Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Jul 1.
Published in final edited form as: Inflamm Bowel Dis. 2011 Nov 13;18(7):1315–1324. doi: 10.1002/ibd.21929

Spironolactone and colitis: Increased mortality in rodents and in humans

Laura A Johnson 1, Shail M Govani 1, Joel C Joyce 2, Akbar K Waljee 1, Brenda W Gillespie 3, Peter D R Higgins 1
PMCID: PMC3288762  NIHMSID: NIHMS329727  PMID: 22081497

Abstract

Background and Methods

Crohn’s disease causes intestinal inflammation leading to intestinal fibrosis. Spironolactone is an anti-fibrotic medication commonly used in heart failure to reduce mortality. We examined whether spironolactone is anti-fibrotic in the context of intestinal inflammation. In vitro, spironolactone repressed fibrogenesis in TGFβ-stimulated human colonic myofibroblasts. However, spironolactone therapy significantly increased mortality in two rodent models of inflammation-induced intestinal fibrosis, suggesting spironolactone could be harmful during intestinal inflammation. Since IBD patients rarely receive spironolactone therapy, we examined whether spironolactone use was associated with mortality in a common cause of inflammatory colitis, Clostridium difficile infection (CDI).

Results

Spironolactone use during CDI infection was associated with increased mortality in a retrospective cohort of 4008 inpatients (15.9% vs. 9.1%, n=390 deaths, p<0.0001). In patients without liver disease, the adjusted OR for inpatient mortality associated with 80 mg spironolactone was 1.99 (95% CI: 1.51 – 2.63) In contrast to the main effect of spironolactone mortality, multivariable modeling revealed a protective interaction between liver disease and spironolactone dose. The adjusted odds ratio for mortality after CDI was 1.96 (95% CI: 1.50 – 2.55) for patients without liver disease on spironolactone vs. 1.28 (95% CI: 0.82 – 2.00) for patients with liver disease on spironolactone, when compared to a reference group without liver disease or spironolactone use.

Conclusions

We propose that discontinuation of spironolactone in patients without liver disease during CDI could reduce hospital mortality by 2-fold, potentially reducing mortality from CDI by 35,000 patients annually across Europe and the US.

INTRODUCTION

The renin-angiotensin-aldosterone system (RAAS) has been implicated in fibrosis of multiple organs including heart, liver, pancreas, and kidney.(1–3) Spironolactone, a competitive aldosterone receptor antagonist, is a potent anti-fibrotic, improves the survival of congestive heart failure patients, and is protective in several rodent models of organ fibrosis.(4–7)

Crohn’s disease (CD) frequently produces intestinal fibrosis and strictures requiring surgical intervention.(8, 9) Current therapies control inflammation and improve symptoms, yet do not alter the development of intestinal fibrosis and the natural history of CD.(10) Intestinal wound healing is mediated by myofibroblasts which are postulated to be the major contributors to intestinal fibrosis. (11) Myofibroblasts are characterized by expression of α-smooth muscle actin (αSMA) and are activated by transforming growth factor-β1 (TGFβ) in fibrotic diseases.(12) Angiotensin II stimulates TGFβ in cardiac myofibroblasts, therefore we hypothesized that downstream aldosterone blockade of this signaling should prevent TGFβ-induced fibrosis in vitro and in vivo.

We investigated whether spironolactone reduces intestinal fibrosis in an in vitro colonic myofibroblast model and in two rodent models of intestinal fibrosis. We determined that spironolactone is anti-fibrotic in vitro. Paradoxically, in two different rodent colitis models, spironolactone therapy during intestinal inflammation produced rapid and significant mortality. To evaluate the clinical relevance of our rodent mortality results, we considered evaluating the effect of spironolactone on mortality of IBD patients with active inflammation, but spironolactone use is rare in patients with IBD. As an alternative, we examined the mortality associated with spironolactone use in an easily identifiable and common source of human colonic inflammation, Clostridium difficile infection. We performed a multivariable logistic regression in a retrospective cohort of inpatients with CDI to evaluate the effect of spironolactone use on CDI mortality.

MATERIALS AND METHODS

In vitro model reagents

Human recombinant TGFβ1 was obtained from R&D Systems (R&D Systems, Minneapolis, MN). Spironolactone and canrenoic acid were purchased from Sigma Aldrich. Enalprilat was acquired from Hopsira (Hopsira, Lake Forest, IL). Eplereone was purchased from Tocris (Tocris, Ellisville, MO). Losartan was obtained from Merck (Merck, Whitehouse Station, NJ) and aliskiren from Novartis (Novartis, East Hanover, NJ).

In vitro myofibroblast culture methods

Early passage (3 to 12) colonic human fibroblast CCD-18Co cells (CRL-1459 from ATCC) were cultured in alpha-MEM (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum and sub-cultured weekly. For in vitro experiments, cells were plated at 30–40% confluence and serum-starved prior to treatment with TGFβ and other compounds. To stimulate a fibrotic phenotype, CCD-18co cells at 30–40% confluence were serum-starved for 24hr prior to treatment with 1 ng/ml TGFβ or 1 ng/ml TGFβ and 100 μM spironolactone or 1 mM canrenone for 48 hours. In the RAAS inhibitor experiments, CCD-18co cells were stimulated with 1 ng/ml TGFβ to which either 50 μM aliskiren, 100 nM enalprilat, 10 μM losartan, or 50 μM eplerenone was added. Cells were harvested after 48 hours.

Protein expression

Total cellular lysates were subjected to SDS-PAGE electrophoresis as previously described.(13) αSMA protein was detected with a mouse monoclonal antibody to human αSMA protein (Sigma, St. Louis, MO). GAPDH protein expression was used as a loading control using a mouse antibody to GAPDH (Chemicon, Temecula, CA). After application and washing of the primary antibodies, membranes were incubated with a HRP-conjugated antibody to mouse IgG (Invitrogen, Carlsbad, CA) and developed using the Pierce Dura detection system (Pierce, Rockford, IL). The resulting autoradiographs were scanned and the images were quantitated using the ImageJ analysis software (available at http://rsbweb.nih.gov/ij/).

Gene expression studies

RNA from CCD-18co cells was extracted using the RNeasy kit (Qiagen, Valencia, CA). cDNA was generated by reverse transcription of 2 μg of total RNA using the Superscript First Strand RT kit (Invitrogen, Carlsbad, CA). Quantitative real-time PCR (qPCR) was performed for Acta2, Col1a1, Ctgf, and Gapdh with the TaqMan gene expression assays (ABI, Foster City, CA) on a Stratagene Mx3000P real-time PCR system (Stratagene, La Jolla, CA). Gene expression was normalized to GAPDH as the endogenous control, and fold-changes (RQ) relative to uninfected controls (no Tx) were calculated using the ΔΔCt-method.(14)

Rodent Models

Chronic colitis and subsequent fibrosis was induced using the rat TNBS enema model as previously described.(15) 0.5, 2.5, 10, or 20 mg/kg/day spironolactone was administered twice daily by oral gavage. Control animals received vehicle twice daily by oral.

In the mouse colitis and intestinal fibrosis model, mice were infected with Salmonella typhimurium as previously described.(16) Briefly, female 8–12-week-old 129S1/SvImJ mice (Jackson Laboratories, Bar Harbor, ME) were divided into two groups. Half the animals received regular rodent chow, the other half received a custom rodent chow containing 2.78g spironolactone per kg (Harlan Teklad, Madison, WI). After 1 week, half of the animals in either the standard chow or spironolactone supplemented chow groups received 20 mg of streptomycin in 0.1 M Hank’s balanced salt solution (HBSS) by oral gavage 24 hours prior to oral infection with 3×106 colony-forming units (CFU) of S. typhimurium strain SL1344 in 100 μl 0.1 M HEPES (pH = 8.0). Uninfected animals received streptomycin and HEPES on the same schedule as the infected groups. Given the 100% mortality in the S. typhimurium/SPIR group, the uninfected spironolactone group was subsequently challenged at day 16 of the 21 day experiment with S. typhimurium as described above. Surviving mice were euthanized at 21 days post-S. typhimurium infection. Mice were monitored for weight loss/gain, general appearance, and health. All animal experiments were conducted with the approval and oversight of the University of Michigan UCUCA (University Committee on Use and Care of Animals).

Retrospective Clostridium Difficile Colitis Study Design

A retrospective cohort of inpatients with a discharge diagnosis of CDI at the University of Michigan Hospital, Ann Arbor, Michigan was reviewed with institutional review board approval. The consent of participants to review of their medical records was waived for this protocol, as the large number of subjects made consent impractical and the risks to subjects were minimal. The electronic medical system was queried for all inpatients older than 18 years of age with a discharge diagnosis of CDI from 1/1/2000 through 12/31/2009. Demographics, labs, discharge diagnoses, and other pertinent data were abstracted electronically. To determine spironolactone dosing, a manual chart review was performed, with the highest in-hospital dose used for modeling. Patients on eplerenone therapy were excluded due to small sample size (n=4). Definitions of heart failure and liver disease as well as descriptions of methods of laboratory value abstraction are detailed in the Supplementary Methods. Selection of the patient sample population for modeling is detailed in Figure 1.

Figure 1. Consolidated Standards of Reporting Trials (CONSORT) flow diagram.

Figure 1

Open in a new tab

Selection of inpatient subjects for modeling among patients with a discharge diagnosis of CDI. Re-admissions for recurrent CDI within 30 days of discharge were treated as a single admission for this analysis.

Statistical Methods

The descriptive statistics are reported as means ± standard deviation (SD) or percentages. Comparisons between the spironolactone users and non-users were made by Student’s t test for quantitative variables and chi-square tests for categorical variables. Single predictor odds ratios were calculated for inpatient mortality using the LOGISTIC procedure in SAS 9.2 (Cary, NC). A multivariable model of inpatient mortality was chosen using the best subset selection method in SAS while maintaining spironolactone dose, liver disease, and heart failure in the model. As 18.6% of patients were hospitalized more than once, the effect of hospitalization on mortality was investigated, and mortality was found to be greater in the first two hospitalizations. Therefore, the single predictor and multivariable models were adjusted for the hospitalization number for each individual, dichotomized as ≤ 2 vs. > 2 hospitalizations. All possible two-way interactions were tested. P values less than 0.05 were considered statistically significant. All tests are 2 sided.

The original MELD formula, which allows for negative values, was used.(17) The optimal functional form of spironolactone dose in the multivariable model was assessed by comparing linear to both logarithmic and quadratic transformations using AIC and likelihood ratio testing. Sensitivity of the model to high doses of spironolactone (> 100mg) was tested by comparing parameter estimates in a model with and without these higher doses. As stated in the Results section, this did not change the significance of any of the parameters.

The logistic regression analysis strategy was based on a model for discrete survival. In our case, the time axis was hospitalization number, with death (yes/no) evaluated for each hospitalization. In such discrete survival models, time is entered as a model term. After considering continuous (linear) versus categorical variables for hospitalization number, we determined that using categories (<= 2 hospitalizations vs. > 2) provided the best fit. To test the robustness of our modeling method, we also performed logistic regression on a random hospitalization for each patient. The results of this alternative modeling method did not change the significance of any of the parameters.

RESULTS

In vitro studies

To determine whether spironolactone is anti-fibrotic in vitro, human colonic myofibroblasts (CCD-18co) were stimulated with TGFβ to induce a fibrotic phenotype characterized by increased actin stress fiber accumulation, αSMA protein expression, and increased expression of pro-fibrotic genes including Acta2, Col1a1, and Ctgf (data not shown). Spironolactone repressed TGFβ induction of αSMA protein expression in a dose-dependent manner (Figure 2A). Spironolactone repressed actin stress fiber formation (data not shown). Both spironolactone and its active metabolite, canrenone, dramatically repressed TGFβ–induced αSMA protein expression (data not shown). Spironolactone repressed the expression of fibrotic genes Acta2, Col1a1, and Ctgf to levels indistinguishable from untreated controls (Figure 2B). Canrenone significantly repressed Acta2 mRNA expression, but its reduction in Col1a1 and Ctgf mRNA expression were not statistically significant (Figure 2B).

Figure 2. Results of in vitro studies.

Figure 2

Open in a new tab

(A) Spironolactone inhibits TGFβ induction of αSMA protein expression in colonic myofibroblasts. A representative Western blot of αSMA expression in protein extracts from CCD-18co colonic myofibroblasts stimulated for 24 h with TGFβ is shown. Increasing amounts of spironolactone (SPIR) from 10 μM to 100 μM reduce αSMA expression to levels comparable to unstimulated cells (no Tx). GAPDH expression serves as the loading control for the amount of protein. (B) Treatment of TGFβ stimulated CCD-18co cells with spironolactone (SPIR) represses expression of Acta2, Col1a1, and Ctgf. A metabolite of spironolactone, canrenone (CAN) partially represses pro-fibrotic gene expression. (C) The role of the RAAS pathway in fibrosis and the relationship of pathway inhibitors. (D) Inhibitors of the RAAS pathway aliskiren (ALK), enalaprilat (ENT), and losartan (LOR) partially repress TGFβ induction of fibrotic genes with partial repression of Acta2 expression but have minimal effect on Col1a1 expression (E). Results are from 9 independent experiments. Asterisks denote statistically significant comparisons between untreated control cells (no Tx) and the treatment groups. Brackets denote comparisons between TGFβ treated and other treatment groups. * P <0.05, *** P <0.001

Given the anti-fibrotic effect of spironolactone, we investigated whether blocking upstream components of the renin-angiotensin (RAAS) pathway (Figure 2C) would block fibrosis in vitro. In TGFβ-stimulated myofibroblasts, other RAAS inhibitors aliskiren (renin inhibitor), enalaprilat (ACE inhibitor), and losartan (angiotensin II type I receptor blocker (ARB)) partially repressed pro-fibrotic gene expression (Figure 2D and Figure 2E) suggesting a role for the RAAS pathway in intestinal fibrosis.

Rodent colitis studies

Unexpectedly, in two animal models of intestinal inflammation and fibrosis, treatment with spironolactone caused significant and rapid mortality. In the rat TNBS chronic colitis model, 20 mg/kg/day spironolactone (10-fold lower than well-tolerated doses in other rat models(18)) produced 44% mortality (95%CI: 0.14–0.79) (Figure 3A). A dose-response experiment in the rat TNBS model demonstrated increased survival with decreasing spironolactone dose. Mortality was 33% at 10 mg/kg/day, and 0% at 2.5 or 0.5 mg/kg/day (Figure 3B). However, lower doses of spironolactone did not reduce the development of fibrosis, as determined by gross pathology, histopathology, fibrotic gene expression, and protein expression (αSMA) (data not shown). Similar experiments with losartan resulted in no mortality and no improvement in fibrosis (data not shown).

Figure 3. Results of rodent colitis studies.

Figure 3

Open in a new tab

(A) Increased mortality in rats with chronic TNBS colitis treated with spironolactone. Kaplan-Meier mortality estimates of rats with chronic TNBS and 20mg/kg/day SPIR (TNBS+SP) compared to no mortality with TNBS alone and in untreated rats (no Tx). Data are from three rats per experimental group. (B) Increased mortality in rats with chronic TNBS colitis treated with spironolactone is dose-dependent. The Kaplan-Meier curve demonstrates increased mortality in rats with chronic TNBS colitis treated with doses of 10 or 20 mg/kg/day SPIR (T+SP 20, T+SP 10) compared to low doses of SPIR (T+SP 0.5, T+SP 2.5), TNBS alone, or rats with no treatment (noTx). Data are from three rats per experimental group. (C) Spironolactone treatment increases mortality in the S. typhimurium mouse model of colitis. Mortality occurred in mice with S. typhimurium-induced colitis treated with 0.7 mg/kg/day SPIR (SP +St), compared to mice infected with S. typhimurium without SPIR (St), uninfected mice (no Tx), or mice receiving 0.7 mg/kg/day SPIR alone (SP). Uninfected mice which received SPIR treatment over 15 days had 0% mortality until subsequent S. typhimurium induction of colitis (SP→SP + St) at day 16 (vertical arrow) which produced 100% mortality by day 5 post-infection. Data are from five mice per experimental group.

In the mouse S. typhimurium infection model of intestinal fibrosis, treatment with 0.7 mg/kg/day of spironolactone produced 80% mortality (95%CI: 0.69–1.00) by day 9 of colitis compared to 20% mortality in the S. typhimurium-infected group (Figure 3C). No mortality occurred in uninfected mice receiving 0.7mg/kg/day of SPIR. However, when the spironolactone-treated cohort was later challenged with S. typhimurium, 100% mortality occurred by day 5 post-infection.

Retrospective study of Clostridium difficile colitis clinical outcomes

To determine whether spironolactone use was associated with increased mortality in humans with intestinal inflammation, we identified a total of 4,008 inpatients with a discharge diagnosis of CDI in University of Michigan Hospital admissions from 1/1/2000 through 12/31/2009. These 4,008 patients accounted for a total of 5,166 CDI-associated hospitalizations in that 10-year period. There were 352 patients on spironolactone during at least one hospitalization; these patients were hospitalized 391 times. The average time between hospitalizations for spironolactone users was not significantly different from non-users while the average number of hospitalizations for the two groups did differ (1.4 vs. 1.3, p=0.02, n=4,008) (Table 1). In patients on spironolactone therapy, the average dose was 79.9 mg (± 71.5). The average age, gender ratio, and race did not differ between spironolactone users and non-users.

Table 1.

Characteristics of Patients with C. difficile Colitis on Spironolactone Therapy.

Characteristic Spironolactone Therapy (N = 352)
Mean (± SD), Mean [Range] or n (%)		 No Spironolactone Therapy (N = 3656)
Mean (± SD), Mean [Range] or n (%)		 P value (N)
Age (years) 57.1 (± 14.2) 57.6 (± 17.9) 0.53
Male 181 (51.4%) 1859 (50.9%) 0.84
Race 0.29
 Caucasian 282 (80.1%) 3,025 (82.7%)
 African-American 37 (10.5%) 369 (10.1%)
 Other 33 (9.4%) 262 (7.2%)
Hospitalizations (individual) 391 4,775
# of Hospitalizations per patient 1.4 [1 – 6] 1.3 [1 – 12] 0.02
Time between hospitalizations (days) 132.5 [1 – 1,483] 129.2 [1 – 2,383] 0.90
Heart Failure 137 (38.9%) 549 (15.0%) <0.0001
Liver Disease 151 (42.9%) 273 (7.5%) <0.0001
Diabetes 90 (25.6%) 835 (22.8%) 0.25
Baseline Measures (per hospitalization)
Sodium (mmol/L) 134.9 (± 5.6) 137.4 (± 4.8) <0.0001 (5,121)
Potassium (mmol/L) 4.2 (± 0.6) 4.1 (± 0.5) 0.0004 (4,582)
Creatinine (mg/dL) 1.3 (± 0.7) 1.5 (± 1.6) <0.0001 (5,118)
BUN (mg/dL) 27.5 (± 18.3) 25.3 (± 20.4) 0.03 (5,119)
INR 1.6 (± 1.0) 1.4 (± 1.0) 0.007 (3,763)
Albumin (g/dL) 3.1 (± 0.7) 3.1 (± 0.7) <0.0001 (4,432)
Total bilirubin (mg/dL) 3.1 (± 5.3) 1.3 (± 3.5) <0.0001 (4,479)
MELD 6.6 (± 9.0) 2.5 (± 10.1) <0.0001 (3,416)
Charlson-Deyo Comorbidity Index 4.1 (± 2.7) 3.9 (± 2.7) 0.25 (3,741)
Outcomes
Mortality (by individual) 56 (15.9%) 334 (9.1%) <0.0001
Mortality (by hospitalization) 49 (12.5%) 341 (7.1%) 0.0001
Length of each hospitalization (days) 25.0 [1 – 171] 17.1 [1 – 851] <0.0001
Cost per Hospitalization (in $1,000’s) 197.9 [0 – 2489.4] 110.4 [0 – 6388.0] <0.0001
Open in a new tab

As expected, the group on spironolactone therapy had significantly more patients with a diagnosis of heart failure (38.9% vs. 15.0%, p<0.0001) and liver disease (42.9% vs. 7.5%, p <0.0001) compared to non-users of spironolactone. The mortality in those patients with heart failure was 77 of 686 (11.2%), in those with liver disease was 71 of 424 (16.7%), and in patients with neither heart failure nor liver disease was 242 of 2,898 (8.4%). The majority of patients taking spironolactone without a history of heart failure or liver disease were doing so due to hypertension or evidence of edema.

Lab values affected by heart or liver dysfunction (creatinine, sodium, INR, total bilirubin, potassium, albumin) were significantly different between users and non-users of spironolactone (see Table 1), and were used as proxies for disease severity in a predictive model. Bilirubin and INR were higher in the spironolactone group while sodium, albumin, and creatinine were lower. Discharge day potassium was higher in the spironolactone group. The MELD(17) score showed a significant difference between the two groups, while there was not a significant difference in Charlson-Deyo comorbidity index.(19, 20) The clinical outcomes in the two groups demonstrated significant differences, with longer length of stay, higher mortality, and higher costs per hospitalization in the spironolactone users.

Figure 4 shows the single predictor odds ratios (OR) for inpatient mortality in all patients after adjustment for hospitalization number. In this analysis (n=5,166, patients on spironolactone had 1.84 (95% CI: 1.34 – 2.53) times the odds of inpatient mortality of those not on the medication. For each additional 25 mg of spironolactone dose, the OR for mortality was increased by 13% (95% CI: 6% – 20%). Liver disease, higher MELD scores, and heart failure were associated with higher inpatient mortality, while Charlson-Deyo scores were not. ACE-I (angiotensin converting enzyme inhibitor) and ARB (angiotensin 2 receptor blocker) use were both associated with decreased mortality.

Figure 4. Single Predictor Odds Ratios for Inpatient Mortality in Patients with C. difficile Colitis.

Figure 4

Open in a new tab

For each predictor, the odds ratio from a logistic model for inpatient mortality was adjusted for hospitalization number (≤ 2 vs. > 2) and is presented with 95% confidence intervals. A vertical dashed line at 1 represents the point of no effect.

These predictors were combined in a multivariable logistic regression model (n=4,415) to predict inpatient mortality (Figure 5A). We found that spironolactone use remained a significant predictor of mortality, and found a statistically significant interaction between spironolactone dose and liver disease, described below. Testing the other univariate predictors, we found that blood urea nitrogen (BUN), total bilirubin, albumin and use of ACE-I and ARB medications were significant contributors to our model of inpatient mortality. No significant interaction between heart failure and spironolactone use was found. 4,415 hospitalizations without missing values were included in this model. Hospitalization number was also included for each individual in the model; patients were more likely to die during one of the first two hospitalizations than in subsequent hospitalizations (OR 1.52, 95%CI: 0.92 – 2.50, p=0.10). This model had fairly high explanatory power, with a c statistic of 0.71. The linear form of spironolactone dosing in the model was statistically superior to a quadratic or logarithmic transformation. The exclusion of high doses of spironolactone (> 100 mg) did not substantially alter the model.

Figure 5. Adjusted Odds Ratios for and Predicted Probabilities of Inpatient Mortality in Patients with C. difficile Colitis.

Figure 5

Open in a new tab

(A) The adjusted odds ratios for inpatient mortality in a multivariable model containing spironolactone, liver disease, heart failure, the interaction between spironolactone and liver disease, ACE-I use, ARB use, total bilirubin, albumin, BUN, and hospitalization number (≤ 2 vs. > 2). The OR for liver diseases with and without spironolactone and the OR of liver disease with spironolactone are compared to patients without liver disease not taking spironolactone (Reference group).

(B) The predicted probability (with 95%CI) of inpatient mortality in patients with liver disease, heart failure and neither liver disease nor heart failure while using 3 possible doses of spironolactone for that disease state. For each disease state, the average total bilirubin, albumin and BUN were used in conjunction with the proportion of patients on ACE-I and ARB medications. Details for each disease state are located in the appendix. In all groups, the probability is shown for hospitalization number ≤2. This demonstrates that spironolactone use is associated with increased mortality in patients without liver disease.

Because of the interaction between liver disease and spironolactone, patients with liver disease had a (non-significant) protective effect from spironolactone of 29% (OR 0.71, 95% CI 0.47 – 1.07), while patients without liver disease had significantly worse outcomes with spironolactone. Compared to the reference group (patients without liver disease not taking spironolactone), patients with liver disease taking the average dose of spironolactone (80 mg) had an adjusted OR for inpatient mortality of 1.07 (95% CI: 0.70 – 1.64). In patients without liver disease, the adjusted OR for inpatient mortality associated with 80 mg spironolactone was 1.99 (95% CI: 1.51 – 2.63) (Figure 5A) compared to the reference group. We show the direct effects of this interaction term on mortality in Figure 5B, where the probabilities of inpatient mortality in patients with liver disease, heart failure, and with neither of these comorbidities are illustrated. To show the dose-response effect, we modeled the predicted probability of inpatient mortality for each disease state for patients on high dose spironolactone, low dose spironolactone, and those not taking the drug. Predicted probabilities are based on the average values of total bilirubin, albumin and BUN, ACE-I/ARB use, and hospitalization number ≤2. Patients with liver disease had a lower probability of inpatient mortality if they were using spironolactone. In contrast, in patients with heart failure, who are typically on lower doses of spironolactone, we demonstrate an increase in inpatient mortality with spironolactone use. The patients without liver disease or heart failure were on higher average doses of spironolactone, and this is reflected in a significantly increased risk of inpatient mortality.

DISCUSSION

The anti-fibrotic effects of aldosterone antagonists have been demonstrated in multiple organs. Clinically, the use of spironolactone has a significant survival benefit in heart failure, but the effect of spironolactone on intestinal fibrosis is unknown. In this study, we found that spironolactone represses fibrotic protein expression and the expression of pro-fibrotic genes in intestinal myofibroblasts. However, we found that spironolactone intervention during inflammatory colitis significantly increased mortality in two rodent models of inflammatory intestinal fibrosis. We extended this finding to patients in a retrospective cohort study of clinical outcomes at a large tertiary center, and found that spironolactone use in patients with CDI was associated with an increased mortality rate in patients without liver disease, even after adjusting for comorbidities and disease severity.

Spironolactone is absorbed through the gut mucosa and metabolized in the liver. While the pharmacokinetics of spironolactone metabolism are complex due to the large number of active metabolites generated, in patients with liver disease, impaired spironolactone metabolism has been reported. (21–23) Altered hepatic spironolactone metabolism may account for the differential mortality outcomes of liver disease compared to non-liver disease patients.

Our final model includes high blood urea nitrogen as a predictor of mortality, which may be a marker of dehydration and therefore severity of CDI. Based on our in vitro results, we would anticipate that ACE inhibitors and ARBs, as inhibitors of the RAAS, would have effects similar to spironolactone, but this was not the case. The protective effect of ACE inhibitors and ARBs suggests that spironolactone may have a distinct mechanism that is associated with mortality in the setting of CDI. Our findings are supported by a recent retrospective study which demonstrated ACEI/ARB use during CDI was associated with decreased mortality.(24)

While we cannot infer causality with a retrospective study, several aspects of our study that provide compelling evidence for the combination of spironolactone and intestinal inflammation affecting mortality. In our rodent models, we did not observe adverse outcomes when animals were treated with spironolactone alone, or when colitis was induced without spironolactone. In addition, we observed a dose-dependent increased mortality in the rodent models. Clinically, we observed a dose-dependent effect of spironolactone on inpatient mortality which was maintained despite adjusting for important covariates associated with spironolactone usage and mortality.

Though entirely speculative, spironolactone therapy in the context of epithelial barrier loss could act synergistically with the C. difficile toxins to increase inflammation severity and mortality. During CDI, toxin A disrupts the epithelial barrier from the apical side, potentiating toxin B access to receptors in the basolateral side.(25) While toxin B is essential for virulence, toxin B alone does not cause symptoms unless mucosal damage has occurred.(26–28) Toxin B, via a Rho/Rac signaling pathway, induces inducible nitric oxide synthase (iNOS), subsequent loss of epithelial barrier function, increased permeability, increased chloride secretion, and diarrhea.(29) In our rodent colitis models spironolactone had no effect in rodents with intact colons, but produced rapid mortality after damage to the intestinal epithelial barrier.

To our knowledge, spironolactone has not been reported to disrupt the epithelial barrier, but spironolactone does affect iNOS signaling. In the heart, spironolactone induces iNOS while aldosterone represses iNOS, suggesting a functional link between the iNOS pathway and spironolactone signaling.(30) In contrast to spironolactone, ACE inhibitors (captopril) and ARBs (candesartan, losartan) have been reported to reduce iNOS expression in the heart and kidney.(31–33) We postulate that during the loss of barrier integrity caused by colonic inflammation, spironolactone can act synergistically with C. difficile toxin B by further inducing iNOS, contributing to additional barrier disruption and increasing disease severity and mortality.

CONCLUSIONS

Clostridium difficile is a major source of hospital acquired infections.(34) Based on our results in the rodent models and the human cohort study, we propose that temporary discontinuation of spironolactone in patients during CDI without liver disease could significantly reduce mortality in the more than 2.7 million hospitalized patients in the US with CDI each year.(35) We estimate that a similar number of patients across Europe are hospitalized with CDI. Assuming that 8.9% of CDI patients in Europe and the US are on spironolactone therapy (as in our sample), a 2-fold reduction in mortality among spironolactone users with CDI would save an estimated 35,000 lives annually across Europe and the US. Prospective clinical studies are needed to determine whether temporary discontinuation of spironolactone in patients with CDI changes clinical outcomes.

Supplementary Material

Supplementary Data

Acknowledgments

This study was funded by a grant K08DK080172 to PDRH and UL1RR024986 to AKW. We thank E. Rodansky of the University of Michigan for critical review of the manuscript.

References

  • 1.Pitt B, Zannad F, Remme WJ, et al. The effect of spironolactone on morbidity and mortality in patients with severe heart failure. Randomized Aldactone Evaluation Study Investigators. N Engl J Med. 1999;341:709–717. doi: 10.1056/NEJM199909023411001. [DOI] [PubMed] [Google Scholar]
  • 2.Tox U, Steffen HM. Impact of inhibitors of the Renin-Angiotensin-aldosterone system on liver fibrosis and portal hypertension. Curr Med Chem. 2006;13:3649–3661. doi: 10.2174/092986706779026138. [DOI] [PubMed] [Google Scholar]
  • 3.Mezzano SA, Ruiz-Ortega M, Egido J. Angiotensin II and renal fibrosis. Hypertension. 2001;38:635–638. doi: 10.1161/hy09t1.094234. [DOI] [PubMed] [Google Scholar]
  • 4.Rajagopalan S, Pitt B. Aldosterone as a target in congestive heart failure. Med Clin North Am. 2003;87:441–457. doi: 10.1016/s0025-7125(02)00183-9. [DOI] [PubMed] [Google Scholar]
  • 5.Gullulu M, Akdag I, Kahvecioglu S, et al. Aldosterone blockage in proliferative glomerulonephritis prevents not only fibrosis, but proliferation as well. Ren Fail. 2006;28:509–514. doi: 10.1080/08860220600779033. [DOI] [PubMed] [Google Scholar]
  • 6.Ersoy R, Celik A, Yilmaz O, et al. The effects of irbesartan and spironolactone in prevention of peritoneal fibrosis in rats. Perit Dial Int. 2007;27:424–431. [PubMed] [Google Scholar]
  • 7.Nishimura H, Ito Y, Mizuno M, et al. Mineralocorticoid receptor blockade ameliorates peritoneal fibrosis in new rat peritonitis model. Am J Physiol Renal Physiol. 2008;294:F1084–1093. doi: 10.1152/ajprenal.00565.2007. [DOI] [PubMed] [Google Scholar]
  • 8.Andres PG, Friedman LS. Epidemiology and the natural course of inflammatory bowel disease. Gastroenterol Clin North Am. 1999;28:255–281. vii. doi: 10.1016/s0889-8553(05)70056-x. [DOI] [PubMed] [Google Scholar]
  • 9.Sands BE, Arsenault JE, Rosen MJ, et al. Risk of early surgery for Crohn’s disease: implications for early treatment strategies. Am J Gastroenterol. 2003;98:2712–2718. doi: 10.1111/j.1572-0241.2003.08674.x. [DOI] [PubMed] [Google Scholar]
  • 10.Louis E, Collard A, Oger AF, et al. Behaviour of Crohn’s disease according to the Vienna classification: changing pattern over the course of the disease. Gut. 2001;49:777–782. doi: 10.1136/gut.49.6.777. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Powell DW, Mifflin RC, Valentich JD, et al. Myofibroblasts. II. Intestinal subepithelial myofibroblasts. Am J Physiol. 1999;277:C183–201. doi: 10.1152/ajpcell.1999.277.2.C183. [DOI] [PubMed] [Google Scholar]
  • 12.Thannickal VJ, Lee DY, White ES, et al. Myofibroblast differentiation by transforming growth factor-beta1 is dependent on cell adhesion and integrin signaling via focal adhesion kinase. J Biol Chem. 2003;278:12384–12389. doi: 10.1074/jbc.M208544200. [DOI] [PubMed] [Google Scholar]
  • 13.Huang EH, Johnson LA, Eaton K, et al. Atorvastatin induces apoptosis in vitro and slows growth of tumor xenografts but not polyp formation in MIN mice. Dig Dis Sci. 2010;55:3086–3094. doi: 10.1007/s10620-010-1157-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(−Delta Delta C(T)) Method. Methods. 2001;25:402–408. doi: 10.1006/meth.2001.1262. [DOI] [PubMed] [Google Scholar]
  • 15.Kim K, Johnson LA, Jia C, et al. Noninvasive ultrasound elasticity imaging (UEI) of Crohn’s disease: animal model. Ultrasound Med Biol. 2008;34:902–912. doi: 10.1016/j.ultrasmedbio.2007.11.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Higgins PD, Johnson LA, Luther J, et al. Prior Helicobacter pylori infection ameliorates Salmonella typhimurium-induced colitis: Mucosal crosstalk between stomach and distal intestine. Inflamm Bowel Dis. 2010 doi: 10.1002/ibd.21489. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Kamath PS, Wiesner RH, Malinchoc M, et al. A model to predict survival in patients with end-stage liver disease. Hepatology. 2001;33:464–470. doi: 10.1053/jhep.2001.22172. [DOI] [PubMed] [Google Scholar]
  • 18.Semler DE, Chengelis CP, Radzialowski FM. The effects of chronic ingestion of spironolactone on serum thyrotropin and thyroid hormones in the male rat. Toxicol Appl Pharmacol. 1989;98:263–268. doi: 10.1016/0041-008x(89)90231-7. [DOI] [PubMed] [Google Scholar]
  • 19.Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
  • 20.Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–619. doi: 10.1016/0895-4356(92)90133-8. [DOI] [PubMed] [Google Scholar]
  • 21.Ochs HR, Greenblatt DJ, Bodem G, et al. Spironolactone. Am Heart J. 1978;96:389–400. doi: 10.1016/0002-8703(78)90052-2. [DOI] [PubMed] [Google Scholar]
  • 22.Overdiek HW, Hermens WA, Merkus FW. New insights into the pharmacokinetics of spironolactone. Clin Pharmacol Ther. 1985;38:469–474. doi: 10.1038/clpt.1985.206. [DOI] [PubMed] [Google Scholar]
  • 23.Sungaila I, Bartle WR, Walker SE, et al. Spironolactone pharmacokinetics and pharmacodynamics in patients with cirrhotic ascites. Gastroenterology. 1992;102:1680–1685. doi: 10.1016/0016-5085(92)91730-r. [DOI] [PubMed] [Google Scholar]
  • 24.Patel SM, Stashefsky E, Maroun MC, et al. The impact of angiotensin-converting enzyme inhibitors and angiotensin-receptor blockers in patients with Clostridium difficile infection. Med Hypotheses. 2011;76:813–815. doi: 10.1016/j.mehy.2011.02.025. [DOI] [PubMed] [Google Scholar]
  • 25.Carter GP, Rood JI, Lyras D. The role of toxin A and toxin B in Clostridium difficile-associated disease: Past and present perspectives. Gut Microbes. 2010;1:58–64. doi: 10.4161/gmic.1.1.10768. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Lyras D, O’Connor JR, Howarth PM, et al. Toxin B is essential for virulence of Clostridium difficile. Nature. 2009;458:1176–1179. doi: 10.1038/nature07822. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Lyerly DM, Saum KE, MacDonald DK, et al. Effects of Clostridium difficile toxins given intragastrically to animals. Infect Immun. 1985;47:349–352. doi: 10.1128/iai.47.2.349-352.1985. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Kuehne SA, Cartman ST, Heap JT, et al. The role of toxin A and toxin B in Clostridium difficile infection. Nature. 2010;467:711–713. doi: 10.1038/nature09397. [DOI] [PubMed] [Google Scholar]
  • 29.Witteck A, Yao Y, Fechir M, et al. Rho protein-mediated changes in the structure of the actin cytoskeleton regulate human inducible NO synthase gene expression. Exp Cell Res. 2003;287:106–115. doi: 10.1016/s0014-4827(03)00129-0. [DOI] [PubMed] [Google Scholar]
  • 30.Chun TY, Bloem LJ, Pratt JH. Aldosterone inhibits inducible nitric oxide synthase in neonatal rat cardiomyocytes. Endocrinology. 2003;144:1712–1717. doi: 10.1210/en.2002-220956. [DOI] [PubMed] [Google Scholar]
  • 31.Ibrahim MA, Ashour OM, Ibrahim YF, et al. Angiotensin-converting enzyme inhibition and angiotensin AT(1)-receptor antagonism equally improve doxorubicin-induced cardiotoxicity and nephrotoxicity. Pharmacol Res. 2009;60:373–381. doi: 10.1016/j.phrs.2009.05.007. [DOI] [PubMed] [Google Scholar]
  • 32.Fan Q, Liao J, Kobayashi M, et al. Candesartan reduced advanced glycation end-products accumulation and diminished nitro-oxidative stress in type 2 diabetic KK/Ta mice. Nephrol Dial Transplant. 2004;19:3012–3020. doi: 10.1093/ndt/gfh499. [DOI] [PubMed] [Google Scholar]
  • 33.Tian B, Liu J, Bitterman P, et al. Angiotensin II modulates nitric oxide-induced cardiac fibroblast apoptosis by activation of AKT/PKB. Am J Physiol Heart Circ Physiol. 2003;285:H1105–1112. doi: 10.1152/ajpheart.01139.2002. [DOI] [PubMed] [Google Scholar]
  • 34.Kyne L, Hamel MB, Polavaram R, et al. Health care costs and mortality associated with nosocomial diarrhea due to Clostridium difficile. Clin Infect Dis. 2002;34:346–353. doi: 10.1086/338260. [DOI] [PubMed] [Google Scholar]
  • 35.Jarvis WR, Schlosser J, Jarvis AA, et al. National point prevalence of Clostridium difficile in US health care facility inpatients, 2008. Am J Infect Control. 2009;37:263–270. doi: 10.1016/j.ajic.2009.01.001. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
NIHMS329727-supplement-Supplementary_Data.doc (55.5KB, doc)

RESOURCES