SGLT-2: another study finding more amputations; are they being oversold??

 a recent US veterans study compared SGLT-2 inhibitors with DPP-4 inhibitors in patients with diabetes and found increased peripheral artery disease (PAD) surgeries in those on the SGLT-2 inhibitors (see dm SGLT2 inc PAD DiabCare2025 in dropbox, or doi.org/10.2337/dc24-1546

 

Details:

-- this was a retrospective cohort study of U.S. veterans with diabetes, aged at least 18yo who received care from the Veterans Health Administration from 1 October 2000 to 31 December 2021

-- data were linked to Medicare, Medicaid, and the National Death Index.

-- included patients: new use of SGLT-2 or DPP-4 medications as an add-on to metformin, sulfonylurea, or insulin treatment alone or in combination

    -- these patients were evaluated for an association of these meds with a PAD surgical procedure for peripheral revascularization and amputation.

-- the median age was 69 years, 95% male, HbA1c at baseline 8.4%, and the median diabetes duration was 10.1 years.

-- 74% white/21% Black, BMI 32

-- comorbidities: ever-smoker 50%, ever alcohol "misuse" 40%, MI 3%, PAD 8%, amputation 1%, COPD 18%, heart failure 13%, malignancy 13%, obstructive coronary disease 33%, serious mental illness 31%

-- blood pressure 134/76, eGFR 76, LDL 80, increased microalbuminuria 8%

-- diabetes co-therapies: insulin 13%, metformin 20%, sulfonylureas 9%, metformin plus insulin 25%, metformin plus sulfonylureas 32%, sulfonylureas plus insulin 4%

-- other meds: ACE 50%, ARB 25%, b-blocker 48%, CCB 32%, non-loop diuretic 30%, loop diuretic 14%, statin 81%, aspirin 23%, other platelet inhibitors 14%, anticoagulant 11%

-- propensity score weighting led to cohorts of 76,072 on SGLT2 vs. 75,833 on DPP4

-- SGLT2 meds: empagliflozin 94%, canagliflozin 4%, dapagliflozin 2%

-- DPP4 meds: saxagliptin 45%, alogliptin 34%, sitagliptin 15%, linagliptin 6%

-- primary outcome: time to first PAD surgical event, a composite of amputation, stent, or bypass surgery

-- secondary outcome: time to each individual PAD surgery

 

-- median follow-up 0.7 years

 

Results:

--Overall number of events (primary outcome):

    -- SGLT2: 874 PAD events; event rate: 11.2 (10.5–11.9) per 1,000 person-yrs

    -- DPP4: 780 PAD events: event rate: 10.0 (9.4–10.6) per 1,000 person-yrs

        -- 18% increase with SGLT2's, adjusted hazard ratio 1.18 (1.08–1.29)

 

--sensitivity analyses:

    -- PAD event rates over the next 360 days after the use of SGLT2s ended: 16% higher in that group, aHR 1.16 (1.06-1.26), consistent with the main outcome above

    -- no significant interaction between medication exposure and use of insulin at baseline (ie, no difference on subgroup analysis in those on vs not on insulin)

 

-- individual PAD outcomes (secondary outcome)

    --revascularization stent and bypass: 

        -- SGLT2: 508  PAD events; event rate: 6.5 (6.0-7.0) per 1,000 person-yrs

        -- DPP4: 438 PAD events: event rate: 5.6 (5.2–6.0) per 1,000 person-yrs

            -- 25% increase with SGLT2's, aHR 1.25 (1.11-1.41)

    -- amputations:

        -- SGLT2: 499 PAD events; event rate: 6.4 (5.9-6.9) per 1,000 person-yrs

        -- DPP4: 451 PAD events: event rate: 5.7 (5.3–6.2) per 1,000 person-yrs

            -- 15% increase with SGLT2's, aHR 1.15 (1.02-1.29)

 

-- The cumulative probability of a PAD event at 4 years was 4.0% for SGLT2’s vs 2.8% for DPP4’s

-- It was noted in the study that the death rate associated with DPP4 inhibitors exceeded that of the SGLT2  inhibitors (as would  be expected given the repeatedly evident cardiovascular benefit from SGLT2's)

 

Conclusions:

-- as we know only too well, diabetes is remarkably prevalent in the US (37 million cases, >90% type 2) and increasing over time, along with the attendant morbidities/mortality (renal, cardiovascular, ophthalmic, neurologic, vascular/need for amputations…)

 

-- The results of this study suggest that there was an increased risk of the composite PAD surgical outcomes in those on SGLT2 inhibitors, which was evident in both those patients receiving amputations and revascularization procedures

-- these results were consistent when evaluating outcomes at 90 and 360 days

    -- these results were also evident even at four years after the study, all of the suggesting that the long-term associations found with PAD after even limited exposure to SGLT2's is a long-acting one, suggesting that early exposure to SGLT2’s lead to vascular changes may not be reversible 

 

-- it was notable that overall mortality was significantly increased with the DPP4 versusSGLT2 inhibitors, confirming that DPP4 inhibitors are not a great diabetes medication (see below for more details)

    -- the concern with DPP4 inhibitors, as noted below, is that they inhibit a wide range of many different bioactive peptides as well as GLP1 (more of a shotgun approach versus a targeted one)

    -- and, SGLT2's do have a significant effect in decreasing heart failure hospitalizations as well as mortality

 

A recent German nationwide study found increased major amputations in patients with diabetes and taking SGLT-2 inhibitors (see dm SGLT2 inc amputation EurJVEndovascSurg2021 in dropbox, or doi.org/10.2337/dc24-1546):

-- >100K patients’ records who had type 2 diabetes were accessed from health claims from BARMER, a large and respected public health insurance provider in Germany.

-- prior diabetes treatment 5 years before: 64% on insulin, 70% metformin, 18% sulfonylureas, 52% DPP-4 inhibitors

-- high rates of comorbidities:  53% coronary artery disease, 43% heart failure, 37% dysrhythmias, 41% chronic kidney disease

-- 44,284 patients initiated an SGLT2 (empagliflozin in 38.4%, dapagliflozin in 60.3%) of whom 13.6% had PAD, and 56,878 initiated a GLP-1 receptor agonist of whom 16.3% had PAD.

-- results, with GLP-1s being the active comparator, after controlling for 35 baseline variables:

      -- heart failure hospitalization: 15% lower with SGLT2 vs GLP-1, HR 0.85 (0.73-0.99) in patients with PAD

      -- lower extremity amputation: 79% higher, HR 1.79 (1.04-2.92) in patients without baseline PAD

 

CANVAS study

-- the CANVAS study of canagliflozin studies (2 trials of 10K participants with type 2 diabetes and high cardiovascular risk) found that this SGLT-2 inhibitor was associated with a doubling  of risk of amputation (6.3 vs 3.4 participants per 1000 person-yrs, HR 1.97 (1.41-2.75)  (https://www.nejm.org/doi/full/10.1056/NEJMoa1611925 and https://gmodestmedblogs.blogspot.com/2017/07/another-study-assessed-role-of-sodium.html )

-- though other studies have not found an increase (eg the CREDENCE and DECLARE-TIMI 58 trials). the reason for this discordance is perhaps because of the well-recognized concomitant risk factors of PAD itself for lower extremity amputation

    -- 20% of those in CANVAS had documented PAD with 72% having a documented history of atherosclerotic heart disease (ie, a quite sick group and likely many more individuals had undetected PAD), and PAD was not a noted comorbidity in many of the prior studies (CREDENCE did not mention baseline PAD rate, DECLARE-TIMI 58 had 6% with PAD; hence, neither of these large studies had nearly the same level of amputation risk as the current study)

    -- this current VA study was explicit in enrolling patients with PAD, perhaps leading to their finding of the association between SGLT2’s and amputations

    -- ie, the issue here regarding the different PAD outcomes associated with SGLT2 use may well be that there were different predispositions to developing PAD (eg, different numbers of patients with more advanced PAD, different uses of the apparently pro-atherogenic meds of insulin and sulfonylureas vs anti-atherosclerotic meds of GLP-1s, and unknown baseline levels of PAD or even having ankle-brachial indices), and patients having no or mild PAD would require a much longer study to see clinical PAD occuring and needing intervention for PAD since those with prevalent clinical PAD are quite likely to need an intervention earlier than those with no or undetected clinical PAD

 

-- the mechanisms by which SGLT2's may increase PAD are unclear. Perhaps there are long-term effects on atherosclerosis and decreased perfusion; others have hypothesized that hemodynamic instability occurs in the early stages of SGLT2 use, perhaps leading to intravascular damage 

-- an effect has been found in some studies about differences in amputations of by different SGLT2's, particularly singling out canagliflozin.

    -- this VA study predominately used empagliflozin (94% of users, accounting for 95% of the PAD events), versus canagliflozin (4% of users accounting for 4% of the events)

    -- although canagliflozin has been singled out as perhaps the worst of the SGLT2's in terms of amputations, a large-scale study found that there was really no difference between them (see dm SGLT‐2 inhibitors all assoc with amputation DiabObesMetab2018 in dropbox, or DOI: 10.1111/dom.13255):

       -- of over 8 million reports from June 2013 to 2017 through the WHO database, there were 79 reports of lower limb amputations associated with SGLT2's, with an overall increase of amputations with SGLT2's being 5.55  (4.23-7.29), specifically with:

          -- canagliflozin: proportional reporting ratio (PRR) 7.09 (5.25- 9.57)

          -- empagliflozin: PRR 4.96 (2.89-8.50)

          -- dapagliflozin: PRR 2.62 (1.33-5.14)

                -- There were overlapping confidence intervals for the SGLT2's suggesting no clear difference between these three SGLT2 medications in amputations

 

my issues concerning SGLT2's:

-- amputations are uncommon with SGLT2's, but are clearly a bad outcome for those so afflicted

-- prior studies on SGLT2's have found similar results regarding increased amputations (http://gmodestmedblogs.blogspot.com/2018/11/sglt-2-inhibitors-twice-risk-of.html)

-- SGLTs2's do have a slew of complications which can be serious: 

    -- there is pretty clear evidence for increasing the incidence of Fournier's gangrene (http://gmodestmedblogs.blogspot.com/2019/05 /sglt-2-inhibitors-and-fourniers-gangrene.html)

    -- there is a pretty significant risk of diabetic ketoacidosis in patients with diabetes and even lower than usual blood sugars (<200 mg/dL)

    -- and the array of other bad outcomes (though several of these have conflicting associations in the medial literature): severe urinary tract infections/urosepsis, bad fungal skin infections, bone fractures. And there was even an FDA black box warning in 2017 related to serious infections and amputations from early use of SGLT2's, though that was rescinded in 2020

 

-- and there is reasonable concern (by me) about their utility for heart failure

    -- it seems that the cardiovascular benefits of  SGLT2's relate to decreased heart failure problems (including hospitalizations and deaths): (http://gmodestmedblogs.blogspot.com/2020/02/diabetes-glp1-agonists-vs-sglt2.html  

    -- this benefit for heart failure may be not so important:

        -- it is likely largely to be from the SGLT2 diuretic effects that SGLT2's prevent heart failure by excreting lots of water along with the sugar (the latter being why diabetes is better)

        -- and, assuming this is the major mechanism leading to decreased heart failure:

            -- why not use a tried-and-true loop diuretics that are titratable to the appropriate level needed to treat the heart failure

                -- and torsemide seems to be the best choice over furosemide, since torsemide is not only is once-a-day, has increased bioavailability, and reduces aldosterone production/sympathetic activation/myocardial fibrosis etc: http://gmodestmedblogs.blogspot.com/2023/01/heart-failure-torsemide-vs-furosemide.html)

            -- so, the equivalent (and likely superior strategy over SGLT2's) would be to test people regularly who are at high risk of heart failure (high cardiovascular risk for atherosclerosis, CKD, diabetes...), even if there are no heart failure symptoms

                 -- this strategy has been found effective in patients with CKD without known heart failure: testing them every 3 months decreases heart failure development and also decreases CKD progression (http://gmodestmedblogs.blogspot.com/2023/12/routine-bnp-assessment-helpful-for.html)

                -- so, a better strategy might be to test patients with diabetes (and the other high risk conditions for developing heart failure) at some frequency (every 3-6 months) for heart failure with routine BNP assessments, and then consider adding torsemide if the BNP rises...

 

 -- SGLT2's do help with diabetes control, though not nearly as much as GLP-1RAs which can lower HgbA1c concentrations much more significantly and also have the benefit of decreasing atherosclerotic outcomes

-- But, as per above, SGLT2’s are associated with a significant increase to several of the known but uncommon severe adverse events that have been reported particularly in patients with diabetes, including increased risk of diabetic ketoacidosis in those with even pretty low blood sugars (<200 mg/dL), urosepsis, severe mycotic infections, Fournier's gangrene (http://gmodestmedblogs.blogspot.com/2019/05 /sglt-2-inhibitors-and-fourniers-gangrene.html), and perhaps bone fractures

 

Limitations:

-- this study chose DPP-4 meds as their comparator to the SGLT2’s. They chose these because DPP-4’s are “cardio-neutral” (ie, no documented clear cardiac benefit), they are contemporary with SGLT2’s as new meds for diabetes, and they are both oral meds.

-- though not mentioned by the authors, SGLT2’s and DPP4’s have similar anticipated A1c reductions of between 0.5% to 1%. However, one could argue (as I would as per above) that the real contender in diabetes meds is between GLP-1 and SGLT2 meds, with DPP4 meds having other issues (eg, blocking many different bioactive peptides as well as GLP1 and having not much effect on A1c, and also a reported increase in developing inflammatory bowel disease: I myself don’t think there is not much of an indication for using DPP4's: https://gmodestmedblogs.blogspot.com/2018/03/dpp-4-inhibitors-increasing-ibd.html )

    -- I would also add that A1c is not a great surrogate marker for diabetes: it does pretty well in assessing blood sugar levels over time (though there are conditions where it is not accurate, such as hemolytic anemia, sickle cell), but the real concern with diabetes is atherosclerotic morbodity and mortality, and these outcomes are largely related to the meds used to treat the diabetes (GLP-1's being the most atherosclerosis protective, insulin/sulfonylureas being proatherogenic in many studies): see https://gmodestmedblogs.blogspot.com/search?q=surrogate+marker

-- amputations are quite infrequent with SGLT2 medications, as noted above, though there does seem to be an increased risk with these medications.

    -- this low frequency of amputations associated with SGLT2 inhibitors does affect the statistical rigor In confirming the association with amputations or whether there are differences between different SGLT2 medications

-- there were several exclusions to the study that could interfere with the generalizability of the results. For example they excluded patients whose initial treatment of diabetes was not metformin, insulin or sulfonylureas. This also excludes patients who were on nonpharmacologic therapy alone for their diabetes. And as a VA study, not so many women were included, limiting generalizability of the results

-- those patients who were put on SGLT2's in this VA study may have been on these because of their benefit for heart failure and chronic kidney disease,  potentially skewing the results to these patients with more serious baseline comorbidities and perhaps higher risk for PAD

-- this was a short-term study with a median follow-up of 0.7 years (though they did have some patients followed up to four years, but these results are suspect since this study was not designed for this outcome). this limits the generalizability of the study to patients who initially have much less PAD (early on the clinical curve; perhaps with no benefit from SGLT2's in this short-term study) versus those with much more atherosclerotic disease who may be more susceptible to early PAD from the SGLT2's. There really needs to be a longer study with very aggressive assessment of all cardiovascular risk factors to provide definitive conclusions

-- They did a statistical analysis to determine the E-value, a measure of potential unmeasured confounding, finding a value of 1.37 which is right on the borderline between exposure and composite PAD-related surgical outcomes. this suggests that a moderate confounder could render the study results inconclusive. they did do propensity scoring to decrease the likelihood of confounding, such as choosing SGLT2's for those with known underlying cardiovascular disease, but there could certainly be other confounders that were not assessed

 

 so,

-- there do continue to be concerns about the whole class of SGLT2 medications, with quite uncommon but also quite serious complications including amputations, diabetes ketoacidosis at low levels of hypoglycemia, severe urinary tract infections, fungal skin infections, bone fractures.  Since these are potentially such serious complications, patients should be aware of these possibilities.

-- I do have concerns about SGLT2's as noted above, including whether they are better than doing routine testing for heart failure and treating with a loop diuretic such as torsemide even before the patients have any heart failure symptoms (and the physical exam for heart failure has a sensitivity of about 50%); and for diabetes, it seems to me pretty clear that GLP-1 agonists are associated with far greater decreases in hemoglobin A1c levels coupled with a general cardiovascular benefit (not just for heart failure) as well as renal protection (also found with SGLT2 medications). it is clear that SGLT2's do have a role in diabetes management, though one could argue that this should be after GLP-1's, if tolerated

 

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