metformin seems okay in patients with severe kidney disease

 

a recent Scottish study found that patients with diabetes who develop stage 4 chronic kidney disease (CKD) and continue metformin had improved survival (see dm metformin with lower GFR AmJKidnDis2025 in dropbox, or doi: 10.1053/ j.ajkd.2024.08.012)

Details:

-- 371,742 Scottish residents with type 2 diabetes were evaluated from the Scottish Diabetes Research Network-National Diabetes Study, from 2010-2019

-- 4,287 prevalent metformin users were identified who had incident stage 4 CKD (ie, eGFR <30); 1,713 of them (40.1%) discontinued the metformin    

-- patients who stopped the metformin did so within 6 months of reaching CKD stage 4; those who continued metformin did so for at least 6 months

    --55% of those who continued the metformin did stop it in the follow-up period (about 60 patients/year from years 7-12, about 25/year from year 13 to end of study): 44% continued throughout the follow-up

-- median age 77, 51% women

-- diabetes duration 15 years, current smoker 10%/ever smoker 44%, BP 136/71, A1c 7%

-- median eGFR was 27, ischemic heart disease in 38%, cerebrovascular disease in 11%

-- diabetes meds: insulin 21%, sulfonylureas 44%, DPP-4 18%, GLP-1 4%, SGLT2 0.4%

-- other meds: ACE/ARB 83%, thiazide 23%, a-blocker 2%, anticoagulant 72%, lipid-lowering med 88%, calcium channel blocker 49%

-- the researchers used a statistical approach called target trial emulation, which attempts to mimic randomized trials in observational studies. What they do to overcome flaws in observational trials is "by first designing a hypothetical randomized trial, the target trial, that would answer the question of interest, then emulating this target trial using the available observational data and appropriate methodology", per https://www.bmj.com/content/378/bmj-2022-071108

-- main outcomes: comparing 3-year survival, MACE (major adverse cardiovascular events), and all-cause mortality in those individuals who continued vs stopped the metformin

-- patients followed a median of 2.5 years

Results:

-- during the follow-up period, 1,702 people died, most commonly from cardiovascular causes (34%), cancer (17%), and respiratory diseases (10%)

-- comparing those who continued vs stopped metformin with stage 4 CKD:

    -- 3-year survival (1,702 deaths): 

        -- continuing metformin: 63.7% ( 60.9%-66.6%)

        -- stopping metformin: 70.5% (68.0%-73.0%)

            -- 26% higher survival in those who continued the metformin, HR 1.26 (1.10-1.44)

    -- MACE (915 events):

            -- no statistically significant difference, with HR 1.05 (0.88-1.26)

the purple line represents mortality over time in those who continued the metformin, the red line for those who discontinued the metformin. note that divergence began around 12 months and continued to increase over time

-- using the target trial emulation in the sensitivity analysis did not show much difference from the above analysis:

     -- all-cause mortality: 34% higher in those stopping the metformin, HR 1.34 (1.08-1.67)

     -- MACE: HR 1.04 (0.81-1.33), still not statistically significant

-- subgroup analysis: discontinuing metformin was associated with 51% higher risk of death from respiratory diseases, HR 1.51 (1.06-2.12), but no difference in the risk of cancer

Commentary:

-- diabetes is the major cause of kidney failure globally, with ¼ of patients having an eGFR<60

-- the historical concern is that metformin is renally excreted and that those with CKD would therefore predisposed to develop lactic acidosis

    -- this concern here is well-founded: the chemically similar biguanide phenformin was used in the US (the prescient Europeans chose metformin), and phenformin led to highly deadly lactic acidosis (in the 1960s and 1970s, the ICUs were filled with patients who were on phenformin and had a 90+% mortality despite maximal ICU care for the time)

    -- but for metformin, huge studies done a few decades ago assessing lactic acidosis only found small increases in lactic acid levels, and there was no increased risk of life-threatening lactic acidosis beyond what was expected (the community baseline). 

    -- a 2018 study of metformin in patients with diabetes and CKD stage 3a, 3b and 4 found that there were no cases of lactic acidosis (defined as >5 mmol/L) in those patients taking metformin with dose reductions (for stage 3a CKD 1500 mg in am and 1000mg in pm; for stage 3b CKD 1000mg in am; and for stage 4 CKD 500mg in am), and the metformin levels in the blood were stable and never exceeded the generally accepted safe level of under 5.0 mg/L: see dm metformin levels ok in CKD DiabCare2018 in dropbox, or doi.org/10.2337/dc17-2231)

-- the Scottish study above found a clear association between stopping metformin in those with GFR <30 and increased risk of death. this did not appear to be due to cardiovascular causes (MACE), though this particular outcome was different from another recent study finding decreased MACE (see below) raising the possibilities: ?did the patients in this current study have more advanced cardiovascular disease and were more likely to have a MACE even with continuing the metformin??; were there substantial differences in diabetes meds (in this current study, a higher percent were on GLP1 and SGLT2 which are cardioprotective)??. ??were the dosages of the different diabetes meds different (the use of insulin/sulfonylureas are associated with increased MACE in most studies, see below)??

    -- notably, the eGFR remained quite stable in those continuing metformin (eGFR was 27.0 at baseline and 26.1 at 6 years out) vs those who stopped the metformin (eGFR was 26.6 at baseline and 16.4 at 6 years out): ie, continued metformin significantly decreased the progression of CKD!!!

 Some other studies on this issue:

   -- a systematic review of 65 studies of patients on metformin who had CKD found that metformin did not increase the incidence of lactic acidosis even in those on metformin above the FDA guidelines of creatinine >1.5 in men and >1.4 in women:   http://gmodestmedblogs.blogspot.com/2015/01/metformin-in-renal-failure.html 

    -- a retrospective Taiwanese study found that there was no increased mortality by continuing to use metformin in patients with diabetes who had stage 5 CKD (these patients had a creatinine of at least 6 mg/dL, with a mean eGFR of 10 in men and 7 in women), as long as the metformin dose was decreased to 500 mg/dL. also, no increase in lactic acid levels in these patients with very severe CKD: http://gmodestmedblogs.blogspot.com/2015/06/metformin-ckd-and-death.html

    -- a new article in Hong Kong of 33,586 metformin users with new onset eGFR <30 (stage 4 CKD, with mean eGFR of 27, A1c 7.4%)  found that those patients who discontinued metformin and were followed 3.8 years had a 40% increased risk of MACE, 52% increase end-stage kidney disease, and 22% increased risk of death. they did have the expected higher A1c levels (increased from 7.7% to 8.2% in those who discontinued the metformin). there was also no increased risk of lactic acidosis despite the stage 4 CKD.  (see dm metformin cont with CKD4 Lancet2024 in dropbox, or doi.org/10.1016/j.eclinm.2024. 102568). the median dose of metformin in this study was 1000mg/d

            -- unclear why there was a large increase in MACE in those who discontinued the metformin in this Hong Kong study as opposed to what was found in the Scottish study above. ??more use of GLP-1s and SGLT2's in the Scottish study, which are cardioprotective?? There was high use of insulin and sulfonylureas in the Scottish study, though this was also found in this Hong Kong study as well. But we do not know medication doses or if the meds changed during the study. And many studies have found increased cardiovascular events in those on insulin or sulfonylureas: http://gmodestmedblogs.blogspot.com/2022/10/diabetes-less-of-cardiovascular-risk.html 

            -- of note in this Hong Kong study as well as the Scottish study above, they also found increased pneumonia-related mortality in those who discontinued their metformin

            -- also another study on metformin in patients with Covid found increased pneumonia deaths in those stopping the metformin

                -- unclear why stopping the metformin led to increased pneumonia deaths in these studies. ??may be attributable to metformin decreasing chronic inflammation

-- in the US, metformin has been contraindicated in patients with estimated glomerular filtration rates (eGFRs) <30 to prevent lactic acidosis, though 2 studies found about 10% of people in the US who had eGFR<30 still continued on metformin and with no bump in lactic acidosis noted

        -- for the formal 2016 FDA guidelines on use of metformin, including not using it when eGFR <30: see http://gmodestmedblogs.blogspot.com/2016/04/fda-changes-metformin-guidelines.html

    -- because the above studies found that metformin taken below this eGFR threshold of <30 is safe and beneficial, the current researchers accessed the huge Scottish national database of patients with diabetes to provide further evidence about the role of continuing metformin in the setting of more advanced CKD of stage 4 or 5 (eGFR <30), confirming the considerable benefit of metformin

-- metformin has been the backbone of diabetes medical management, for several reasons:

    -- it inhibits hepatic gluconeogenesis and increases post-prandial insulin-mediated glucose utilization by major tissues, especially muscle and liver

    -- it lowers serum free fatty acids, which helps decrease gluconeogenesis

    -- it lowers serum lipid concentrations (as well as improving endothelial function, decreasing inflammation and oxidative stress),  which may be part of the reason that metformin seems to be associated with decreased cardiovascular events: https://pmc.ncbi.nlm.nih.gov/articles/PMC9549498/ , though the studies are not totally consistent on this

    -- and, notably, it seems to help improve the gut microbiome:

            -- a part of metformin's mechanism of action may be related to creating a healthier gut microbiome. A study of diabetic mice found that those on metformin improved their glycemic profile (as expected), but there was a significant increase in the gut bacterium Akkermansia. In a parallel experiment, just increasing this bacterium in the gut (in the absence of metformin) also enhanced glucose tolerance and decreased adipose tissue inflammation, suggesting that an additional mechanism of action for metformin may be through its direct effect on the microbiome (see http://gmodestmedblogs.blogspot.com/2024/07/microbiome-in-cardiovascular-disease.html ). 

            -- a recent review revealed the complexity of metformin's effects on the microbiome: https://pmc.ncbi.nlm.nih.gov/articles/PMC10518565

            -- as an aside, there are several effects of the gut microbiome that seem to play a role in cardiovascular health: http://gmodestmedblogs.blogspot.com/2024/07/microbiome-in-cardiovascular-disease.html

    -- bottom line: metformin is a great drug for patients with diabetes for a variety of reasons

Limitations:

-- this study (and others sited) are through retrospective data-mining and not randomized controlled trials, thereby leading to associations and not clear causality. for the latter, a well-conducted RCT is necessary

    -- for example, we do not know why some individuals with CKD continued metformin and some did not. were there important differences between these groups that could explain why some stopped vs those who continued the metformin?? could those differences explain the different outcomes??

    -- this study did use the target trial framework to try to mimic an RCT in observational studies by equalizing the expected covariates (but though this is better than just an observational study, there may well be unmeasured confounders not assessed). it was reassuring that doing this did not change the actual adverse outcomes by much

-- there was some "contamination" in the study: many people (55% as noted above, though at different times of follow-up) discontinued the metformin in the "continuation" group and 4% in the discontinuation group restarted their metformin during the followup. these changes may have blunted the results they found (ie, the effect of actually continuing metformin might have been even more)

-- this study was in a single population in the UK (Scottland) which has a universal health care system and the results may not be generalizable to other areas

-- there was no information about medication adherence in the study, only the medications dispensed

-- it is not clear exactly what was going on in the group who discontinued the metformin. their A1c did increase, as noted above. Were there changes in other diabetes meds to compensate for this increase? they did allude to that there may be changes in these meds in the study, but no data was given (and though they referenced the supplementary tables to review this A1c data and presumably changes in the meds, these supplementary tables actually did NOT exist (and, by the way, i have found this to be true in several studies where data was supposed to be available in the supplementary tables or graphs, even in the best peer-reviewed journals). if there were changes in the meds in those who stopped the metformin, were they to more or less atherogenic meds?? this could clearly alter their results for MACE

-- another general concern I have, as per many prior blogs, is that creatinine is not the best measure of actual GFR or as a predictor of bad outcomes; cystatin c seems to be much better predictor (https://gmodestmedblogs.blogspot.com/2023/12/cystatin-c-better-predictor-of-bad.html

so, another in the slew of studies suggesting that continuing the metformin may continue to be beneficial in those with progressive renal dysfunction, and seems to actually improve the eGFR.  this is hugely important given that diabetes is the major cause of CKD, CKD is remarkably common, CKD is a bad disease for so many reasons, and continuing the metformin seems to decrease total mortality as well as major cardiovascular events (at least in some studies) and renal deterioration.

    -- of course, it would be very important to have a randomized controlled trial to really determine the efficacy of continuing metformin to see if there is a different lower limit (or none??) of eGFR that might be the appropriate cutpoint (vs the current eGFR of 30)

-- given the many known (and perhaps others unknown) benefits of metformin throughout the body, I am personally very leery of stopping metformin in patients, even in those who are doing really well on GLP-1 or SGLT2 therapies, or using one of these newer agents as my initial therapy (there was an article suggesting using SGLT2s as first-line therapy: https://gmodestmedblogs.blogspot.com/2022/07/diabetes-sglt-2-inhibitors-as-first.html , though I thought through a pretty flawed argument to support this)

        -- i have certainly succumbed to medical information that have had lots of observational data supporting a clinical conclusion in the past (these articles presented very convincing mechanisms of action to support the potential clinical benefit), only to find that in well-conducted RCTs this was not really true (eg, assessing and treating patients with high homocysteine levels, prescribing hormone therapy to postmenopausal women, accepting the nutrition guidelines that to lower lipids and decrease cardiovascular disease it was most important to lower saturated fats, typically through consuming more carbs...)

-- so, we really do need a randomized controlled trial to be sure. And, if we want to use metformin at more severe stages of CKD, we should do so gingerly: reducing the metformin dose, following closely the renal function, aggressively addressing other cardiovascular risk factors, and following heart failure both clinically, as well as by BNP levels (there was an impressive study finding that checking BNP levels every 3 months or so in patients with CKD but no known heart failure did lead to interventions to decrease both cardiovascular and renal future outcomes: http://gmodestmedblogs.blogspot.com/2023/12/routine-bnp-assessment-helpful-for.html

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them