Prior appendectomy and cerebral infarction as potential risk factors for recurrent ischemic colitis: a case control study

Background: Ischemic colitis (IC) is a benign disease associated with acute lower gastrointestinal bleeding and frequent recurrence. While several studies have investigated risk factors for IC onset, few have assessed the risk factors for recurrent IC. This study aimed to identify risk factors for recurrent IC. Methods: Potential risk factors for recurrence were assessed by examining medical records and laboratory ndings in this single-center retrospective study. We extracted the following data from the patients’ medical records: patient characteristics, clinical signs and symptoms, laboratory ndings, method of treatment, length of hospital stay, disease course, and the frequency of IC morbidities. Patients with IC were selected from a total of 439,312 patients over an 11-year period. Patients were divided into recurrent and non-recurrent IC groups. Results: In total, 225 patients met the diagnostic criteria for IC during the specied study period; of these, 204 (90.7%) and 21 (9.3%) were included in the non-recurrent and recurrent IC groups, respectively. Univariate and multivariate analyses showed a signicant association between IC recurrence and both cerebral infarction (p = 0.01, odds ratio=5.41) and history of appendectomy (p = 0.0009, odds ratio = 5.19). The median (interquartile range [IQR]) follow-up time for all patients was 1556 (353–2768) days; this was much longer than the median (IQR) time to recurrence of 291 (64–907) days in the recurrent IC group. Conclusions: Prior cerebral infarction and appendectomy are potential risk factors for IC recurrence. However, the pathogenic mechanism of IC remains to be determined.


Background
Ischemic colitis (IC) is one of the most common causes of acute lower gastrointestinal bleeding, and its diagnostic criteria were proposed by Marston et al. in 1966 (1). IC accounts for 8.7%-18.0% of all cases of acute lower gastrointestinal bleeding (2)(3)(4), and according to a previous systematic review, its incidence in the general population ranges from 4.5 to 44 cases per 100,000 person-years (5). IC is generally more common in older women (6)(7)(8)(9)(10)(11)(12)(13)(14) and is characterized by a sudden onset of left lower quadrant pain accompanied by diarrhea and bloody stools. While these symptoms often rapidly resolve with bowel rest, some cases involve irreversible changes that require surgery (e.g., gangrenous colitis, fulminant colitis, stricture formation, and chronic IC) or may be fatal. IC may also be caused by other severe conditions. For example, systemic hypotension due to sepsis, hypokalemia, cardiogenic shock, or third-space uid shift causing intestinal hypoperfusion, thereby progressing to IC. Vascular (e.g., hypertension, dyslipidemia, diabetes, cardiovascular disease, cerebral infarction, and chronic renal failure) and intestinal (e.g., irritable bowel syndrome, constipation, colorectal cancer, history of abdominal surgery, and history of vascular surgery) factors have been reported to be associated with IC onset (6-18). In recent years, the use of drugs, such as contraceptives and vasoconstrictors, has been reported to be a risk factor for IC onset (17,19,20). In addition, diseases such as systemic lupus erythematosus and antiphospholipid syndrome, which cause vasculitis, have also been reported as risk factors for IC (6, [21][22][23]. However, as these diseases can also cause enteritis, care must be taken in formulating the differential diagnosis of IC. IC reportedly recurs at a frequency of 6.8%-16.0% after symptom resolution (24)(25)(26)(27). While several studies have assessed the risk factors for IC onset, few have investigated the potential risk factors for recurrent IC. A previous study reported smoking and abdominal aortic aneurysms (AAAs) as risk factors for recurrent IC; however, its sample size was relatively small (118 patients) (24).
A recent audit of patients treated for IC at our hospital revealed the absence of AAAs. However, many patients had a history of appendectomy. Therefore, we conducted a retrospective study of patients diagnosed with IC at our hospital and investigated the potential risk factors for recurrent IC.

Study design and setting
The study protocol was approved by the appropriate ethics committee (Yokohama City University Certi ed Institutional Review Board) (Approval No: CRB3180007) and was exempted from obtaining consent directly from patients by using an opt-out method. We conducted a case-control study of patients previously diagnosed with IC at our hospital during an 11-year interval between January 1, 2009, and December 31, 2019. Patients diagnosed with IC were selected from a total of 439,312 patients registered in the electronic medical record database during this period. The follow-up period was from the onset of IC to the date of the last visit (up to December 31, 2019). Patients who had a relapse after resolution of the initial IC episode during the follow-up period were included in the recurrent IC group, whereas those who experienced a single episode of IC during the study period were included in the nonrecurrent IC group.

Patient characteristics
We identi ed patients with IC in the database using the following International Classi cation of Diseases-10 codes: K550 (NDNL, GK7L, PR7K), K551 (F1A5, BPJQ, G440), and K559. We counted cases with multiple IC disease classi cations as a single case, and a total of 316 patients with IC were identi ed. The diagnosis of IC was further con rmed based on clinical symptoms and laboratory ndings, as there are no internationally standardized diagnostic criteria for this disease. We used the diagnostic criteria proposed by Marston et al., (1) which are commonly followed in Japan: 1) sudden-onset left lower abdominal pain accompanied by diarrhea and bloody stools, 2) typical ndings on contrast-enhanced computed tomography (CT) (edematous thickening of the intestinal wall and increased peri-intestinal fatty tissue concentration) or on colonoscopy (CS) (ulceration/erosion, redness, edema, biopsy image), predominantly on the left side of the colon, 3) no autoimmune disease or infection, 4) no trauma or mechanical factors (e.g., hernia), 5) no abdominal surgery within the past 6 months, 6) no evidence of severe acute ischemic changes in other organs, and 7) absence of a secondary onset of IC. We excluded patients who were younger than 20 years, pregnant, or opted out of the study; we also excluded suspected cases that could not be de nitively diagnosed, and patients diagnosed with IC at other hospitals, as it was not possible to con rm what diagnostic criteria were used.

Data extraction
We extracted the following data from the patients' medical records: patient characteristics (sex, age, body mass index [BMI], drinking history, smoking history, medical history, comorbidities, medications, and vital signs), clinical signs and symptoms (abdominal pain, diarrhea, bloody stools, and peritoneal irritation symptoms), laboratory ndings (contrast-enhanced CT ndings, endoscopic features, and histological features), method of treatment, length of hospital stay, disease course (time of resolution of abdominal pain, time of resolution of bloody stools, and initiation of meals), and the frequency of IC morbidities.
Contrast-enhanced CT and/or CS was performed to assess each segment of the colon. Contrastenhanced CT was performed to detect poor contrast, bowel wall thickening, mesenteric fat stranding, pericolonic free uid, mesenteric gas, and portal vein gas. CS was performed to assess the presence of mucosal erosion, ulceration, edema, redness, necrosis, and stricture. The frequency of IC morbidities was de ned as the number of times the patient was diagnosed with IC at our hospital.

Statistical analysis
All patient data were entered into a Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA) spreadsheet. The computer containing the data was password-protected and stored in a locked cabinet in the medical o ce. Continuous variables are expressed as mean ± standard deviation, and categorical variables are expressed as numbers and percentages. The chi-square test was performed to compare proportions between the recurrent IC and non-recurrent IC groups, and the t-test was used to compare means. All statistical analyses were performed using JMP Pro 15 (SAS, Cary, NC, USA), and the level of statistical signi cance (p) was set at ≤ 0.05. Two separate multivariate analyses were performed including 1) three items that were previously reported to be signi cantly different (current smoking (24) and two items that were signi cantly different in the univariate analyses) and 2) all items. These three items were also analyzed for the time to IC recurrence using the Kaplan-Meier method.

Results
Of the 316 patients diagnosed with IC from the medical records database, 91 were excluded based on our diagnostic criteria; therefore, we nally included the data of 225 patients in our analyses. The recurrent and non-recurrent IC groups comprised 21 (9.3%) and 204 (90.7%) patients, respectively ( Figure 1).

Characteristics of patients with IC
The mean age of the patients was 64.6 ± 15.0 years and most were female (70.22%) ( Table 1). BMI and drinking/smoking history data were only available for 178 (79.1%) and 203 (89.8%) patients, respectively, because of the inclusion of emergency patients in addition to general internal medicine outpatients.
Smoking history was a risk factor for IC, and 129 (57.3%) out of the 225 patients were non-smokers. While abdominal pain was the rst symptom in all patients, bloody stool was not the rst symptom in any patient. Hospitalization was required for 44.4% of the patients, and the mean (± SD) duration of hospitalization was 8.9 (± 6.3) days. There was no signi cant difference in the length of hospitalization between the recurrent and non-recurrent IC groups, and there was no mortality among either inpatients or outpatients. The medical records documented the entire disease course (i.e., time from IC onset to the resolution of symptoms and initiation of dietary therapy) for hospitalized patients. The time (from IC onset) to resolution of bloody stools and abdominal pain, as well as the resumption of oral intake, were 3.6 (± 1.4), 4.9 (± 1.8), and 5.8 (± 2.5) days (± SD), respectively. No signi cant differences were observed between the recurrent and non-recurrent IC groups in this regard.  Figure 2, Figure 3, Figure 4).  Table 3 Multivariate analysis of risk factors for IC.  Figure 5). The follow-up duration was de ned as the period from the onset of IC until the last day of treatment (documented up to December 31, 2019, in the electronic medical records) for patients whose follow up ended and for those who were discharged or referred to other hospitals during the follow-up period.
Comparison of imaging results between the study groups CT was performed in 218 out of the 225 patients within 5 days after IC onset. The ndings indicative of IC were most commonly observed in the descending colon, followed by the sigmoid colon and transverse colon; only two patients did not exhibit signs of IC in the descending colon. Edematous thickening of the intestine was observed in all patients, and more than 90% of patients had disproportionate fat stranding around the intestine. There was no signi cant difference in the proportion of positive ndings in each colon segment between the non-recurrent and recurrent IC groups (Table 4).
CS was performed in 155 patients between the date of IC onset and 95 days later. CS performed 23 days after IC onset often did not yield ndings indicative of IC. In contrast, positive ndings for IC were observed in all cases when CS was performed within 22 days of IC onset; in two of these cases, insertion was limited to the descending colon, and deep insertion was deferred because of in ammatory ndings or severe pain. Among the patients who underwent CS within 22 days, no signi cant differences were observed in the proportion of positive ndings in each colon segment between the non-recurrent and recurrent IC groups (Table 4). Table 4 Univariate analysis of computed tomography and colonoscopy ndings.

Discussion
The results of this study indicate that a history of appendectomy or cerebral infarction signi cantly increased the risk of IC recurrence. In contrast to previous studies, a signi cant association was not found between IC recurrence and either smoking or AAAs. Prior studies have identi ed a history of abdominal surgery as a risk factor for IC onset (6-18). However, a history of appendectomy has not been reported as a risk factor for IC onset or recurrence.
The appendix, located in the right lower abdomen and connected to the cecum, is anatomically distant from the site most frequently affected by IC. Recent studies have suggested that the appendix functions as an immune organ (28, 29) and that it is associated with intestinal micro ora (30,31) and autoimmune bowel diseases (32)(33)(34). Ulcerative colitis is an autoimmune bowel disease characterized by lesions extending to the rectum; it has been reported to have a relationship with the appendix, which is anatomically distant. Patients who undergo appendectomy at a young age are less likely to develop ulcerative colitis (32,33). This has been attributed to the fact that appendectomy markedly reduces the proliferative response of lymphocytes in the mucosal intrinsic layer to autologous intestinal bacteria. The appendiceal lymphoid tissue has a vital role in producing immunoglobulin A (IgA)-positive cells that are mobilized to the colon. Intestinal IgA induced by appendiceal lymphoid tissue contributes to the regulation of the intestinal microbiota. A previous study found that a decrease in the IgA titer greatly disrupted the intestinal microbiota balance in appendectomized mice (30). Therefore, we speculate that disturbances to the intestinal microbiota balance render patients with a history of appendectomy more prone to repeated intestinal in ammation than those without a history of appendectomy.
Cerebral infarction has previously been reported as a risk factor for the development of IC (6-18). Cerebral infarction can be broadly classi ed as cardiogenic cerebral embolism, lacunar infarction, or atherothrombotic cerebral infarction. Among these, the cause of cardiogenic cerebral embolism is atrial brillation, while lacunar infarction and atherothrombotic cerebral infarction are caused by arteriosclerosis. Both atherosclerotic disease and cerebral infarction have been reported to be risk factors for IC onset. In the present study, we did not observe differences in risk factors for atherosclerosis (e.g., smoking, hypertension, obesity, diabetes, and dyslipidemia) between the recurrent and non-recurrent IC groups. One of the reasons may be that the sample size was too small to provide su cient statistical power to detect a signi cant difference.
Vascular risk factors for IC are attributed to local microcirculatory disturbances rather than a thromboembolic effect. Therefore, in terms of cerebral infarction, the pathogenic mechanism can be more similar to lacunar and atherothrombotic cerebral infarctions, which are caused by a deterioration of circulatory dynamics due to atherosclerosis, than to cardiogenic cerebral embolism.
The characteristics of the patients in this study (mean age, 64.6 ± 15.0 years; 70.22% women) were similar to those reported in previous studies (35) and the ACG clinical guideline (16); this supports the generalizability of the results of this study. In addition, although there was some concern that patients in the non-recurrent IC group may have subsequently experienced IC recurrence, this was unlikely as the median (IQR) follow-up time for all patients was 1556 (353-2768) days; this was much longer than the median (IQR) time to recurrence of 291 (64-907) days in the recurrent IC group.
This study has some limitations. First, as this was a single-center study, the distribution of comorbidities may have been biased. Second, the retrospective study design relied on previously registered disease codes from a database; therefore, our results may have been affected by omissions in the medical records and incomplete reporting of medical histories. Third, the multivariate analysis was limited owing to the small number of IC cases relative to the number of investigated risk factors. Fourth, we excluded patients with secondary IC, as the inclusion criteria were based on the diagnostic criteria used in Japan.
Thus, we were unable to assess patients with IC and hemodynamic deterioration after surgery for AAAs or during treatment of chronic renal failure, as well as those with stenosis or obstruction caused by colorectal cancer. This may be one of the reasons why the number of patients with AAA was zero and that with chronic renal failure and colorectal cancer was low.

Conclusions
This study identi ed cerebral infarction and a history of appendectomy as risk factors for recurrent IC.
Previously reported risk factors for IC onset, including hypertension, diabetes mellitus, and dyslipidemia, were not found to be related to IC recurrence in our study. These ndings may be used to facilitate the identi cation of patients at high risk of IC recurrence and the institution of appropriate intervention measures in clinical practice. The results of this study warrant further con rmation by multicenter prospective studies with larger samples sizes.