1/2}$ is negative. For example, this occurs when concentration is increasing in time in a peripheral vein shortly after intravenous injection. It easier to understand a negative half-life of an increasing function of time when some maximum total value of a measurement is being approached in time by subtracting the measured values from that maximum and viewing the result as a positive half-life of a decreasing function in time, as is done, for example, using total dosage administered for a urinary drug recovery $t_{1/2}$ calculation. However, growth of any function in time, including drug mass accumulating in urine, actually has a negative half-life."

The only thing that is "hypothetical" is that sums of exponential terms can form good models of metformin. That hypothesis was tested and found wanting. I will continue this later. CarlWesolowski (talk) 19:30, 6 July 2020 (UTC)

Any model in biology requires experimental proof and a testable hypothesis, I'm unclear as to why a mathematical proof is being discussed here. This may all be true of mathematical models but you have entered the world of biology, where a model requires significant experimental evidence before being accepted. As a note, I actually think the concentration of a drug after intravenous injection would rather rapidly be the same in the blood probably negligible with respect to total half life (in the blood at least since it rapidly mixes). I assumed your situation was more where there is a difference in the cell or tissue such as the physical properties within it that makes it accumalate differently or enter at a slower rate or a SC injection where there is more limited distribution. For instance the non-solid phase components of a cell will have a different rate of accumulation due to density. Either way, the model hasn't gained widespread use or acceptance at this stage. That would be shown by synthesis reviews of many papers discussing approaches to pharmacokinetics using the method. As an aside none of this is relevant to this metformin article, it wouldn't help the reader to include a mathematical proof PainProf (talk) 20:43, 6 July 2020 (UTC)

You are fighting me rather than asking, there is a point here: I wish to weigh in on the points of discussion here that I consider dangerous to patient health. From the Comparison... paper "bolus intravenous metformin plasma half-life was 1.5 h as last sampled at 8 h \cite{Sirtori1978}, 4.5 h as last sampled at 12 h \cite{tucker1981metformin}, and 20.4 h, as last sampled at 72 h (dogs) \cite{johnston2017pharmacokinetics}, such that there is good evidence that metformin plasma half-lives increase with elapsing time." What this let us say "potentially" if I am "correct" leads to is overdosing patients with reduced renal function. I consider it my duty to warn people rather than let potentially dangerous information go unquestioned. Now, are you suggesting that I should keep quite because it takes a long time to explain things, or, do you want me to continue? CarlWesolowski (talk) 21:05, 6 July 2020 (UTC)

Now, to continue, if I did not show that half-life is in general variable, the alternative hypothesis that metformin half-life is poorly characterized would be given some credence: "rather than just accept that metformin plasma concentration half-lives are poorly characterised \cite{Graham2011}" This is what physicists would call hand waving. Nowhere in the metformin article here does anyone even note that there is any difficulty determining a metformin half-life. Continuing the quote "we note that Xie et al. \cite{xie2015metformin} as well as Sambol et al. \cite{sambol1996pharmacokinetics}, the latter for oral multidosing, have commented that metformin half-life increased with elapsing time." It is because of this increase in half-life with elapsing time that metformin build-up could proceed to dangerous drug levels in renal failure. Finally, I don't care if anyone ever duplicates the model, or it comes into general usage. I am not wearing my ego on my sleeve, this is not about me. Some of my colleagues cannot read a model like this:

${\text{GPC}}(a,b,\alpha ,\beta ;t)=\theta (t-\beta ){\frac {\alpha b^{a}\beta ^{\alpha }}{\Gamma (a)}}\sum _{n=0}^{\infty }{\frac {1}{n!}}(-1)^{n}b^{n}t^{a-\alpha +n-1}B_{1-{\frac {\beta }{t}}}(a+n,-\alpha )$

Such that requiring them to learn what a unit step function, $\theta (x)$, is, or an incomplete Beta function, $B_{x}(g,h)$, is, could take a long time. Concerning half-life, if I rewrote half-life for PK, I dare say it would meet with much more resistance than if I rewrote it for physics on this same website. I can go on to include more evidence, but, at this point, consider yourself warned, be wary of dosing metformin in renal failure, build up to toxic levels could take a very long time and have an apparently erratic appearance, so much so that cause and effect may be hard to piece together.