DPP-4 inhibitors increasing IBD

A recent data-mining study found a significant association between starting DPP-4 (dipeptidyl peptidase-4) inhibitors for diabetes and the later development of inflammatory bowel disease (see dm dpp4 and ibd bmj2018 in dropbox, or doi.org/10.1136/bmj.k872.

Details:
--141,170 patients >18yo starting antidiabetic meds, from 2007-2016, and followed a median of 3.6 years. from the United Kingdom Clinical Practice Research Datalink of  >700 general practices . 

--mean age 62, 15% with BMI >30, 15% alcohol-related disorders, 16% current smokers, 31% A1c>8%

--primary outcome: adjusted hazard ratio for incident inflammatory bowel disease (IBD) with use of DPP-4 inhibitors vs other diabetes meds

--30,488 (21.6%) received script for DPP-4 inhibitors; median duration of use =1.6 yrs

--patients on DPP-4 inhibitors: older (66 vs 61yo),  more likely to have higher A1c concentration (A1c>8% in 44% vs 30%), longer duration of diabetes (4.2 vs 2.5 yrs), more microvascular complications (twice as many), more likely to have taken aspirin (66% vs 42%) and NSAIDs (64% vs 54%) and less likely to have taken oral contraceptives (6 vs 8%). no difference in other autoimmune conditions or mean number of diabetic drugs

Results:

--208 cases of incident IBD during 552,413 person-yrs of follow-up: crude incidence rate of 37.7/100K person-yrs. All results adjusted for age, sex, year of cohort entry, BMI, alcohol-related disorders, smoking status, as well as diabetes-related variables: Hgb A1c, micro- and macrovascular complications, duration of diabetes, and prior diabetes meds used 

--DPP-4 inhibitor use was associated with 75% increased incidence of IBD: HR 1.75 (1.22-2.49), and absolute increase from 34.5 to 53.4/100K person-yrs, number-needed-to-harm = 2291 patients over 2 years and 1177 over 4 years

--the hazards ratio increased with longer duration of DPP-4 inhibitor use: HR 2.90 (1.31-6.41) after 3-4 years, then decreasing to 1.45 (0.44-4.76) [though the numbers of patients on the meds for >3 years was pretty small]

--IBD association with DPP-4 inhibitors was signficantly higher only for ulcerative colitis, HR 2.23 (1.32-3.76) and not Crohn's, HR 0.87 (0.37-2.09) [though there were really too few events to know for sure: 44 total with Crohn's vs 96 with UC]

--comparison to insulin use (as a means to control for harder-to-treat diabetes, since insulin is the med typically used in advanced disease):

    --no difference in IBD risk in those on vs not on insulin    

    ​--head-to-head comparison:  DPP-4 inhibitors vs insulin had a 2.28-fold increase risk of IBD, adjusted HR of 2.28 (1.07-4.85)

--sensitivity analyses: no difference in outcomes by any of the analyses done, including more rigorous definition of clinical events, stratification by age < vs >60yo, exclusion of TZDs (glitazones), multiple imputations (to control for missing information), disease risk score (an alternative to propensity scoring), or use of GLP-1 agonists (to control for the incretin effect, since both drugs increase GLP-1 levels)

Commentary:

--this retrospective data-mining study has the limitations of all such studies, including the fact that they can only posit an association and not causality. in particular, there is always concern about unknown confounders, or ones not measured in the database. But this study found a pretty large association, and per the authors, "a hypothetical confounder would need to be strongly associated with both the exposure (odds ratio >4.7) and the outcomes (relative risk >5.0) to move the point estimate towards the null", ie it would have to be a pretty substantial confounder to alter the results of the above association.

--a major concern I have had is that these DPP-4  inhibitors block many different cellular functions and are not specific in their blocking GLP-1 (their presumed anti-diabetic effect is that they block the breakdown of GLP-1, so there is more glucose-mediated stimulation of insulin release from the pancreas). However, the DPP-4 enzyme, a serine protease, is on the surface of many different cells, and affects many different bioactive peptides including those which affect the immune response. It seems that DPP-4 affects T-cell function in particular: “maintaining lymphocyte composition and function, T cell activation and co-stimulation, memory T cell generation  and thymic emigration patterns during immune-senescence”  and promotes Th1 cytokine response (per Klemann C. Clin Exp Immunol. 2016; 185: 1-21, or go to https://onlinelibrary.wiley.com/doi/epdf/10.1111/cei.12781 ), who also notes “therefore, inhibition of DPP-4 might represent a double-edged sword”. Other studies have found that lower concentrations of DPP-4 serum levels are associated with higher clinical disease score in those with Crohn's but not ulcerative colitis (see Hildebrandt M. Xcand J Gastroenterol 2001; 36:1067), though not clear whether higher IBD clinical scores led to lower DPP-4 levels or vice versa. DPP-4 also modulates gastric hormones

    --i am also concerned about using the TZDs (glitazones),  though there may be cardioprotection (found in PROactive study, but then not in the  POSCA-IT study), since they also have myriad nondiabetes effects by activating a group of peroxisome proliferator-activated-gamma nuclear receptors. though i am more inclined to use pioglitazone than DPP-4 inhibitors because they may be cardioprotective and have been around for a long time.​

--see a recent blog on A1c goals, emphasizing using cardioprotective diabetes meds, with some negative comments of DPP-4's: http://gmodestmedblogs.blogspot.com/2018/03/loosening-a1c-goal-is-low-a1c-really.html . another blog reviewed data on DPP-4 inhibitors and increased hospitalizations for heart failure ( http://gmodestmedblogs.blogspot.com/2016/04/diabetes-dpp-4-inhibitors-and-risk-of.html  )

So, the DPP-4 association with IBD may well be significant, though the actual number of IBD cases is pretty low. But, this study does bring up a pretty general issue: it is more likely that a very targeted drug (eg the GLP-1 agonists) will be safer than drugs which affect multiple diverse targets (as with DPP-4 inhibitors).  This does seem pretty obvious on the surface, but my concern is that many of the clinical studies are pretty short-term (and it may take more time to uncover the drug’s nefarious effects), many studies are stopped early because of demonstrated benefit (which further short-changes the likelihood of finding one of these adverse effects, which then have even less time to manifest themselves), the drug companies unabashedly promote the positives (my guess is that they do not say “this is a really pretty mild/mediocre diabetes drug, though it affects huge numbers of non-diabetes related bioactive peptides all over the body”), and it really took me some digging in the non-false news medical literature to find out about the shotgun effects of these drugs.​