Antihyperlipidemic drugs mitigate the elevated incidence of peptic ulcer disease caused by A cohort study

Authors: Pei-Hsien Chen, Chiu-Lin Tsai, Yow-Wen Hsieh, Der-Yang Cho, Fuu-Jen Tsai, Cheng-Li Lin, Hsien-Yin Liao

Abstract

Background:

Several risk factors for peptic ulcer disease (PUD) have been identified; however, the recurrence rate of PUD remains high even with standard ulcer treatments. High cholesterol levels have been proposed as a risk factor for PUD, but clinical evidence remains limited. Therefore, this database study investigated whether hyperlipidemia increases PUD risk and whether antihyperlipidemic drugs reduce this risk.

Methods:

A long-term cohort design was adopted, and Taiwan’s National Health Insurance Research Database was used to enroll patients diagnosed with hyperlipidemia between 2000 and 2016. Patients without hyperlipidemia were randomly matched based on variables such as age and gender to establish a comparison cohort at a 1 ratio. Another cohort study was conducted to determine whether antihyperlipidemic drugs or red yeast rice prescriptions can reduce the incidence of PUD in patients with hyperlipidemia.

Results:

The overall incidence of PUD was 1.48 times higher in the hyperlipidemia cohort (203,235 patients) than in the nonhyperlipidemia cohort (adjusted hazard ratio, 1.48; 95% CI, 1.46-1.50; p < 0.001). Among the patients with hyperlipidemia, those who used antihyperlipidemic drugs with or without red yeast rice prescriptions exhibited a lower risk of developing PUD relative to those who did not use them; the adjusted hazard ratios were 0.33 (95% CI, 0.21-0.52) and 0.81 (95% CI, 0.78-0.84), respectively. When the cumulative exposure to antihyperlipidemic drugs and red yeast rice prescriptions increased, the risk of developing PUD showed a decreasing trend, which was statistically significant for antihyperlipidemic drugs but not for red yeast rice.

Conclusion:

Hyperlipidemia is associated with a higher risk of PUD, which can be reduced through the administration of antihyperlipidemic drugs with or without red yeast rice prescriptions.

1. INTRODUCTION

Several risk factors are associated with peptic ulcer disease (PUD); however, recurrence rates of PUD remain high even with appropriate treatment. A 5-year study of patients with Helicobacter pylori–negative and noneradicated H. pylori revealed PUD recurrence rates of 36.4% and 43.8%, respectively.^1^ Common risk factors for PUD include H. pylori infection, the use of acetylsalicylic acid, and the use of nonsteroidal anti-inflammatory drugs (NSAIDs).^2^ Each risk factor induces PUD through distinct mechanisms. H. pylori produces urease to create an alkaline environment, allowing it to survive in the stomach under the mucosal barrier;^3^ the organism also expresses several adhesins, which help it to attach to the gastric mucosa.^4^ Acetylsalicylic acid may induce gastric mucosal damage by altering mucosal microvessels in an acid-dependent manner.^5^ NSAIDs cause PUD primarily by systemically inhibiting the expression of cyclooxygenase-1-derived prostaglandins, thereby damaging the gastroduodenal mucosa.^6^ Although treatments for PUD are available, the prevention of ulcer recurrence has yet to be fully achieved.^2^ A prospective study highlighted that PUD was idiopathic in one-fifth of the cases it analyzed.^7^ Therefore, identifying other risk factors for PUD and the corresponding treatments is necessary.

High cholesterol levels may be associated with an increased risk of PUD, but the clinical evidence for this association is weak. Two studies have indicated that cholesterol can impair H. pylori eradication therapy.^8,9^ H. pylori converts cholesterol into cholesteryl 6′-O-acyl-α-D-glucopyranoside and subsequently delivers this compound to host epithelial cells. After translocation, cholesteryl 6′-O-acyl-α-D-glucopyranoside can enhance the adhesion of H. pylori to the gastric epithelium.^10^ The modification of cholesterol level enhances the ability of H. pylori to defend against antibiotics, T-cell activation, phagocytosis, and bacterial clearance.^10–13^ These mechanisms suggest that hyperlipidemia causes PUD. However, to the best of our knowledge, no studies have reported clinical evidence indicating that high cholesterol leads to PUD. Only two database studies have reported an indirect relationship between statin use and PUD,^14,15^ and no big data analysis has directly verified the role of hyperlipidemia in causing PUD.

Therefore, we conducted a database study to investigate whether hyperlipidemia increases the risk of PUD and whether antihyperlipidemic drugs and red yeast rice prescriptions (LipoCol Forte^®^, manufactured by Chuang Song Zong Pharmaceutical Co., Ltd., Pingtung County, Taiwan; abbreviated as RYR hereinafter), reduce the risk of PUD.

2. METHODS

2.1. Data source

Taiwan’s National Health Insurance (NHI) program was launched in 1995, and it now provides coverage to over 99.5 % of the population. The National Health Insurance Research Database (NHIRD) is a long-term claims database derived from the NHI program. The database contains patient information such as sex, date of birth, date of clinic visits, drug prescriptions, and diagnostic codes assigned in accordance with the International Classification of Diseases, 9th and 10th Revisions, Clinical Modification (ICD-9-CM and ICD-10-CM). In our study, we used ICD-9 and ICD-10 codes to identify the diseases that patients suffer from, and analyzed the medications they use through drug codes, namely anatomical therapeutic chemical (ATC) codes. We represented ICD codes using integers, regardless of whether they are whole numbers or include decimals, with the exception of those related to H. pylori infection and chronic liver disease, which remain with decimals. Therefore, informed patient consent was not required to access the NHIRD.

2.2. Study cohorts

The present study was approved by the Central Regional Research Ethics Committee of China Medical University, Taichung, Taiwan (CMUH109-REC2-031(CR-4)). From the NHIRD, we obtained a random sample of 2 million beneficiaries (approximately 8.6% of Taiwan’s population). This cohort comprised patients newly diagnosed with hyperlipidemia (ICD-9-CM: 272, ICD-10-CM: E78) between 2000 and 2016, with the index date being the date of first diagnosis of hyperlipidemia. During this period, patients without hyperlipidemia were matched with the hyperlipidemia cohort at a 1 ratio on the basis of sex, age (5-year range), and index year to form the nonhyperlipidemia cohort. We excluded patients who were diagnosed having PUD (ICD-9-CM: 531-535, ICD-10-CM: K25-K29) or H. pylori infection (ICD-9-CM: 041.86, ICD-10-CM: B96.81) before the index date, patients who took proton pump inhibitors (PPI) before PUD diagnosis, patients aged <20 years, and patients who withdrew from the NHI program before PUD diagnosis. Patients infected with H. pylori were excluded prior to the index date, as H. pylori may induce PUD and its treatment often includes PPI.

The primary endpoint was a diagnosis of PUD (ICD-9-CM: 531-535, ICD-10-CM: K25-K29) with the concurrent use of PPI (ATC A02BC). All included patients were observed from the index date until the diagnosis of PUD, the occurrence of death, or the end of the study (year 2017). Furthermore, the patients were stratified according to age, sex, comorbidity and medication, and the relative risk of PUD in the hyperlipidemia cohort compared with the nonhyperlipidemia cohort was analyzed using Cox models.

The second cohort study, extending from the first cohort study, was brought out to compare whether the use of antihyperlipidemic drugs and/or RYR could lower the incidence of PUD in the hyperlipidemia patients. This treatment cohort analysis used the hyperlipidemia patient group which was used in the cohort above. Hyperlipidemia patients who were prescribed antihyperlipidemic drugs—including statins, fibrates, bile acid sequestrants, nicotinic acid derivatives, and other lipid-modifying agents—and/or RYR, accumulating more than 30 days of prescription, were matched with patients who used neither antihyperlipidemic drugs nor RYR. The antihyperlipidemic drugs are classified under ATC codes C10A and C10B, with RYR classified as A047152. The primary endpoint was a diagnosis of PUD with the concurrent use of PPI. The exposure of medicine during the first diagnosis of hyperlipidemia to the diagnosis of PUD was also calculated.

We conducted a literature review based on previous studies^14,15^ to identify comorbidities associated with PUD for further analysis. The selected comorbidities esophageal reflux (ICD-9-CM: 530; ICD-10-CM: K21 and K22), hypertension (ICD-9-CM: 401; ICD-10-CM: I10), ischemic heart disease (ICD-9-CM: 410-413; ICD-10-CM: I20-I22, I24), chronic liver disease (ICD-9-CM: 571.4-571.9; ICD-10-CM: K72-K74), diabetes mellitus (ICD-9-CM: 250; ICD-10-CM: E10-E14), stroke (ICD-9-CM: 430-438; ICD-10-CM: I60-I69), chronic obstructive pulmonary disease (ICD-9-CM: 491, 492, 494, and 496; ICD-10-CM: J40-J44), and end-stage renal disease (ICD-9-CM: 585 and 586; ICD-10-CM: N18 and N19).

2.3. Statistical analysis

Basic demographic data are presented as numbers (percentages) for categorical variables and means ± SDs for continuous variables. A chi-squared test was used to compare demographic information, comorbidities, and medication use of the hyperlipidemia and nonhyperlipidemia cohorts at baseline. Univariable and multivariable Cox proportional-hazards models were used to analyze the risk of PUD, which is presented using hazard ratios (HRs) and 95% CIs. The relationships between cumulative antihyperlipidemic drug and RYR exposure with PUD incidence were also analyzed. The Kaplan–Meier method was used to calculate the cumulative incidence of PUD; the cumulative incidence results of the two study cohorts were then compared using the log-rank test. A p-value of <0.05 was considered a statistically significant result. Statistical analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC); we used the survfit package of R to construct Cox proportional-hazards models.

3. RESULTS

3.1. Higher incidence of PUD among patients with hyperlipidemia than among those without hyperlipidemia

No significant difference was detected between patients with and without hyperlipidemia in terms of their mean age, sex, baseline comorbidities, and medication use (Table 1). Table 2 presents the results pertaining to the relationships between the patients’ age, sex, comorbidities, and medication use; it also indicates the incidence rate, crude HR, and adjusted HR for PUD in the hyperlipidemia cohort. After adjusting the multivariable Cox model was adjusted, 47 954 of the 203 235 patients with hyperlipidemia were identified as having developed PUD, yielding an incidence rate of 27.7 per 1000 person-years. Among the matched 203 235 patients without hyperlipidemia, 30 500 developed PUD, yielding an incidence rate of 18.3 per 1000 person-years. The hyperlipidemia cohort had a higher adjusted HR (1.48; 95% CI, 1.46-1.5) for PUD relative to the nonhyperlipidemia cohort. The Kaplan–Meier model revealed that the cumulative incidence of PUD increased with time in both cohorts (Fig. 1) but was consistently higher in the hyperlipidemia cohort than in the nonhyperlipidemia cohort (log-rank test, p < 0.001). Table 2 also indicates a significantly higher incidence of PUD among older patients (40-65 vs >65 adjusted HR, 1.51 vs 2.43; 95% CI, 1.48-1.53 vs 2.38-2.49) and among male patients (adjusted HR, 1.13; 95% CI, 1.12-1.15). The following comorbidities were determined to be associated with an increased adjusted HR for PUD: esophageal reflux, hypertension, ischemic heart disease, chronic liver disease, diabetes mellitus, stroke, chronic obstructive pulmonary disease, and end-stage renal disease. Regardless of the presence of comorbidities, however, the hyperlipidemia cohort exhibited a higher adjusted HR for PUD relative to the nonhyperlipidemia cohort (Table 3). Medications including NSAID (adjusted HR, 1.06) and aspirin (adjusted HR, 1.04) showed a higher adjusted HR for PUD while clopidogrel (adjusted HR, 0.98) did not (Table 2).

3.2. Reduction of PUD incidence through the use of antihyperlipidemic drugs

In the treatment cohort analysis, as illustrated in Table 4, we examined how the use of antihyperlipidemic drugs or RYR contributed to the prevention of PUD in the hyperlipidemia cohort. At baseline, no significant difference was observed between patients with hyperlipidemia who took antihyperlipidemic drugs, RYR, or both, and those who did not (Table 4). Compared with patients who did not use antihyperlipidemic drugs or RYR, those who used antihyperlipidemic drugs (adjusted HR, 0.81), or both (adjusted HR, 0.33) exhibited a significantly lower adjusted HR for PUD, while those who used RYR (adjusted HR, 0.92) did not show a significant difference (Table 5). Comorbidities except for hypertension was associated with higher HR for PUD. Clopidogrel (adjusted HR, 1.27) and aspirin (adjusted HR, 1.03) was associated with higher adjusted HR for PUD, while NSAIDs (adjusted HR, 0.92) were not.

3.3. Reduction in HR of newly diagnosed PUD through longer exposure to antihyperlipidemic drugs

Table 6 presents the relationships of PUD with cumulative antihyperlipidemic drugs and RYR exposure. Longer antihyperlipidemic drugs exposure was associated with a lower incidence rate of PUD. When antihyperlipidemic drugs exposure increased from ≤60 days (adjusted HR, 1.38; CI, 1.32-1.45) to 61 to 182 days (adjusted HR, 1.16; CI, 1.10-1.23), 183 to 462 days (adjusted HR, 0.81; CI, 0.77-0.86), and >462 days (adjusted HR, 0.35; CI, 0.33-0.37), the adjusted HR exhibited a significant decreasing trend. Among the patients who used RYR, it showed a decreasing trend for HR in ≤28 days (adjusted HR, 0.81; CI, 0.41-1.63), 29 to 50 days (adjusted HR, 0.88; CI, 0.37-2.12), and >105 days (adjusted HR, 0.60; CI, 0.25-1.43), but did not reach statistical significance. A similar trend is presented in Fig. 2.

4. DISCUSSION

To the best of our knowledge, this is the first study to directly investigate the relationship between hyperlipidemia and PUD. Our first main finding is that the hyperlipidemia cohort had a higher incidence of developing PUD relative to the nonhyperlipidemia cohort. Compared with the 203,235 matched patients without hyperlipidemia, the 203,235 patients with hyperlipidemia exhibited a higher risk of PUD (30,500 vs 47,954 patients; adjusted HR, 1.48; Table 2). Fig. 1 also reveals that, relative to the patients without hyperlipidemia, those with hyperlipidemia exhibited a higher overall incidence of PUD (Kaplan–Meier model, log-rank test, p < 0.001). Even after other risk factors that promote PUD (e.g., aspirin, NSAIDs, clopidogrel, and related comorbidities) were excluded (Table 3), hyperlipidemia was still the main factor affecting the incidence of PUD.

The second main finding is that in the hyperlipidemia cohort, the use of antihyperlipidemic drugs reduced the risk of PUD. In the hyperlipidemia cohort, those who used antihyperlipidemic drugs exhibited a lower risk of being newly diagnosed PUD relative to those who did not use antihyperlipidemic drugs or RYR (Table 5, adjusted HR, 0.81). Longer exposure to antihyperlipidemic drugs was linked to reduced PUD incidence, exhibiting a significant decreasing trend (Table 6; ≤60 days; adjusted HR, 1.38; CI, 1.32-1.45; 61-182 days; adjusted HR, 1.16; CI, 1.10-1.23; 183-462 days; adjusted HR, 0.81; CI, 0.77-0.86; >462 days; adjusted HR, 0.35; CI, 0.33-0.37). Studies have identified statins as 3-hydroxy-3-methyl-glutarylcoenzyme A reductase inhibitors, and statins are the main class of drugs used to reduce the cholesterol levels of patients with hyperlipidemia. In addition to reducing blood lipid levels, statins have been reported to have gastroprotective^16,17^ and antimicrobial effects.^9,18^ Statin can promote autophagy in macrophages toward H. pylori, increase gastric mucosal NO and PGE2 levels, reduce oxidative stress, maintain vascular integrity, and stimulate gastric juice mucin.^16–19^ The second most commonly used lipid-lowering drug, the fibrate class, has also demonstrated properties in animal experiments in promoting the healing of gastric ulcers and preventing their development.^20,21^ These mechanisms support our finding that antihyperlipidemic drugs can reduce the risk of developing PUD.

The third main finding is that the combined use of RYR and antihyperlipidemic drugs significantly reduces the incidence of PUD by 67% among the hyperlipidemia cohort, as shown in Table 5. This observation has not been reported in other studies. Fig. 2 also shows a decreasing trend in the incidence of PUD among patients using RYR and/or antihyperlipidemic drugs. RYR, derived from Monascus purpureus and used in traditional Chinese medicine has been shown to reduce total cholesterol, low-density lipoprotein, and triglyceride levels.^22–28^ Moreover, our analysis reveals that using RYR alone as a treatment was associated with a lower incidence of PUD, though it did not reach statistical significance (adjusted HR: 0.92, p = 0.63). Patients who took RYR for various durations (≤28, 29-50, and >105 days) also exhibited a lower adjusted HR relative to those who did not. However, these differences were not statistically significant, likely due to the small sample size of RYR users (RYR non-treated = 45,597; Table 6). The limited use of RYR can be attributed to its coverage under NHI only starting in 2010, along with the prevailing preference for Western medicine as the primary treatment for hyperlipidemia in our cohort from 2000 to 2016.

A strength of the present study is its methodological quality. Few epidemiologic studies or clinical trials have examined the incidence of PUD in patients with hyperlipidemia. To the best of our knowledge, only two studies^14,15^ have used the NHIRD to analyze the relationship between hyperlipidemia and PUD. The first study^14^ examined 48 562 patients with hyperlipidemia and reported that patients with hyperlipidemia who took statins exhibited a lower risk of PUD relative to a control group who did not take statins. The second study^15^ adopted a case–control design to examine the data of 35 194 patients with PUD, and reported that using statins reduced the incidence of PUD. Our study improved upon previous NHIRD studies by updating the inclusion period, which resulted in a larger sample size of 406 470. Furthermore, the two aforementioned studies did not involve any cohort analysis to determine whether hyperlipidemia leads to PUD. In the present study, we applied a sound methodology for achieving our objective; we performed a cohort analysis to verify that hyperlipidemia leads to PUD and then conducted a case–control study to determine whether antihyperlipidemic drugs can reduce the risk of PUD.

Nevertheless, the present study has several limitations. First, the NHIRD does not include data on variables such as occupation, body mass index, smoking, alcohol consumption, and environmental exposure, which are possible confounders. We applied 1 matching to minimize the selection bias between the two groups examined in the present study. Second, clinical variables, such as serum laboratory data, images, and pathological findings, are not available in the NHIRD. Third, we defined PUD as ulcers in the stomach and duodenum, excluding those in the esophagus.^2^ However, some studies propose that PUD also includes esophageal ulcers.^29^ This may affect the size of the endpoint patient population. Fourth, hyperlipidemia patients may have increased doctor visits and may receive more intensive treatment and more correlated examination if their condition is more severe. This may affect the incidence of PUD which we cannot exclude. This may also explain why the adjusted hazard ratio of PUD, caused by aspirin, NSAID, and clopidogrel, did not increase substantially as expected. Finally, because only data on physician prescriptions were available, we could not verify the included patients’ medication adherence; such information could reveal the true duration of drug exposure.

In conclusion, our findings suggest that people with hyperlipidemia have an increased incidence of developing PUD, which can be reduced by the administration of antihyperlipidemic drugs with or without RYR.

ACKNOWLEDGMENTS

The authors thank Health Data Science Center, China Medical University Hospital, for providing administrative, technical and funding support. This manuscript was edited by Wallace Academic Editing. The authors specially thank Ms. Yu-Cih Yang and Ms. Mei-Chen Lin for their helps in the data collection.