The Role of Appendicectomy in Ulcerative Colitis: Systematic Review and Meta-Analysis

Abstract

Background

This updated systematic review and meta-analysis investigates the putative role of the appendix in ulcerative colitis as a therapeutic target.

Methods

Ovid Medline, Embase, PubMed and CENTRAL were searched with MeSH terms (“appendectomy” OR “appendicitis” OR “appendix”) AND (“colitis, ulcerative”) through October 2020, producing 1469 references. Thirty studies, including 118 733 patients, were included for qualitative synthesis and 11 for quantitative synthesis. Subgroup analysis was performed on timing of appendicectomy. Results are expressed as odds ratio (OR) with 95% confidence intervals (CIs).

Results

Appendicectomy before UC diagnosis reduces the risk of future colectomy (OR, 0.76; 95% CI, 0.65-0.89; I² = 5%; P = .0009). Corresponding increased risk of colorectal cancer and high-grade dysplasia are identified (OR, 2.27; 95% CI, 1.11-4.66; P = .02). Significance is lost when appendicectomy is performed after disease onset. Appendicectomy does not affect hospital admission rates (OR, 0.87; 95% CI, 0.68-1.12; I² = 93%; P = .27), steroid use (OR, 1.08; 95% CI, 0.78-1.49; I² = 36%; P = .64), immunomodulator use (OR, 1.04; 95% CI, 0.76-1.42; I² = 19%; P = .79), or biological therapy use (OR, 0.76; 95% CI, 0.44-1.30; I² = 0%; P = .32). Disease extent and risk of proximal progression are unaffected by appendicectomy. The majority (71% to 100%) of patients with refractory UC avoid colectomy following therapeutic appendicectomy at 3-year follow-up.

Conclusions

Prior appendicectomy reduces risk of future colectomy. A reciprocal increased risk of CRC/HGD may be due to prolonged exposure to subclinical colonic inflammation. The results warrant further research, as consideration may be put toward incorporating a history of appendicectomy into IBD surveillance guidelines. A potential role for therapeutic appendicectomy in refractory left-sided UC is also identified.

Lay Summary

This article was written as part of a higher degree with the University of Edinburgh. The first author received the Association of Surgeons in Training (ASiT) Edinburgh Surgery Online Bursary during the completion of the degree and this journal article.

This updated systematic review finds appendicectomy before ulcerative colitis (UC) diagnosis reduces risk of future colectomy but increases the risk of colorectal malignancy. Incorporating a history of appendicectomy into IBD surveillance guidelines could be considered. A potential role for therapeutic appendicectomy in left-sided treatment refractory UC is also identified.

appendicectomy, ulcerative colitis, inflammatory bowel disease

Introduction

Ulcerative colitis (UC) is a form of inflammatory bowel disease (IBD) characterized by chronic idiopathic colonic mucosal inflammation.¹ It has a worldwide prevalence of 7.6 to 245 per 100 000, and both the disease and its treatments are a source of significant morbidity.² The etiology of UC is multifactorial and poorly understood, challenging the identification of therapeutic targets. Despite best efforts, 30% of patients progress to requiring colectomy.³ The appendix, previously considered a vestigial organ, is a current focus in IBD research with a view to utilizing its putative influence on the course of UC for therapeutic benefit.¹

The role of the appendix in ulcerative colitis was first observed in 1987 when patients diagnosed with UC were noted to have a lower incidence of previous appendicectomy compared with non-appendicectomized controls.⁴ Meta-analysis has since demonstrated appendicectomy protects against developing UC (odds ratio [OR], 0.44; 95% confidence interval [CI], 0.30-0.64; P < .0001).⁵ Follow-up studies have suggested this protective effect is only significant when appendicectomy was performed for appendicitis or mesenteric adenitis and that the protective effect is lost if appendicectomy is performed after the age of 20 years.^6–8 Subsequent systematic reviews and meta-analyses have produced variable and conflicting results as new data become available.^9–11

It remains unclear whether the appendix offers protection by an intrinsic immunomodulatory role, through reactive pro-inflammatory pathways in response to acute appendicitis or both. T-cell receptor-α knockout murine models of colitis have shown overexpression of CD4⁺ T cells in appendicular lymphoid follicles.¹² This is associated with a cytokine imbalance, particularly of interleukin (IL)-4, a key regulator in humoral and adaptive immunity responsible for the proliferation and differentiation of T cells and B cells.¹² This appendicular cytokine imbalance is thought to contribute toward auto-sensitization processes that precede the development of UC.^12,13 Appendicectomy in these mice, before 1 month of age, supressed the development of UC.¹³ This led to the suggestion that the appendix acts a priming site for the development of UC in genetically susceptible individuals. Corresponding abnormal appendicular CD4⁺ T-cell expression has been confirmed in vivo in humans diagnosed with UC in active and inactive disease.^14–16 The observation of peri-appendicular red patches (PARPs) in patients prior to developing UC supports the appendix’s proposed role as primer in disease onset.¹⁷ Histological examination of PARP reveals a disease pattern similar to UC rather than acute appendicitis.^17,18 Peri-appendicular red patches have also been observed synchronously in established left-sided UC; its presence might demonstrate a continued immunological role of the appendix in established disease. Conversely, pro-inflammatory pathways in acute appendicitis might act protectively against developing UC. Murine models demonstrate an activation of regulatory T-cell Forkhead box Protein 3 (FoxP3⁺CD25⁺) in appendicitis.¹⁹ Their expression is age-limited, occurring mostly in childhood and is thought to protect against autosensitization by promoting T-regulatory cell activity.^6,18–20 This may account for the loss of protective effect when appendicectomy is performed after 20 years of age. There are limitations in murine models in UC where mice do not have a true appendix but instead cecal lymphoid patches interpreted as the murine equivalent. Accepting this, research suggests biological plausibility of an immunomodulatory role of the appendix in UC.

The primary objective of this PROSPERO-registered review is to analyze the effect of appendicectomy on the clinical course of UC in adults.

Materials and Methods

Criteria for Considering Studies and Outcome Measures

This review is compliant with the Methodological Expectations of Cochrane Intervention Reviews. In the absence of randomized trials, this review includes controlled observational studies, single-arm cohort studies, and case series/reports.²¹ Both retrospective and prospective studies are included. Systematic and literature reviews were manually reference-checked to identify original studies. Non-English and nonhuman studies were excluded. Studies were not excluded on the basis of publication status.

Patients diagnosed with ulcerative colitis and older than 18 years at the time of participating in a study were eligible for inclusion in this review.

Intervention is defined as appendicectomy performed for any indication, at any age, both before and after UC diagnosis. The control is defined as patients with UC who did not have a history of appendicectomy.

Primary outcomes felt to objectively represent disease activity and severity were selected:

1. Risk of colorectal cancer (CRC) and high-grade dysplasia (HGD)

2. Risk of colectomy

3. Medical therapy requirements

4. Disease extent

5. Risk of proximal disease progression

6. Disease activity

Search Strategy

Ovid Medline, Embase, PubMed, Cochrane Central Register of Controlled Trials (CENTRAL), and clinicaltrials.gov were searched using the MeSH terms “appendectomy” OR “appendicitis” OR “appendix” AND “colitis, ulcerative” up to October 2020 (Appendices 1-5). Articles describing the same study or utilizing the same database were collated, including the most updated study where necessary. All included articles underwent citation checking on Web of Science. References were stored in EndNote prior to importing into Covidence for deduplication and review.²²

Data Collection and Analysis

Inclusion decisions

Articles were independently screened by 2 authors (S.W. and Z.S.), resolving disagreement by consensus. Title and abstract screening categorized articles as, “include,” “maybe,” or “exclude.” Full texts were reviewed for those labelled “include” or “maybe” to determine study eligibility.

Data collection process

Data were independently extracted by authors using a standardized data extraction template. In addition to the specified outcome measures, this review collected information to populate a table of characteristics, including study design, date of study, and participant demographics.

Assessing risk of bias and study quality

Authors independently assessed the quality of case-control and cohort studies using the Newcastle-Ottawa Scale (NOS).²³ The NOS awards stars for high-quality studies over 3 domains: “selection” (maximum 4 stars), “comparability” (maximum 2 stars), and either “exposure” or “outcome” (maximum 3 stars) for case-control or cohort studies, respectively. These scores were used to assign studies as “good,” “fair,” or “poor” levels of quality. A “good” study required >2 stars in selection, >0 stars in comparability, and >1 star in outcomes. A “fair” study required 2 stars in selection, >0 stars in comparability, and >1 star in outcomes. A “poor” required: 0-1 stars in selection, 0 stars in comparability, and 0-1 stars in outcomes. The quality of case series and single-arm cohort studies were assessed using the checklist from Murad et al.²⁴ This tool assesses 4 categories (“selection,” “ascertainment,” “causality,” and “reporting”) using 8 leading questions. For the purpose of this review, questions 4-6 were not utilized. This study scored each category a maximum of 1 point each. A “good” study scored 4 points, a “fair” study scored 2-3 points, and a “poor” study scored 0-1 points. Where discrepancy in scoring arose, the highest score was accepted.

Effect measures

Dichotomous outcome variables and count data are displayed as odds ratios (ORs) with 95% confidence intervals (CIs).

Unit of analysis issues

All data were collected from included studies that reported outcome measures over multiple time frames. For data sets reporting multiple results over time, the long-term data were selected for data pooling to reduce the risk of reporting bias.

Assessing heterogeneity and data synthesis

Studies were pooled for meta-analysis if methodological heterogeneity was deemed acceptable and if NOS/Murad scores were “good” or “fair.” Statistical heterogeneity was assessed by calculating the I² value, pooling data with I² values >50% by random effects model and data with I² values <50% by fixed effects model. Where feasible, subgroup analysis was performed on timing of appendicectomy relative to UC diagnosis. Outcome variables with unacceptably high methodological heterogeneity or “poor” qualitative assessment outcomes were not pooled for meta-analysis, instead presented by narrative review. Case reports were included for narrative review only. Statistical analysis was performed by Review Manager 5.4.²⁵ Sensitivity analysis was performed on studies excluded from meta-analysis. Data reporting conforms to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (Appendix 6).²⁶ Results are summarized in a summary of findings table with the certainty of body of evidence assessed according to the 5 GRADE domains.²⁷

Results

Literature Search

Systematic literature search yielded 1469 references. From these, 172 full text reviews were retrieved to assessing eligibility. Thirty-four literature and systematic reviews were manually reference-checked, identifying 1 additional study. Eighteen articles could not be assessed due to non-English text. Thirteen references were collated into 6 references due to reporting on the same or updated data sets. Full texts could not be located for 11 references; authors were contacted without response. In total, there were 30 references eligible for qualitative synthesis: 28 manuscripts and 2 registered ongoing trials (Figure 1). Citation checking of included references did not identify additional studies.

Figure 1.

PRISMA flow diagram of study selection process.

From the 28 references in this review, there were 30 studies eligible for inclusion.^28–55Table 1 summarizes included case-control and cohort studies with NOS score.^{29,31,32,34,35,38,39,41,44,45,47,49,50,55} Two studies described be collecting relevant data that were incompletely reported on and that authors did not respond when contacted for additional data.^53,54 These studies are included for narrative review. There were 2 single-arm cohort studies, 2 case series, and 2 case reports—all presented in Table 2 with Murad scores.^{11,28,30,37,40,46} The ongoing clinical trials are elaborated on in the Discussion.^56,57

Table 1.

Characteristics of included case-control and cohort studies.

Author	Year	Country	Study Design	Follow -up	n		Female (%)	Timing of A- relative to UC Diagnosis	Age at A-	A- < 20 years	Age at UC Diagnosis		Disease Duration		NOS Score
					A-	A+					A-	A+	A-	A+	S	C	E/O	Total
Barreiro29	2009	Spain	Cohort, P	1 year	541 total		-	Before	-	-	-	-	-	-	3	1	2	Fair
Chen31	2018	China	Cohort, R	2 years	11	391	181 (45.0%)	Before	Mean (SD) 29.97 (9.4)	1 (9%)	Mean (SD) 48.64 (5.39)	Mean (SD) 41.72 (15.70)*	Median (IQR) 5.66 (3.12)	Median (IQR) 6.04 (5.08)	4	1	3	Good
Cosnesa,32	2002	France	Cohort, R	10 years	49	589	340 (53.3%)	49 (8%) Before	Median (range) 11 (6–43)	35 (71.4%)	Median (IQR) 35.7 (6.6)	Median (IQR) 32.9 (14.4)	Median (IQR): 10.1 (8.1)	Median (IQR): 7.2 (8.3)*	3	1	2	Good
Cosnesb,32	2002	France	Cohort, P	3 years	41	466	-	-	-	-	-	-	-	-
Florin34	2004	Australia	Case-control, R	7 years	19	275	142 (48.3%)	Beforec	-	8 (42.1%)	Mean (SEM) 37.9 (3.07)	Mean (SEM) 32.4 (0.89)	-	-	4	1	3	Good
Gordillo55	2015	Spain	Cohort, R	4 years	60	771	-	Before	-	-	-	-	-	-	4	1	3	Good
Hallas35	2004	Denmark	Case-control, R	> 10 years	202	808	590 (58.4%)	After	Mean (SD) 43.3 (17.8)	0	Mean (SD) 38.6 (18)	Mean (SD) 38.7 (17.7)	-	-	2	1	3	Fair
Jo38	2003	Japan	Case-control, P	6 months	9	9	8 (44%)	After	Mean (range) 32.5 (13–48)	-	Median (range) 32.5 (13–48)	Median (range) 32.5 (13–47)	Median (range): 4.12 (0.75–14)	Median (range): 4.18 (0.92–15)	2	1	2	Fair
Khoo39	2019	Australia	Cohort, R	> 3 Years	215 total		-	-	-	-	-	-	-	-	4	1	2	Good
Leea,41	2014	Korea	Cohort, R	7 years	68	2544	1208 (46.2%)	Before	Median (range) 29 (10–64)	6 (8.8%)	41 (18–68)	36 (11–80)	-	-	3	1	2	Good
Leeb,41	2014	Korea	Case-control, R	12–13 years	36	144	-	After	Median (range) 42 (19–79)	1 (2.8%)	35 (11–73)	-	-	-	2	1	3	Fair
Myrelida,44	2017	Sweden	Cohort, R	> 10 years	1537	62174	30333 (48%)	Before	Mean (SD) 32.2 (16.3)	250 (16.3%)	Mean (SD) 45.9 (17.4)	Mean (SD) 44.6 (20.3)*	-	-	3	2	3	Good
Myrelidb,44	2017	Sweden	Cohort, P	> 10 years	603	41980	20135 (47%)	After	Mean (SD) 40.6 (17.0)	-	Mean (SD) 33.6 (16.9)	Mean (SD) 42.4 (19.1)*	-	-	4	1	3	Good
Naganuma45	2001	Japan	Case-control	1 year	21	304	155 (47.7%)	Before	Mean (SD) 25.7 (10.9)	-	-	-	Mean (SD) = 12.1 (12.3)	Mean (SD) = 9.9 (7.8)	2	2	3	Fair
Parian47	2013	USA	Case-control, R	-	81	1722	-	49 (60%) Before. 32 (40%) After.	-	-	-	-	Value not given	Value not given	2	1	3	Fair
Selbya,49	2002	Australia	Cohort, P	-	12	239	126 (49%)	Before	Mean (SEM) 26.6 (4.14)	4 (33.3%)	Mean (SEM) 42.5 (6.45)	Mean (SEM) 32.1 (0.84)*	-	-	2	1	2	Fair
Selbyb,49	2002	Australia	Cohort, P	-	8	239	126 (49%)	After	Mean (SEM) 31.8 (3.7)	0%	Mean (SEM) 24.6 (3.44)	Mean (SEM) 32.1 (0.84)	-	-
Stellingwerf50	2020	The Netherlands	Cohort, P	>10 years	63	763	452 (54.7%)	24 (38.1%) Before. 33 (52.4%) after. 6 (9.5%) undeclared	Median (range) 31 (8–70)	-	Median (range) 31 (8–70)	Median (range)31.5 (5–82)	Median (range) 31 (8–70)	Median (range) 31.5 (5–82)	4	1	3	Good
Harnoy36	2015	France	Cohort, R	10 years	15	217	-	-	-	-	-	-	-	-	2	0	2	Poor
Martinez Lozano42	2018	Spain	Cohort, R	7 years	3	68	38 (53.5%)	-	-	-	-	-	-	-	3	1	1	Poor
Masser43	2013	Austria	Cohort, R	-	22	215	-	Before	-	-	-	-	-	-	2	1	1	Poor
Park48	2014	Korea	Cohort, R	-	66	280	-	Beforec	-	-	-	-	-	-	3	0	1	Poor
van der Slootd,53	2020	The Netherlands	Cohort, R	-	20	327	188 (54.2%)	-	-	-	-	-	-	-	3	0	1	Poor
Walshd,54	2017	USA	Cohort, R	4.3 (3.3–15.1)	9	60	93 (55%)	-	-	-	-	-	-	-	3	0	2	Poor
Toader52	2012	Romania	Cohort, P	2–15 years	183 total		-	-	-	-	-	-	-	-	1	0	1	Poor
Elias33	2016	Brazil	Cohort, P	6 months	160 total		70 (44%)	Before	Mean (SD) 47.6 (14.1)	-	-	-	-	-	3	0	1	Poor

Modified table from Stellingwerf et al.¹¹ Abbreviations: A-, appendicectomy; A+, no appendicectomy; P, prospective; R, retrospective; NOS, Newcastle-Ottawa Scale; S, Selection; C, Comparability; E/O, Exposure/Outcome.

Author^a—Appendicectomy before UC diagnosis. Author^b—appendicectomy after UC diagnosis. ^cPatients who underwent appendicectomy after UC diagnosis were excluded following appendicectomy. *P < .05. ^dUnreported outcome measures, authors did not respond when contacted.

Abbreviations: SD, Standard deviation; IQR, Interquartile range; SEM, standard error of the mean.

Table 2.

Characteristics of single arm cohort, case series, and case report studies.

Author	Year	Country	Study Design	Follow -up	Center(s)	n	Female (%)	Age (y)	Disease Duration (y)	Timing of A- relative to UC Diagnosis	Age at A- (y)	A- < 20 years	Disease Location	Medication	UC Refractory to Medical Therapy	Murad
																S	A	C	R	Quality
Bolin30	2010	Australia	Single-arm cohort, P	3 Years	1 Center	50	-	-	Median (range) 5 (0–30)	100% After	-	-	100% Proctitis	ASA, CS, IM, B and T	Yes	1	1	1	1	Good
Stellingwerf51	2019	The Netherlands	Single-arm cohort, P	3 months—5 years	2 Centers	28	13 (46.4%)	Mean (range) 40.5 (33.3–47.8)	Mean (range) 8 (3.3–14.8)	100% After	Mean (range) 40.5 (33.3–47.8)	0%	All	ASA, CS, IM, B	Yes	1	1	1	1	Good
Bageacu28	2011	France	Case series	3.6 years	1 Center	8	3 (37.5%)	Mean 42.6	-	100% After	Mean 42.6	-	100% Proctitis	ASA	Yes	0	1	0	0	Poor
Järnerot37	2001	Sweden	Case series	2–4 years	1 Center	6	4 (66.7%)	Mean 51.5	Median (range) 2.5 (0–12)	100% After	Mean 47.3	-	All	ASA, CS, IM	Yes	0	1	0	0	Poor
Kong40	2015	USA	Case study	-	1 Center	1	0	25	-	100% Before	25	0%	Pancolitis	-	No	-	-	-	-	N/A
Okazaki46	2000	Japan	Case study	-	1 center	1	0	21	3	100% After	21	0%	100% Proctitis	-	No	-	-	-	-	N/A

Modified table from Stellingwerf et al.¹¹ Abbreviations: A-, appendicectomy; A+, no appendicectomy; P, prospective; R, retrospective; Murad assessment; S, selection; A, ascertainment; C, causality; R, reporting.

In total, there were 118 733 patients with ulcerative colitis included in this review. Of these, 4148 (3.49%) underwent appendicectomy; 1938 (46.72%) patients underwent appendicectomy prior to UC diagnosis; 2152 (51.88%) patients underwent appendicectomy after UC diagnosis; and 94 (2.27%) appendicectomies did not have their timing clarified. Eleven studies reported age at appendicectomy.^{31,32,35,38,41,44,45,49–51} Of these, 7 described the proportion of those undergoing appendicectomy prior to 20 years of age.^{31,32,34,41,44,49} However, only 1 performed subgroup analysis on this.⁴⁴ A total of4 studies described the indication for appendicectomy,^35,41,44 of which 2 performed subgroup analysis on this.^35,44 This review did not perform subgroup analysis on age or indication for appendicectomy. Lastly, 4 studies^32,38,47,50 out of 6 studies^{31,32,38,45,47,50} reported longer disease duration in appendicectomized patients compared with control. Two studies reported no significant difference.^31,38 The final results are presented in the summary of findings (Table 3).

Table 3.

Summary of findings table with certainty of body of evidence according to the 5 GRADE domains.

Certainty Assessment							No. Patients		Effect		Certainty	Importance
No. Studies	Study Design	Risk of Bias	Inconsistency	Indirectness	Imprecision	Other Considerations	Appendicectomy	No Appendicectomy	Relative (95% CI)	Absolute (95% CI)
Colorectal cancer and high-grade dysplasia
4	observational studies	not serious	not serious	not serious	seriousa,b	none	13/191 (6.8%)	102/2398 (4.3%)	OR 1.90(0.93 to 3.85)	35 more per 1,000 (from 3 fewer to 104 more)	⨁◯◯◯ VERY LOW
Colectomy
9	observational studies	not serious	seriousc	not serious	not seriousb	none	283/2689 (10.5%)	10966/111868 (9.8%)	OR 0.93(0.58 to)	21 fewer per 1,000 (from 32 fewer to 10 fewer)	⨁◯◯◯VERY LOW
Steroid requirement
4	observational studies	not serious	not serious	seriousd	not serious	none	87/195 (44.6%)	1943/3964 (49.0%)	OR 1.08(0.78 to 1.49)	19 more per 1,000(from 62 fewer to 99 more)	⨁◯◯◯VERY LOW
Immunomodulator requirement
6	observational studies	not serious	not serious	seriousd	seriousb	none	56/251 (22.3%)	1021/4793 (21.3%)	OR 1.04(0.76 to 1.42)	7 more per 1,000 (from 43 fewer to 65 more)	⨁◯◯◯VERY LOW
Biological requirement
3	observational studies	not serious	not serious	seriousd	seriousb	none	17/167 (10.2%)	338/3440 (9.8%)	OR 0.76(0.44 to 1.30)	22 fewer per 1,000 (from 53 fewer to 26 more)	⨁◯◯◯VERY LOW

Table modified from GRADEpro GDT. Abbreviations: CI, confidence interval; OR, odds ratio.

^aSmall sample sizes, ^bImprecise due to wide confidence intervals with potential for important harm, ^cHigh level of heterogeneity, I² >50%, ^dVariation in medication protocols between studies.

Risk of Colorectal Cancer and High-grade Dysplasia

Five studies assessed the risk of CRC and/or HGD in 2589 UC patients.^{32,34,36,50,55} Two studies combined results for HGD and CRC incidences.^34,55 Two studies reported HGD and CRC incidences separately,^36,50 and 1 study only reported on CRC.³² This review considered CRC and HGD equivalent and combined all data for analysis. Four studies were suitable for meta-analysis.^32,34,50,55 follow-up period ranged from 4 to 11 years.

Meta-analysis on risk of colorectal cancer and high-grade dysplasia

Appendicectomy in UC does not significantly affect the risk of developing CRC and/or HGD compared with non-appendicectomized controls (OR, 1.70; 95% CI, 0.94-3.07; I² = 18%; P = .08; Figure 2a). Subgroup analysis on appendicectomy before UC diagnosis shows a significantly increased risk of CRC/HGD compared with non-appendicectomized controls (OR, 2.27; 95% CI, 1.11-4.66; I² = 0%; P = .02; Figure 2b).^32,34,55

Figure 2.

Meta-analysis of risk of colorectal cancer (CRC) and high grade dysplasia (HGD) according to timing of appendicectomy.

Poor-quality studies and sensitivity analysis in risk of colorectal cancer and high-grade dysplasia

The study that was not pooled for meta-analysis did not describe controlling confounding factors, resulting in a poor NOS score.³⁶ In this study, blinded pathologists examined 232 UC colectomy specimens. Fifteen were noted to have previous appendicectomy of uncertain timing relative to UC diagnosis. This was associated with a significantly increased risk of CRC/HGD (OR, 17.16; 95% CI, 3.5-84.2; P = .005).³⁶ Sensitivity analysis saw a significantly increased risk of CRC/HGD in appendicectomized patients (OR, 2.71; 95% CI, 1.10-6.67; I² = 59%; P = .03; Figure 2c).

Risk of Colectomy

Ten studies, involving 112 754 UC patients, assessed the risk of future colectomy, of which 9 were pooled for meta-analysis.^{31,32,34,35,41,44,47–50} Three studies reported separate results for patients who underwent appendicectomy prior to UC diagnosis (indicated by “author, a”) compared with those appendicectomized after diagnosis (indicated by “author, b”).^41,44,49 The incidence of colectomy in appendicectomized UC patients compared with control was 0% to 28% and 5.2% to 37.6%, respectively, over variable follow-up periods (6 months to 46 years).

Meta-analysis on risk of colectomy with subgroup analysis on timing

Appendicectomy in UC does not significantly affect the risk of colectomy (OR, 0.93; 95% CI, 0.58-1.49; I² = 85%; P = .76; Figure 3a). Subgroup analysis on 6 studies saw a significantly reduced risk of future colectomy when appendicectomy was performed before UC diagnosis (OR, 0.76; 95% CI, 0.65-0.89; I² = 5%; P = .0009; Figure 3b).^{31,32,34,41,44,49} Appendicectomy after UC diagnosis did not significantly affect the risk of future colectomy (OR, 1.37; 95% CI, 0.61-3.07; I² = 63%; P = .45; Figure 3c).^35,41,44,49

Figure 3.

Meta-analysis of risk of colectomy according to timing of appendicectomy.

Subgroup analysis on variables other than timing of appendicectomy

One high-quality study, including 63 711 UC patients, performed subgroup analysis on the indication and age of appendicectomy in UC patients.⁴⁴ Appendicectomy performed prior to UC diagnosis for appendicitis or other pathology reduced the risk of colectomy when performed on patients younger than 20 years old (hazard ratio [HR], 0.44; 95% CI, 0.27-0.72; P = .001; and HR, 0.62; 95% CI, 0.43-0.90; P = .011).⁴⁴ Conversely, appendicectomy for appendicitis after UC diagnosis increased the risk of colectomy even when performed younger than 20 years old (HR, 1.56; 95% CI, 1.20-2.03; P = .001).⁴⁴

Treatment refractory disease and risk of colectomy

Two single-arm prospective cohort studies observed the risk of colectomy in treatment refractory disease following appendicectomy.^30,58 Both studies scored highly on quality assessment. One observed 28 adult patients with disease of all locations with inadequate response to 5-aminosalicylates (5-ASA), corticosteroids, immunomodulators, and biologicals (as defined by multidisciplinary team discussion) who would otherwise proceed to pan-proctocolectomy.⁵¹ Following therapeutic appendicectomy, the proportion of patients who proceeded to colectomy at 3 and 6 months were 3 (10.71%) and 6 (21.42%), respectively. Long-term follow-up (median 3.7 years) observed 71.43% of patients avoided colectomy following therapeutic appendicectomy. Histological examination found patients were more likely to benefit if there was active appendiceal inflammation compared with none/extensive ulcerations (85% and 20%; P = .001).⁵⁸ Additionally, greater therapeutic benefit was observed in patients with left-sided/proctitis compared with those with pancolitis (63% and 25%; P = .05).⁵⁹ The second study observed 50 patients with UC proctitis over 3 years, noting none went on to require colectomy following appendicectomy.³⁰ This study did not clearly define what constituted “treatment refractory” disease.

Poor-quality studies and sensitivity analysis in risk of colectomy

One study did not control for confounding factors resulting in a poor NOS score.⁴⁸ This single-center study of 2346 UC patients found the incidence of colectomy was not affected by a history of previous appendicectomy compared with control (10.6% and 8.6%; P = .56).⁴⁸ Sensitivity analysis on the excluded data did not significantly affect results (OR, 0.96; 95% CI, 0.62-1.49; I² = 84%; P = .85).

Medical Therapy Requirements

Eight cohort studies, 2 case-control studies, and 1 single-arm cohort assessed UC medication requirements in 7832 patients.^{30–32,34,41,43,48–50,53} Five studies presented results for patients appendicectomized before UC diagnosis.^{31,34,41,43,49} Three studies presented results for patients appendicectomized after diagnosis.^30,41,49 Two studies presented combined results for patients appendicectomized before and after diagnosis.^32,50 Two studies did not describe timing of appendicectomy relative to diagnosis.^48,53 Of the eleven cohort and case-control studies, 8 were pooled for meta-analysis.^{31,32,34,41,49,50} This review assessed the following UC medication requirements: corticosteroids, immunomodulators, and biologicals. Immunomodulators included in this review are azathioprine, mercaptopurine, methotrexate, and 5-ASAs.

Appendicectomy does not significantly affect corticosteroid requirements (OR, 1.08; 95% CI, 0.78-1.49; I² = 36%; P = .64),^32,41,50 immunomodulator requirements (OR, 1.04; 95% CI, 0.76-1.42; I² = 19%; P = .79),^{32,34,41,43,49,50} or biological requirement therapies (OR, 0.76; 95% CI, 0.44-1.30; I² = 0%; P = .32; Supplemental Figures 4a, 4b, and 4c, respectively).^41,50 Subgroup analysis could only be performed on immunomodulator requirements where appendicectomy before UC diagnosis did not significantly alter results (OR, 0.85; 95% CI, 0.51-1.42; I² = 63%; P = .84). Of note, only one of the high-quality studies defined a minimum duration of medication course in their inclusion requirements (at least 1 year).³⁴ This was the only study observing a significant reduction in immunomodulator requirements following appendicectomy (OR, 0.13; 95% CI, 0.02-0.99; P = .049).³⁴

One high-quality study reported the overall medication requirements, combining steroids, immunomodulators, and biological agents. This study found no significant difference in medication requirements between appendicectomized and non-appendicectomized controls (27.3% and 49.1%; P = .153).³¹

Pre- and postappendicectomy medication requirements

Two of the 11 studies compared pre- and postappendicectomy medication requirements. The first, a prospective single-arm cohort study, examined the effect of appendicectomy on 50 patients with treatment refractory ulcerative proctitis.³⁰ Patients were followed up to 3 years. Following appendicectomy, 15 (30%) became medication free. The second, a retrospective nested case-control study, compared 36 appendicectomized patients with 144 non-appendicectomized controls. Appendicectomy reduced steroid requirements from 58.33% to 26.67%. However, azathioprine and biological requirements increased from 11.11% to 18.75%, and from 0% to 5.56%, respectively.⁴¹ Corresponding changes in medication requirements were observed in the control group with no statistically significant difference observed when comparing intervention to control groups.

Poor-quality studies and sensitivity analysis in medication requirements

Three studies scored poorly on quality assessment, 2 found no significant difference in medication requirements, and 1 study had incomplete data sets.^43,48,53 Sensitivity analysis did not significantly affect results for any medication: corticosteroid requirements (OR, 1.09; 95% CI, 0.79-1.48; I² = 15%; P = .60), immunomodulator requirements (OR, 1.00; 95% CI, 0.75-1.35; I² = 14%; P = .98), and biological requirements (OR, 0.74; 95% CI, 0.46–1.18; I² = 0%; P = .20).

Disease Extent

Eight studies described maximal disease extent at the time of diagnosis in 5536 UC patients (Supplementary Table 4).^{31,32,34,41,45,49,50} Seven found no significant difference.^{32,34,41,45,49,50} One study observed greater numbers of ulcerative proctitis and lesser numbers of pancolitis in appendicectomized compared with non-appendicectomized controls (ulcerative proctitis, 36.3% and 9.2%; pancolitis, 18.2% and 49.1%; P = .0009).³¹

Risk of Proximal Disease Progression

Six studies examined the effect of appendicectomy on the risk of proximal disease progression in proctitis and left-sided disease with contradicting results.^{39,41,42,52,54} Meta-analysis could not be performed due to poor NOS scores^42,52,54 or incomplete data sets, despite attempts to contact authors.^39,52

High-quality studies and risk of proximal disease progression

Three studies including 2015 patients scored satisfactorily on quality assessment.^39,41 Two studies compared the effect of appendicectomy prior to UC diagnosis, and 1 assessed the effect of appendicectomy in established UC. All studies found appendicectomy does not affect the rate of proximal disease progression. The first included 215 patients and compared the 34 experiencing disease progression against those without and found no difference in incidence of previous appendicectomy between groups (2.9% and 4.2%; P = .722).³⁹ The second compared left-sided disease progression in 44 patients with prior appendicectomy against 1720 without and found no significant difference (OR, 0.64; 95% CI, 0.30-1.38; P = .26).⁴¹ The same authors prospectively observed 36 patients and found a non-significant reduction in the rate of proximal disease extension following appendicectomy from 11 of 27 (40.7%) to 3 of 16 (18.8%).⁴¹

Poor-quality studies and risk of proximal disease progression

The results from 3 studies with poor-quality assessment scores were conflicting.^42,52,54 One reported appendicectomy protected against proximal progression.⁵² One found no significant difference in proximal progression in appendicectomized group compared with control (7.1% and 3.8% respectively; P = .588).⁴² One found appendicectomy significantly increased risk of proximal progression (adjusted hazard ratio 2.74; 95% CI, 1.07-7.01; P = .04).⁵⁴ These studies scored poorly due to poor comparability,^52,54 self-reported outcome measures without independent verification,⁴² or concerns regarding study design and risk of detection bias.⁵²

Disease Activity

Thirteen studies assessed disease activity in 2873 patients.^{28–33,37,38,45,50,51} Eight scored satisfactorily on quality assessment.^{29–32,38,45,50,51} Three studies scored poorly,^28,33,37 and 2 case reports were not suitable for quality assessment.^40,46 Methodological heterogeneity precluded meta-analysis.

Satisfactory quality assessment studies on disease activity

From 8 studies including 2697 patients,4 found appendicectomy improved disease activity,^30,32,45,51 but 4 found appendicectomy had no effect.^29,31,38,50 No study reported appendicectomy exacerbated disease activity. The first study assessed disease severity by comparing the incidence of hospitalization with disease symptoms and/or patient reports of abdominal pain ≥10 stool/day. Appendicectomy prior to diagnosis was found to significantly reduce disease severity compared with control, (12 of 21 [57%] and 239 of 303 [79%] respectively; P = .031).⁴⁵ The second study involved 507 patients with UC and assessed severity by comparing “active years.” These were defined as any patient-reported flare-ups in a year constituting 1 active year. There were 41 patients with a history of appendicectomy before or after diagnosis. The study observed 47 active years in 98 person years in this group (48%). This was significantly less than the 466 patients without appendicectomy: 631 active years in 1024 patient years (62%; P < .01).³² Two studies prospectively examined the effect of appendicectomy on treatment-refractory disease.^30,51 One observed 50 patients with UC proctitis after appendicectomy over 3 years.³⁰ Severity was assessed by the simple clinical colitis activity index (SCCAI). This validated index scoring system attributes higher scores to more severe disease (range 0-19).⁶⁰ In total, 40 (80%) patients experienced a significant reduction in SCCAI scores from a median of 9 (range 7-12) to 2 (range 0-12; P < .0005). The majority (37 of 40) of responders displayed a sustained effect at 3 year follow-up, and all patients avoided colectomy.³⁰ The second study included 28 patients with refractory UC of all locations and assessed Mayo and patient-reported Inflammatory Bowel Disease Questionnaire (IBD-Q) scores.⁵¹ There are 3 related Mayo scoring systems assessing up to 4 domains (each rated 0-3), including stool frequency, rectal bleeding, physician’s global assessment, and endoscopic appearance.⁵¹ This study investigated partial Mayo (PM), endoscopic Mayo (EM) and total Mayo (TM) scores. Patients were assessed at 3 months, 6 months, and at long-term follow-up (median 3.7 years, range 2.3-5.2 years). Overall, TM and PM scores saw a significant and sustained reduction following appendicectomy.⁵¹ The pre-operative median TM score was 9 (interquartile range [IQR], 8-11). This reduced to 5.3 (IQR, 4.1-6.6), 4.7 (IQR, 3.0-6.3) and 2.9 (IQR, 1.3-4.6) at 3 months, 6 months, and long-term follow-up (P < .001), respectively. Median pre-operative PM score was 7 (IQR, 6-8). This reduced to 3.5 (IQR, 2.7-4.3), 2.7 (IQR, 1.7-3.8), and 1.4 (IQR, 0.3-2.5) at 3 months, 6 months, and long-term follow-up, respectively (P < .001 for all). At 3 months, 14 of 28 (50%) were deemed to have a clinical response, as defined by a decrease of >3 points in the PM score .⁵¹ The clinical response was sustained at 12 months and long-term follow-up in 13 of 14 (92.9%) responders. These findings were associated with a significant improvement in self-reported IBD-Q scores at long-term follow-up.⁵¹

Four studies of 1787 patients with satisfactory quality assessment scores observed appendicectomy did not affect disease severity.^29,31,38,50 The largest involved 826 patients, of which 63 underwent appendicectomy before or after diagnosis.⁵⁰ Severity was assessed by SCCAI and EM scores. Rather than reporting mean SCCAI scores, this study reported the number of patients with SCCAI <5 as a representation of inactive disease. No significant difference was noted between the appendicectomy and control groups (12.9% and 14.1%; P = .810).⁵⁰ Similarly, EM scores >1 were considered active disease. Again, no significant difference was noted between appendicectomy and control groups (30.5% and 35.3%; P = .91).⁵⁰ The second study assessed 402 UC patients, of which 11 had appendicectomy before diagnosis, by comparing Truelove and Witts severity index scores.³¹ This index considers bowel frequency, the presence of blood in stool, clinical observations, and anemia or raised erythrocyte sedimentation rate to categorize disease as, “mild,” “moderate” or “severe.”³¹ Over a 2-year follow-up, there was no significant difference in the distribution of patients across the 3 categories (P = .444). Lastly, a case-control study matched 18 patients with UC by age, sex, duration, disease extent, and medication requirements.³⁸ Nine patients underwent appendicectomy, and response was assessed by ulcerative colitis activity index (UCAI) and endoscopic Blackstone criteria.³⁸ Patients were assessed 1 week before intervention and followed up at 3 intervals: 4 weeks, 3 months, and 6 months. Results found a statistically significant drop in UCAI scores from pre-intervention baseline at 4-week follow-up for both intervention and control groups (from 147.2 ± 29 to 118.8 ± 29; P < .05, and from 149.7 ± 28.7 to 114.3 ± 36.0; P < .05, respectively).³⁸ The degree of reduction was not significantly different between the 2 groups (appendicectomy group, 28.6 ± 30.6; control group, 35.3 ± 36.1; P > .2).³⁸ Overall, the endoscopic scores showed no significant change for either group throughout the study. However, 5 patients with distal colitis were noted to experience endoscopic and histological improvement following appendicectomy. Patients with pancolitis did not show corresponding improvement.³⁸ The final study found that in 541 UC patients, appendicectomy did not condition disease severity (P = .63).²⁹ This study used an undisclosed severity index including surgical resection, steroid-dependency, or resistance and immunosuppressive/biologic therapy requirements.

Poor-quality studies in disease activity

Three studies scored poorly on quality assessment, 2 reported beneficial effects, and 1 reported equivocal results.^28,33,37 Results are summarized in appendix 7.

Discussion

To our knowledge, this is the largest meta-analysis assessing the effect of appendicectomy on UC disease course. It also introduces new data by performing subgroup analysis on timing of appendicectomy relative to UC diagnosis on several of the suitable variables that were analyzed. When performed at any time relative to UC diagnosis, appendicectomy does not affect the risk of future colectomy, CRC/HGD, medication requirements, disease extent, or risk of proximal disease progression. Substantial methodological heterogeneity precluded meta-analysis of disease activity. However, 4 of 8 studies describe appendicectomy (before or after UC diagnosis) to be associated with either subjective or objective improvement in disease activity. Four studies report no significant effect, whereas none report exacerbating disease activity levels. However, when appendicectomy is performed prior to UC diagnosis, this significantly reduces the risk of future colectomy by 24%. This acts as a surrogate marker suggesting the disease may be more amenable to medical treatment in appendicectomized patients. Further subgroup analysis on timing of appendicectomy relative to UC diagnosis was limited to the risk of CRC/HGD and immunomodulator requirements; the latter noted no difference. The reduced risk of colectomy is at the cost of a disproportionately increased risk of CRC and HGD, which is more than double compared with non-appendicectomized controls. Previous meta-analysis has demonstrated the proportion of colectomies performed for colitis symptoms was less in those with previous appendicectomy compared with control (40.9% and 86.3%, respectively).¹¹ This will increase the apparent risk of colectomy for CRC and HGD, yet does not account for the increased absolute risk. The latter might be due to an extended at-risk exposure to subclinical colonic inflammation, demonstrated by the observed longer disease durations in appendicectomized patients.^32,38,47,50 This subsequently increases the risk of dysplastic and/or metaplastic progression. If these patients exhibit milder disease courses, their risk of malignant progression may be exacerbated by being less likely to require 5-ASA therapy with its proven protective effect against dysplastic progression.⁶¹ This review found only 2 studies reporting 5-ASA requirements, with neither study observing a difference in appendicectomized vs non-appendicectomized patients; subgroup analysis on timing of appendicectomy could not be performed. The small sample sizes and lack of power calculations from these studies justify further research in this area. There is also a possibility that patients who undergo appendicectomy but still develop UC represent a phenotypical subtype of UC that inherently increases the risk of HGD and CRC. This review identifies a paucity in data attempting to elucidate possible pathophysiological mechanisms of UC onset unique to this patient cohort. This justifies further research in this area, as this may create a diagnostic opportunity for identifying at-risk patients or those with undiagnosed mild disease.

The absolute risk of colonic malignancy and high-grade dysplasia observed in this review was 6.80% and 4.25% for appendicectomized and non-appendicectomized patients, respectively. This is much greater than previous reports of a 1.26% total risk of developing CRC in all UC patients.⁶² This discrepancy is likely attributable to this review’s decision to incorporate HGD into CRC analysis. This was included for 2 reasons. First, some studies did not differentiate between HGD and CRC, so combining results facilitated data pooling for meta-analysis. Secondly, the management plan for both conditions are equivalent; surgical resection is advocated where patient suitability permits due to the risk of metaplastic progression in HGD.⁶³

The results observed in this review suggest the putative immunomodulatory roles of the appendix are primarily involved in pathophysiological pathways of disease onset. The potential role for therapeutic appendicectomy in established UC is therefore contentious. However, limited evidence suggests a therapeutic advantage when performed specifically in patients with treatment-refractory disease. Appendicectomy in this cohort has demonstrated improvement in both objective and subjective disease severity indices, with a sustained benefit in a significant proportion at long-term follow-up. There will be a degree of subjectivity to the included studies’ definitions of “treatment-refractory” disease. However, by incorporating decision-making from multidisciplinary team meetings, this apparent subjectivity reflects standard clinical practice. Laparoscopic appendicectomy could delay, or obviate, the need for colectomy in a significant proportion of this cohort, relieving patients of the associated psychosocial, financial, and quality of life costs. With this in mind, the results from the ongoing COlonic Salvage by Therapeutic Appendicectomy (COSTA) trial are much anticipated.⁵⁷ Furthermore, the observation that patients with left-sided treatment-refractory disease have a greater likelihood of benefiting from appendicectomy than pan-colonic UC sufferers introduces another question: do these patients represent a separate phenotype of this disease? If so, this may account in part for conflicting results across included studies. This review recommends future relevant research must consider stratifying according to disease extent.

Sensitivity Analysis

Sensitivity analysis was performed on 3 outcome variables: incidence of CRC/HGD, risk of colectomy, and medication requirements. The overall risk of CRC and HGD, regardless of timing of appendicectomy, became significant when including the poor-quality study. This study was excluded from data pooling based on a poor NOS score due to a lack of description regarding the control of confounding factors, including age, sex, smoking status, family history, disease severity, and duration. An attempt at contacting authors for additional information was unsuccessful. This was the only study to retrospectively compare the incidence of appendicectomy in colectomy specimens, noting a disproportionately high number of CRC/HGD in colectomy specimens with previous appendicectomy. These results might support this study’s proposed hypothesis that appendicectomy increases the risk of CRC due to disease quiescence and prolonged exposure to subclinical colonic inflammation. The results for the 2 other outcomes did not significantly change following sensitivity analysis.

Review Limitations

The majority of included studies were retrospective cohort studies, and none were randomized. This inherently limits the certainty of body of evidence, as displayed in the summary of findings table. Furthermore, variability existed in the extent of controlling for confounding factors, likely contributing toward divergent results. Results from the ongoing randomized-controlled ACCURE and ACCURE-UK 2 trials are awaited.^56,64 A lack of translation service mandated the exclusion of non-English studies, introducing publication bias. This is exacerbated by the failure of authors to respond to this review when seeking full-text articles or additional data. A small number of studies previously reported that the protective effect of appendicectomy in UC is lost when performed for indications other than appendicitis or in adulthood.^6–8 This review identifies a paucity in later studies reporting on these variables and recommends consideration be put toward exploring these variables by subgroup analysis in future research.

The medication requirement results must be interpreted with caution as this review recognizes unavoidable methodological heterogeneity driven primarily by progressive pharmacological advancements and continual updating of differing national guidelines at various rates.^65,66 This is exemplified by the identification of UC-specific pro-inflammatory immune pathways amenable to selective targeting by biological agents like tumor necrosis factor-alpha (TNF-a) inhibitors.⁶⁷ Although the Food and Drug Administration has approved the use of 3 such biological agents for use in treatment-refractory UC, this has yet to be incorporated into some national guidelines.⁶⁸ Of the 7 high-quality studies included in the review, drug protocols, medication agents, doses, and course durations were not described in 5. Furthermore, the data on medication use was collected over a 29-year period and internationally. This introduces a degree of unavoidable methodological variability that reflects global clinical practice. Despite this, the trend observed by 6 of 7 studies was that medication requirements were unaffected by a history of appendicectomy. This and the observed low statistical heterogeneity led to the decision to pool for meta-analysis. It is also important to consider the reciprocal influence that pharmacological advancements may have on the outcome measures in this review, potentially contributing to the observed divergent results.

Conclusion

To conclude, the putative role of the appendix in UC is complex and multifactorial. Although pathophysiological mechanisms remain to be elucidated, it is clear that appendicectomy prior to UC diagnosis reduces the risk of future colectomy. The resultant extended in vivo subclinical colonic inflammation may be the cause for the reciprocal increased risk of metaplastic progression. Further research in this field is justified, as consideration could be put toward incorporating a history of previous appendicectomy into current IBD surveillance guidelines. A potential role for therapeutic appendicectomy in left-sided treatment-refractory UC is also identified, where it may preserve functional colon and quality of life in a significant proportion of patients. Immunological pathways underpinning this are uncertain but could relate to the reduction of auto-sensitization processes, pro-inflammatory pathway amelioration, and alterations in intestinal microbiome.

Acknowledgments

No ethical approval was sought for this research. No patient consent was sought for this research. All material has been referenced where presented and/or modified. This article is PROSPERO registered (CRD42021251927).

Conflicts of Interest

Authors declare no conflict of interest or further financial disclosures.