Beneficial effect of statins on total mortality in abdominal aortic aneurysm (AAA) repair

Abstract

In this single center, retrospective cohort study we wished to compare early and total mortality for all patients treated for abdominal aortic aneurysms (AAA) with open surgery who were taking statins compared to those who were not. A cohort of 640 patients with AAA was treated with open surgery between 1999 and 2012. Patients were consecutively recruited from a source population of 390,000; 21.3% were female, and the median age was 73 years. The median follow-up was 3.93 years, with an interquartile range of 1.79–6.58 years. The total follow-up was 2855 patient-years. An explanatory strategy was used. The propensity score (PS) was implemented to control for selection bias and confounders. The crude effect of statin use showed a 78% reduction of the 30-day mortality. A stratified analysis using the Mantel–Haenszel method on quintiles of the PS gave an adjusted effect of the odds ratio equal to 0.43 (95% CI: 0.18–0.96), indicating a 57% reduction of the 30-day mortality for statin users. The adjusted rate ratio was 0.62 (95% CI: 0.45–0.83), indicating a reduction of long-term mortality of 38% for statin users compared to non-users for a median follow-up of 3.93 years. This retrospective cohort study showed a significant beneficial effect of statin use on early and long-term survival for patients treated with open surgery. To be conclusive, our results need to be replicated by a randomized clinical trial.

Keywords

abdominal aortic aneurysm (AAA)mortality statins surgical repair

Introduction

Statin therapy is associated with improved survival after open surgery and endovascular aortic repair (EVAR) for patients with abdominal aortic aneurysms (AAA). Lipid-modifying medication (LMM) is recommended for all patients who have coronary and non-coronary arteriosclerotic disease. In this single center, retrospective study we wished to compare early and total mortality for all patients treated for AAA who were taking statins compared to those patients who were not. Patients treated with EVAR were excluded from this study in order to decrease the known beneficial effect of EVAR on perioperative survival.

Materials and methods

A cohort of 640 patients with AAA was operated on between 20 July 1999 and 15 July 2012, with a closing date of 15 July 2012. Patients were consecutively recruited from a source population of 390,000; 21.3% (137) were female, and the median age was 73 years. A total of 23.4% (150) of the patients had emergency ruptured AAA (rAAA), verified by computed tomography angiography (CTA).

Early and total mortality were evaluated. The median follow-up was 3.93 years, with an interquartile range of 1.79–6.58 years; total follow-up was 2855 patient-years. An explanatory strategy was used. We used the propensity score (PS) to control for selection bias and confounders.

Epidemiological design and statistical methods

This was an exposed (statin user) versus a non-exposed cohort with the outcomes of early mortality and total mortality with long-term survival. The crude effect of statin use on the outcome of early (30-day) mortality was estimated by the odds ratio (OR) and its 95% confidence interval (CI). For total mortality, we used the incidence rate ratio (IRR) and its 95% CI using the patient-years model.¹ In addition, survival curves using the Kaplan–Meier method and the log-rank test were implemented to compare equality of the curves.¹

The adjusted effect of statin use on outcomes was done using stratification of the PS into quintiles using the Mantel–Haenszel method.²

This was complemented by use of the logistic model by controlling the PS as a continuous variable for the 30-day mortality end point. For the total mortality end point, we complemented by using the Cox model and estimating the hazard ratio (HR) of statin use and stratifying the Cox model into quintiles of the PS.

The PS is the conditional probability to use statins given the individual covariates, adjusting for observed use of statins with the probability of non-use (propensity) creates a quasi-randomized experiment. Stratification of the PS into quintiles will remove 90% of measured selection bias due to the difference in baseline covariates between a user and non-user of statins.^3,4

Our model is in agreement with the limitations by Peduzzi et al., which pinpoints the need to have no less than 10 events per covariate, especially for the outcome of total mortality.⁵

As we are in a prognostic strategy in the PS estimation analysis, the predictive accuracy of the model was evaluated by calibration and discrimination. Calibration was evaluated by the Hosmer–Lemeshow (H-L) goodness-of-fit test. Discrimination was evaluated by the analysis of the area under the ROC (receiver operating characteristic) curve. If the area under the curve (AUC) is greater than 0.7, it can be concluded that the model has an acceptable discriminatory capability. The reporting and analysis of our study followed the suggestions for improvement of Weitzen et al.⁶ when combining classical methods and the PS to remove 90% of measured selection bias due to baseline differences using quintiles stratification.³ Unlike randomization, PS techniques cannot control for unknown or unmeasured potential confounding factors.

Results

Generation of the propensity score (PS)

Because of a difference in distribution of covariates between statin users and non-statin users (Table 1), we constructed a PS. We generated the PS for probability of statin use by the logistic model considering 13 variables: diabetes mellitus, American Society of Anesthesiology (ASA) physical status classification system group, beta-blockers, hypertension, previous myocardial infarction (MI), atrial fibrillation, chronic obstructive pulmonary disease (COPD), renal failure, smoker, sex, age, diameter of AAA, and emergency rAAA.

Table 1.

Distribution of clinical profile between users and non-users of statins.

Variables	Statin (+)n=369	Percent	Statin (–)n=271	Percent	p-value
Female sex	83	22.5	54	20.0	0.4340
Age over 70 years	218	59.0	195	72.0	0.0010
Smoking	174	47.0	110	40.6	0.0990
Type 2 diabetes	41	11.0	12	4.40	0.0020
COPD	44	12.0	13	5.0	0.0390
Aspirin	238	65.0	82	30.0	0.0001
Beta-blockers	290	79.0	112	41.0	0.0001
AF	41	11.0	31	11.4	0.9330
Angina	86	23.0	52	19.0	0.2110
Previous MI	147	40.0	61	22.5	0.0001
Bad ASA^a	42	11.4	75	28.0	0.0001
Hypertension	306	83.0	151	56.0	0.0001
Aneurysm diameter >72 mm^b	64	17.0	106	39.0	0.0001
Emergency rAAA	54	15.0	96	35.4	0.0001
Renal failure	33	9.0	21	8.0	0.5910
Inflammatory AAA	36	10.0	30	11.0	0.5890

COPD, chronic obstructive pulmonary disease; AF, atrial fibrillation; MI, myocardial infarction; ASA, American Society of Anesthesiology physical status classification system; rAAA, ruptured abdominal aortic aneurysm.

Poor ASA score (level ASA score 4+5).

72 mm is the 75% percentile of the aneurysm diameter.

The generated model was satisfactory as far as discrimination of the area under the ROC curves, with a value of 0.8082, which was markedly over the 0.70 borderline (Figure 1).

Figure 1.

ROC curve of the discrimination of the propensity score.

Also, the calibration of the model was satisfactory with the H-L goodness-of-fit test and had a p=0.9946 indicating a goodness-of-fit for the model data.

Outcome: Early 30-day mortality

The crude effect of statin use showed a 78% reduction of 30-day mortality (Table 2) for patients using statins compared to non-users. A stratification analysis using the Mantel–Haenszel method on quintiles of the PS gave an adjusted effect on the 13 variables included in the PS (adjusted odds ratio (ORa) = 0.43, 95% CI: 0.18–0.96), indicating a 57% reduction of the 30-day mortality for statin users versus non-statin users. There was homogeneity of the effect in the five strata (the test of homogeneity had a p-value = 0.2659).

Table 2.

End-point: 30-day mortality.

Crude effect

30-day mortality	Statin (+)	Statin (–)	Total
Yes	10	29	39
No	359	242	601
	369	271	640

OR=0.23 (0.11–0.47), p=0.0001.

Logistic model

OR=0.23

0.11–0.48

Adjusted effect

Estimated by stratification of the PS into quintiles using the Mantel–Haenszel method.

Adjusted OR (ORa)

ORa=0.43

0.18–0.96

Test of homogeneity over five strata: p=0.3919.

Logistic model adjusted for PS score

OR=0.41

0.04–0.91

OR, odds ratio; ORa, adjusted odds ratio; PS, propensity score.

A complementary analysis using the logistic model gave a very close result (OR=0.41, 95% CI: 0.04–0.91) when adjusting for the PS in the model (Table 2).

Outcome: Total mortality

The survival curves were markedly different between statin users and non-statin users (Figure 2). This difference was highly significant (p=0.0001) according to the log-rank test.

Figure 2.

Survival of statin users versus non-users.

The crude effect of statin use on total mortality showed an IRR=0.56 (95% CI: 0.42–0.73; p=0.0001). An adjusted effect using the person-years stratification of the PS into quintiles using the Mantel–Haenszel method is shown in Table 3.

Table 3.

End-point: Total mortality.

Crude effect

Total mortality	Statin (+)	Statin (–)	Total
Yes	90	135	225
Patient-years	1552	1303	2855

IRR=0.56 (0.42–0.73), p=0.00001.

Univariate Cox model

HR=0.55

0.43–0.73

Adjusted effect

Estimated by stratification of the PS into quintiles using the Mantel–Haenszel method.

Adjusted IRR on PS

IRR=0.62

0.45–0.83

Test of homogeneity over five strata, p=0.3987.

Cox model stratified on quintiles of the PS score

HR=0.60

0.44–0.82

IRR, incidence rate ratio; HR, hazard ratio; PS, propensity score.

The results showed an adjusted rate ratio using the Mantel–Haenszel of 0.62 (95% CI: 0.45–0.83). This indicates a reduction of the long-term mortality of 38% for statin users compared to non-statin users. This result was adjusted for the 13 variables when generating the PS. The test of homogeneity had a p-value of 0.3987, which indicates homogeneity of the effect on statin use in the five strata of the PS.

A complementary analysis was done using the Cox model to stratify the PS into quintiles, and this gave close results (HR=0.60, 95% CI: 0.44–0.82).

Discussion

AAA patients were not optimally treated with the following prophylactic medications: beta-blockers, platelet inhibitors, angiotensin converting enzyme (ACE) and statins. In particular, where 75% of the patients should have been taking statins, only 41% were.⁷ In a Danish study, AAA repair patients had an annual MI incidence of 2.5% (HR=2.1%; 95% CI: 1.9–2.2) compared to the general population. Stroke had an annual incidence of 2.9% (HR=1.8; 95% CI: 1.6–1.9). There was also a 2.4-fold 95% CI increase (2.3–2.4) in the HR of all-cause mortality compared with the general population.⁸

What are the plausible biological mechanisms of the prophylactic protective effect of statin medication? Reduced incidence of MI and stroke is, by some authors attributed to the pleiotropic effect on the vascular wall. It has been shown in a pilot study that statin use had a pleiotropic effect on aortic wall atheroma that both stabilized and reduced the size of the plaque.⁹ When looking at the effect of statin medication on the expansion rate and the risk of rupture in the aortic wall of AAA patients, irrespective of the decrease in serum cholesterol levels, there was also cellular activity. This activity included effects on endothelial cells, smooth muscle cells, platelets, monocytes/macrophages, and the inflammatory process/cascade.¹⁰ Further pleiotropic effects were documented. Statins decrease the levels of matrix metalloproteinase (MMP-9, MMP-3) in the AAA wall. MMPs are implicated in reducing extracellular matrix proteins (ECM) in the AAA wall.¹¹ It has been postulated that an imbalance between MMPs and their inhibitors (TIMPS) is a cause of aneurysmal and atherosclerotic disease.^12,13 In addition, an association between statin therapy and reduced expansion in patients with small AAA has been shown.^14,15 In post deployment of EVAR, it has been documented that statin therapy can be predictive of AAA sac diameter reduction.¹⁶

Going from the cellular level to the clinical setting, there are quite a number of investigative teams that have shown reduced mortality when using statins. Several studies have suggested a beneficial effect of perioperative statin therapy in reducing cardiovascular complications and death directly after vascular surgery.^17,18

The Dream Study group has shown that statin therapy upon enrollment in their study was independently associated with better overall survival after AAA open surgery and EVAR (HR=0.5, 95% CI: 0.3–0.8; p=0.004). Statin therapy was especially associated with fewer cardiovascular deaths (HR=0.4, 95% CI: 0.2–0.9; p=0.025).¹⁹ Statin therapy has been shown to give better perioperative outcomes in several studies with patients treated for AAA.^20–24 Previously, only a few studies have shown a possible beneficial effect of statin therapy on long-term survival after AAA repair, but these were all retrospective.^25–27

Statin use undeniably reduces coronary events and stroke in AAA patients. Our study also demonstrated a protective effect of statin use. This was true both for early (30-day) and long-term survival. We registered statin use at the time of admission: 369 patients were taking statins and 271 were not. Five patients had unknown status. Our elective perioperative work-up includes a cardiologist-reviewed heart duplex study, stress test, and spirometry. Medications are adjusted at this time. In an elective setting, those patients requiring coronary angiography receive either percutaneous coronary intervention (PCI) or a coronary artery bypass graft (CABG), if required.

The PS is quite extensively used in thoracic surgery. Bias and confounding factors are thus reduced. But, it does not control for unmeasured confounders that always exist in an observational design. Here, we chose to use the stratified Cox model and the logistic model to complement our analysis.

We have shown that the adjusted effect for confounders and the selection bias of statin therapy reduces early adjusted 30-day mortality by 57% and total mortality in long-term survival is reduced by 38%.

Conclusion

This retrospective cohort study showed a significant beneficial effect of statin use on early mortality and long-term survival for patients treated for AAA with open surgery. The PS was used to control for major selection bias and confounding factors. To be conclusive, our results need to be replicated by a randomized clinical trial.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Kleinbaum

Klein

Survival Analysis: A Self Learning Text. 2nd ed. New York, NY: Springer Science + Business Media, 2005.

Kleinbaum

Kupper

Morgenstern

Epidemiologic Research: Principles and Quantitative Methods. New York: Van Nostrand Reinhold, 1982.

D’Agostino

Jr.

Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998; 17: 2265–2281.

Rosenbaum

Rubin

DB.

The central role of propensity score in observational studies of causal effect. Biometrica 1983; 70: 41–45.

Peduzzi

Concato

Kemper

et al . A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996; 49: 1373–1379.

Weitzen

Lapane

Toledano

et al . Principles for modeling propensity scores in medical research: A systematic literature review. Pharmacoepidemiol Drug Saf 2004; 13: 841–853.

Lloyd

Newton

Norwood

MGA

et al . Patients with aortic aneurysm: Are we missing the opportunity for cardiovascular risk reduction? J Vasc Surg 2004; 40: 691–697.

Eldrup

Budtz-Lilly

Laustsen

et al . Long-term incidence of myocardial infarct, stroke, and mortality in patients operated on for abdominal aortic aneurysms. J Vasc Surg 2012; 55: 311–317.

Nemoto

Hoshina

Takayama

et al . Statins reduce extensive aortic atheromas in patients with abdominal aortic aneurysms. Ann Vasc Dis 2013; 6: 711–717.

10.

Van Kuijk

Flu

Wiiteveen

et al . The influence of statins on the expansion rate and rupture risk of abdominal aortic aneurysms. J Cardiovasc Surg 2009; 50: 599–609.

11.

Wilson

WRW

Evans

Bell

PRF

et al . HMG-CoA reductase inhibitors (statins) decrease MMP-3 and MMP-9 concentrations in abdominal aortic aneurysms. Eur J Vasc Surg 2005; 30: 259–262.

12.

Freestone

Turner

Coady

et al . Inflammation and matrix metalloproteinases in the enlarging abdominal aortic aneurysm. Arterioscler Thromb Vasc Biol 1995; 15: 1145–1151.

13.

Loftus

Naylor

Goodall

et al . Increased matrix metalloproteinase-9 activity in unstable carotid plaques. A potential role in acute plaque disruption. Stroke 2003; 31: 40–47.

14.

Takagi

Matsui

Umemoto

A meta-analysis of clinical studies of statins for prevention of abdominal aortic aneurysm expansion. J Vasc Surg 2010; 52: 1675–1681.

15.

Takagi

Yamamoto

Iwata

et al . Effects of statin therapy on abdominal aortic aneurysm growth: A meta-analysis and meta-regression of observational comparative studies. Eur J Vasc and Endovasc Surg 2012; 44: 287–292.

16.

Raux

Cochennec

Becquemin

J-P.

Statin therapy is associated with aneurysm sac regression after endovascular aortic repair. J Vasc Surg 2012; 55: 1587–1592.

17.

O’Neil-Callahan

Katsimaglis

Tepper

et al . Statins decrease perioperative cardiac complications in patients undergoing noncardiac vascular surgery: The Statins for Risk Reduction in Surgery (STaRRS) study. J Am Coll Cardiol 2005; 45: 336–342.

18.

Le Manach

Ibanez

Bertrand

et al . Impact of preoperative statin therapy on adverse postoperative outcomes in patients undergoing vascular surgery. Anesthesiology 2011; 114: 98–104.

19.

De Bruin

Baas

Heymans

et al . Statin therapy is associated with improved survival after endovascular and open aneurysm repair. Vasc Surg 2014; 59: 39–44.

20.

Poldermans

Bax

Kertaj

et al . Statins are associated with a reduced incidence of perioperative mortality in patients undergoing major noncardiac vascular surgery. Circulation 2003; 107: 1848–1851.

21.

Durazzo

Machado

Ikeoka

et al . Reduction in cardiovascular events after vascular surgery with atorvastatin: A randomized trial. J Vasc Surg 2004; 39: 967–975.

22.

Kertaj

Boersma

Westerhout

et al . A combination of statins and beta-blockers is independently associated with a reduction in the incidence of perioperative mortality and nonfatal myocardial infarction in patients undergoing abdominal aortic aneurysm surgery. Eur J Vasc Endovasc Surg 2004; 28: 343–352.

23.

Schouten

Boersma

Hoeks

et al .; Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography Study Group. Fluvastatin and perioperative events in patients undergoing vascualar surgery. N Engl J Med 2009; 361: 980–989.

24.

McNally

Agle

Parker

et al . Perioperative statin therapy is associated with improved outcomes and resource utilization in patients undergoing aortic aneurysm repair. J Vasc Surg 2010; 51: 1390–1396.

25.

Leurs

Visser

Laheij

et al . Statin use is associated with reduced all-cause mortality after endovascular abdominal aortic aneurysm repair. Vascular 2006; 14: 1–8.

26.

Ketai

Boersma

Westerhout

et al . Association between long-term statin use and mortality after successful abdominal aortic aneurysm surgery. Am J Med 2004; 116: 96–103.

27.

Diehn

Becher

Karzen

Statins are associated with decreased mortality in abdominal, but not thoracic aortic aneurysm patients undergoing endovascular repair: Propensity score-adjusted analysis. Vasa 2008; 37: 241–249.