Hypoalbuminaemia, Not Biologic Exposure, Is Associated with Postoperative Complications in Crohn’s Disease Patients Undergoing Ileocolic Resection

Abstract

Background

There are limited data on the postoperative outcomes in Crohn’s disease patients exposed to preoperative ustekinumab or vedolizumab. We hypothesised that preoperative biologic use in Crohn’s disease is not associated with postoperative complications after ileocolic resection.

Methods

Crohn’s disease patients who underwent ileocolic resection over 2009–2019 were identified at a large regional health system. Preoperative biologic use within 12 weeks of surgery was categorised as no biologic, anti-tumour necrosis factor, vedolizumab, or ustekinumab. The primary endpoint was 90-day intra-abdominal septic complication. Risk factors included preoperative medical therapies, demographics, disease characteristics, laboratory values, and surgical approach. Regression models assessed the association of biologic use with intra-abdominal septic complication.

Results

A total of 815 Crohn’s disease patients who underwent an ileocolic resection were included [62% no biologic, 31.4% anti-tumour necrosis factor, 3.9% vedolizumab, 2.6% ustekinumab]. Primary anastomosis was performed in 85.9% of patients [side-to-side 48.8%, end-to-side 26%, end-to-end 25%] in primarily a stapled [77.2%] manner. Minimally invasive approach was used in 41.4%. The 90-day postoperative intra-abdominal sepsis rate of 810 patients was 12%, abscess rate was 9.6%, and anastomotic leak rate was 3.2%. Multivariable regression modelling controlling for confounding variables demonstrated that preoperative biologic use with anti-tumour necrosis factor [p = 0.21], vedolizumab [p = 0.17], or ustekinumab [p = 0.52] was not significantly associated with intra-abdominal septic complication. Preoperative albumin < 3.5 g/dl was independently associated with intra-abdominal septic complication (odds ratio [OR] 1.76 [1.03, 3.01]).

Conclusions

In Crohn’s disease patients undergoing ileocolic resection, preoperative biologics are not associated with 90-day postoperative intra-abdominal septic complication. Preoperative biologic exposure should not delay necessary surgery.

1. Introduction

Crohn’s disease [CD] management has been revolutionised by biologic therapy. Although anti-tumour necrosis factor [TNF] therapy has been the mainstay of CD management, newer biologics including vedolizumab [VDZ; an anti-integrin monoclonal antibody] and ustekinumab [UST; an anti-interleukin 12/23 antibody], are being increasingly used. Despite the developments in medical therapy, surgery remains necessary in up to 60–80% of CD patients at some point during their disease course, most often for medically refractory stricturing or penetrating disease.¹ Biologics have potentially reduced these rates, but surgery remains common.^2–10 Ileocaecal resection with anastomosis remains the most common intra-abdominal surgery for CD.

Due to potential influence of immune modification on operative outcomes, early retrospective studies assessing the impact of preoperative biologic exposure with peri- or postoperative complications (eg, intra-abdominal septic complications [IASC] and superficial surgical site infections) suggested a potential association.^7–9 However, subsequent studies have called these associations into question.^10,11 Furthermore, data in CD patients treated with the newer biologics, VDZ and UST, have been heterogeneous, controversial, and limited.^12–19 Many of these studies included heterogeneous patient populations [both CD and ulcerative colitis] undergoing a wide spectrum of surgical interventions [eg, small bowel resection, colectomy, stricturoplasty]. Such variability in study populations and procedures with known differences in IASC risk limits interpretation. Thus, the influence of preoperative VDZ and UST use before abdominal surgery in CD patients undergoing ileocaecal resection remains unknown. We aimed to assess the association of postoperative IASC rate in CD patients undergoing ileocaecal resection [ICR] with preoperative biologic therapy.

2. Methods

A retrospective review was performed on adult [>18 years] CD patients undergoing ICR at Cleveland Clinic Foundation [18 hospitals in Northeast Ohio and Florida] between January 1, 2009, and January 1, 2019. CD population was defined using standard ICD-9 [555.0, 555.1, 555.2, 555.9] and ICD-10 codes [K50.0, K50.1, K50.8, K50.9] for Crohn’s disease. Surgical intervention for ileocaecal resection or neo-ileocaecal resection was identified using CPT codes [44160, 44140, 44204, 44205]. Manual chart review, including clinician documentation, operative reports documented at time of surgery, preoperative laboratory values, and imaging reports, was performed to confirm the accuracy of the CD diagnosis, surgical intervention, and relevant clinical data. Surgeries utilizing a primary anastomosis and anastomosis with temporary diverting ileostomy were included in the study population, whereas those with end ileostomy were excluded.

Preoperative biologic exposure within 12 weeks was categorised as no biologic therapy, anti-TNF [infliximab, adalimumab, certolizumb pegol, or golimumab], VDZ, or UST. The preoperative period of 12 weeks was determined based on biologic pharmacokinetics and previous studies on postoperative complications in the setting of preoperative biologic exposure.^15–17 Exposure to biologics during this time period was confirmed via manual chart review. If a patient received multiple biologics during this 12-week time period, then the most recent biologic exposure was assigned.

2.1. Clinical data

Data collected included patient demographics (age, gender, active tobacco use at surgery, diabetes mellitus, body mass index [BMI]), disease characteristics [disease duration, disease behaviour, disease location, perianal disease according to Montreal Classification],²⁰ preoperative biologic use [none, anti-TNF, VDZ, or UST], number of biologics used prior to surgery, preoperative systemic steroid or immunomodulator use within 4 weeks of the surgery, preoperative laboratory values (haemoglobin, white blood cells [WBC], platelets, albumin, C-reactive protein [CRP]). Surgical data included anastomosis type [end-end, end-side, side-side], anastomotic technique [handsewn or stapled], surgical approach [hand-assist, laparoscopic, and/or open], and diversion technique if applicable.

2.2. Outcomes

The primary endpoint was the 90-day rate of IASC, defined as intra-abdominal abscess or leak. These outcomes were defined by the presence of clinical signs [fever, abdominal pain], radiographic evidence of intra-abdominal free fluid, gas, and/or abscess on cross sectional imaging obtained postoperatively or clinician designation of an anastomotic leak, or those recognised on reoperation. Secondary outcomes included superficial surgical site infection, 90-day mortality, 90-day ileus defined by symptomatology, radiographic evidence and clinician documentation, 90-day return to operative room , 90-day hospital readmission, and non-surgical infectious complications as defined by review of clinical documentation. In the case of ICR with diversion, the outcomes were measured from the initial surgical intervention.

2.3. Statistical analysis

Categorical variables were presented as proportions and compared with chi square or Fisher’s exact test where appropriate. Continuous variables were presented as mean with standard deviation or median with interquartile range based on normality. Comparisons were made via analysis of variance [ANOVA] or Kruskall‐Wallis tests where appropriate. Pairwise group comparison [no biologic vs anti-TNF, VDZ, or UST] were performed at a Bonferroni-adjusted p-value of 0.017 when overall p-values were < 0.05. Outcomes were compared across biologic exposure groups. Univariate logistic regression modelling was conducted to identify factors associated with occurrence of IASC. Multivariable generalised estimating equation logistic regression models were created to assess the association of preoperative biologic exposure on IASC. Covariates were selected by the best subset method and accounted for multiple surgeries for the same patient. Based on a clinical decision, preoperative systemic steroid use was forced into the final model. The number of independent variables in the model was determined following the 10 events per variable rule. All comparisons were made at a significance level of 0.05. All analyses were performed with R version 3.6.1.

2.4. Ethical considerations

The Cleveland Clinic Institutional Review Board approved this study.

3. Results

3.1. Study population

Of the overall 815 adult CD patients who underwent ICR that were included, 810 patients were assessed for the primary endpoint, IASC. The median age at surgery was 36 [27, 49] years, and the age at diagnosis was 23 [17, 30] years [Table 1]. Active tobacco use was present in 22.4% patients. The median BMI was 23.6, with 49% of patients having a normal BMI and 40.3% being obese. Disease behaviour was primarily stricturing [39.4%], penetrating [22.3%], or a combination of the two [31.2%]. Approximately 17% of patients had perianal disease at the time of surgery and about one-third of patients had previous perianal disease. Rates of preoperative corticosteroid and immunomodulator use within 4 weeks of surgery were 29.1% and 20.2%, respectively.

Table 1.

Demographics and clinical characteristics by current biologic status.

	All N = 815□	No bio N = 505	Anti-TNF N = 257	VDZ N = 32	UST N = 21	p-value
Median age at surgery [years, IQR]	36.0 [27.0;49.0]	38.0 [27.0;51.0]	33.0 [25.0;44.0] a	36.0 [31.0;44.5]	31.0 [25.0;39.0]	0.002
Median age at diagnosis [years, IQR]	23.0 [17.0;30.0]	24.0 [18.0;33.0]	21.0 [16.0;28.0] a	19.0 [15.8;23.0] a	16.0 [12.0;25.0] a	<0.001
Median disease duration [years, IQR]	10.0 [4.00;19.0]	10.0 [3.00;19.0]	9.00 [4.00;17.0]	17.5 [13.8;22.2] a	13.0 [7.00;21.0]	0.001
Female [N, %]	444 [54.5%]	280 [55.4%]	134 [52.1%]	18 [56.2%]	12 [57.1%]	0.84
Median BMI [kg/m2, IQR]	23.6 [20.6;27.7]	23.9 [20.8;27.8]	22.9 [20.3;26.2]	23.9 [20.9;27.1]	25.9 [20.7;28.9]	0.04
BMI category [N, %]:						0.16
Underweight	86 [10.6%]	47 [9.40%]	34 [13.3%]	4 [12.5%]	1 [4.76%]
Normal weight	396 [49.0%]	239 [47.8%]	134 [52.5%]	15 [46.9%]	8 [38.1%]
Obese	326 [40.3%]	214 [42.8%]	87 [34.1%]	13 [40.6%]	12 [57.1%]
Tobacco Use [N, %]:						0.17
Former	156 [19.2%]	98 [19.5%]	47 [18.3%]	9 [28.1%]	2 [10.0%]
Active	182 [22.4%]	126 [25.1%]	48 [18.7%]	4 [12.5%]	4 [20.0%]
Diabetes mellitus [N, %]	50 [6.16%]	38 [7.57%]	9 [3.50%]	3 [9.38%]	0 [0.00%]	0.07
CD location [N, %]:						0.35
Colon	33 [4.05%]	24 [4.76%]	6 [2.33%]	2 [6.25%]	1 [4.76%]
Ileocolon	484 [59.5%]	291 [57.7%]	157 [61.1%]	21 [65.6%]	15 [71.4%]
TI	297 [36.5%]	189 [37.5%]	94 [36.6%]	9 [28.1%]	5 [23.8%]
Upper GI [N, %]	154 [18.9%]	91 [18.1%]	51 [19.8%]	7 [21.9%]	5 [23.8%]	0.75
CD behaviour [N, %]:
Inflammatory	58 [7.13%]	40 [7.95%]	16 [6.23%]	0 [0.00%]	2 [9.52%]
Penetrating	181 [22.3%]	114 [22.7%]	56 [21.8%]	4 [12.5%]	7 [33.3%]
Stricturing	320 [39.4%]	192 [38.2%]	108 [42.0%]	14 [43.8%]	6 [28.6%]
Stricturing + penetrating	254 [31.2%]	157 [31.2%]	77 [30.0%]	14 [43.8%]	6 [28.6%]
Preoperative steroid use [N, %]						0.60
Budesonide/No	577 [70.9%]	356 [70.5%]	187 [72.8%]	19 [61.3%]	15 [71.4%]
Systemic	237 [29.1%]	149 [29.5%]	70 [27.2%]	12 [38.7%]	6 [28.6%]
Preop IMM use [N, %]	164 [20.2%]	91 [18.1%]	56 [22.0%]	9 [28.1%]	8 [38.1%]	0.06
Median number of prior biologics [N, IQR]	1.00 [0.00;2.00]	0.00 [0.00;1.00]	1.00 [1.00;2.00]	3.00 [2.00;4.00	4.00 [3.00;4.00]	<0.001
Preop Hb [g/dL, IQR]	11.9 [10.3;13.3]	11.8 [10.4;13.1]	12.2 [10.3;13.5]	12.0 [10.0;13.1]	10.8 [9.9;12.0]	0.27
Preop WBC [cells/L, IQR]	9.6 [7.1;12.9]	9.6 [7.2;13.0]	9.6 [6.9;12.9]	9.1 [7.5;12.0]	8.3 [5.8;10.3]	0.49
Preop Plt [Plt/L, IQR]	305 [240;383]	306 [243;384]	298 [237;373]	324 [262;444]	349 [299;407]	0.08
Preop Alb [g/dL, IQR]	3.8 [3.4;4.2]	3.9 [3.4;4.2]	3.7 [3.3;4.1]	3.6 [3.2;4.1]	3.90[3.5;4.4]	0.14
Preop Alb						0.50
<3.5 g/dL	175 [29.6%]	101 [27.6%]	60 [32.6%]	10 [37.0%]	4 [26.7%]
≥3.5 g/dL	417 [70.4%]	265 [72.4%]	124 [67.4%]	17 [63.0%]	11 [73.3%]
Preop CRP [mg/L, IQR]	2.10 [0.80;6.43]	1.90 [0.80;6.40]	3.00 [1.60;7.15]	1.70 [1.10;2.10]	4.20 [2.20;6.50]	0.64
Prior perianal disease [N, %]	275 [33.9%]	152 [30.3%]	93 [36.2%]	18 [56.2%]	12 [60.0%]	0.001
Perianal disease at time of surgery [N, %]	139 [17.1%]	76 [15.1%]	49 [19.1%]	8 [25.0%]	6 [28.6%]	0.12
Prior ileocolonic resections [N, %]	374 [45.9%]	220 [43.6%]	119 [46.3%]	22 [68.8%]	13 [61.9%]	0.02

BMI, body mass index; No bio, no biologic treatment; IMM, immunomodulator; Hb, haemoglobin; WBC, white blood cell; Alb, albumin; CRP, C-reactive protein; CD, Crohn’s disease; VDZ: vedolizumab; UST: ustekinumab; TNF: tumour necrosis factor; TI, terminal ileum; Plt, platelets; IQR, interquartile range; GI, gastrointestinal; preop, preoperative.^aSignificantly different from ‘no bio’, a significance level of 0.017 was used for pairwise ad-hoc comparisons [no bio vs anti-TNF, no bio vs VDZ, and no bio vs UST].

^□Of 815 patients, there is a lack of data regarding the presence of anastomosis and/or ostomy in eight patients and on data regarding IASC in five patients.

Of the 815 included patients, 62% [n = 505] received no biologic therapy, 31.4% [n = 257] received anti-TNF, 3.9% [n = 32] received VDZ, and 2.6% [n = 21] received UST within 12 weeks of surgery [Table 1]. There were significant differences in age at diagnosis [p < 0.001], previous perianal disease [p = 0.001], and BMI [p = 0.04] by biologic exposure, and previous ileocolic resections [p = 0.02]. Anti-TNF treated patients were older compared with those treated with no biologics [p < 0.017] [Table 1]. Patients treated with VDZ had longer disease durations compared with those treated without biologics [p < 0.017]. Those on UST or VDZ were more likely to be previously on various anti-TNF therapies before initiation [p < 0.001]. Preoperative steroid use was similar among the groups [p = 0.6], and UST-exposed patients had the highest rate of preoperative immunomodulator use [38.1%, vs 28.1% VDZ, vs 22% anti-TNF, vs 18.1% no biologic] [p = 0.06].

A majority of patients [85.9%] underwent an anastomosis. There was a significant difference in rates of anastomosis among the biologic-exposed groups [p = 0.003], and UST-treated patients had significantly fewer anastomoses when compared with the no biologic group [p < 0.017] [Table 2]. The VDZ and UST group had significantly more anastomoses with a temporary ostomy compared with the no biologic group [p < 0.017]. Surgical approach was evenly distributed between laparoscopic [46.7%] or open [53.1%]. The majority of patients had a side-to-side anastomosis performed [49%] and most anastomoses were stapled [77.2%], but this did not differ by biologic type [p = 0.07].

Table 2.

Operative characteristics.

	All	No bio	Anti-TNF	VDZ	UST	p-value
	N = 815	N = 505	N = 257	N = 32	N = 21
Diversion technique [N, %]:						0.03
End ileostomy	94 [28.5%]	49 [27.4%]	30 [25.4%]	7 [31.8%]	8 [72.7%]
Loop ileostomy	236 [71.5%]	130 [72.6%]	88 [74.6%]	15 [68.2%]	3 [27.3%]
Surgical approach [N, %]:
Hand assist	1 [0.14%]	0 [0.00%]	1 [0.42%]	0 [0.00%]	0 [0.00%]
Lap	337 [46.7%]	199 [45.6%]	115 [48.7%]	11 [39.3%]	12 [57.1%]
Open	383 [53.1%]	237 [54.4%]	120 [50.8%]	17 [60.7%]	9 [42.9%]
Anastomosis[N, %]	700 [85.9%]	441 [87.3%]	222 [86.4%]	25 [78.1%]	12 [57.1%]	0.003
Anastomosis +/- ostomy [N, %]	807
Anastomosis with temporary ostomy	223 [27.6%]	122 [24.5%]	84 [32.8%]	15 [46.9%] a	2 [9.5%] a
Anastomosis without ostomy	477 [59.1%]	319 [64.1%]	138 [53.9%]	10 [31.2%]	10 [47.6%]
No anastomosis with temporary ostomy	107 [13.3%]	57 [11.4%]	34 [13.3%]	7 [21.9%]	9 [42.9%]
Anastomosis type [N, %]:						0.80
End-to-end	173 [25.0%]	108 [24.8%]	59 [26.9%]	4 [16.0%]	2 [18.2%]
End-to-side	180 [26.0%]	120 [27.5%]	52 [23.7%]	5 [20.0%]	3 [27.3%]
Side-to-side	337 [49.0%]	208 [47.8%]	108 [49.3%]	16 [64.0%]	6 [54.5%]
Anastomosis technique [N, %]:						0.07
Handsewn	159 [22.8%]	106 [24.1%]	51 [23.1%]	1 [4.00%]	1 [9.09%]
Stapled	537 [77.2%]	333 [75.9%]	170 [76.9%]	24 [96.0%]	10 [90.9%]

No bio, no biologic treatment; VDZ: vedolizumab; UST: ustekinumab; TNF: tumour necrosis factor; Lap, laparoscopic.

^aSignificantly different from ‘no bio’, a significance level of 0.017 was used for pairwise ad-hoc comparisons [no bio vs anti-TNF, no bio vs VDZ, and no bio vs UST]

3.2. Postoperative complications

IASC, the primary endpoint, occurred in 12% of the patient cohort [n = 814] [Table 3]. The most common IASC was intra-abdominal abscess, occurring in 9.6%. There was no significant difference in overall IASCs rates in the UST [4.8%], VDZ [18.8%], anti-TNF [14.6%], and no biologic [10.6%] groups [p = 0.17]. However, rate of intra-abdominal abscesses was significantly different among biologic-exposed groups [p = 0.04], and the highest rates occurred in VDZ [18.8%] patients compared with anti-TNF [12.5%], UST [4.8%], or no biologic [7.7%] patients. There were no anastomotic leaks in VDZ- and UST-treated patients. There were similar rates of superficial wound infections [p = 0.89] and non-surgical infectious complications across the groups [p = 0.78]. There was one 90-day mortality due to septic shock from bacterial peritonitis in a 36 year-old patient with cirrhosis, who underwent ileocolic anastomosis with diverting ostomy without use of preoperative biologic therapy. The rates of 90-day ileus were highest in the VDZ-treated group [34.4%] and the lowest in the UST-treated group [9.5%], but there was no significant difference among the groups [p = 0.08].

Table 3.

90-day outcomes among preoperative biologic groups.

	All	No bio	Anti-TNF	VDZ	UST	p-value
	N = 814	N = 504	N = 257	N = 32	N = 21
Intra-abdominal sepsis [N, %]	97 [12.0%]	53 [10.6%]	37 [14.6%]	6 [18.8%]	1 [4.76%]	0.17
Intra-abdominal Abscess [N, %]	78 [9.58%]	39 [7.74%]	32 [12.5%]	6 [18.8%]	1 [4.8%]	0.04
Leak [N, %]	26 [3.22%]	20 [3.98%]	6 [2.37%]	0 [0.00%]	0 [0.00%]	0.51
Superficial wound infection [N, %]	61 [7.49%]	39 [7.74%]	20 [7.78%]	1 [3.12%]	1 [4.76%]	0.89
Mortality [N, %]	1 [0.12%]	1 [0.20%]	0 [0.00%]	0 [0.00%]	0 [0.00%]	1.00
Ileus [N, %]	213 [26.2%]	142 [28.2%]	58 [22.6%]	11 [34.4%]	2 [9.52%]	0.08
Return OR [N, %]	63 [7.75%]	37 [7.34%]	22 [8.59%]	2 [6.25%]	2 [9.52%]	0.85
Readmit [N, %]	180 [22.1%]	98 [19.5%]	63 [24.5%]	12 [37.5%]	7 [33.3%]	0.03
Any infectious complication [N, %]	200 [24.6%]	116 [23.0%]	70 [27.2%]	11 [34.4%]	3 [14.3%]	0.21
Non-surgical infectious complications [N, %]	95 [11.7%]	55 [10.9%]	34 [13.3%]	4 [12.5%]	2 [9.52%]	0.78
Surgical infectious complications [N, %]	140 [17.2%]	80 [15.9%]	52 [20.2%]	7 [21.9%]	1 [4.76%]	0.17

OR, operating room; VDZ: vedolizumab; UST: ustekinumab; TNF: tumour necrosis factor;

3.3. Factors associated with intra-abdominal septic complications

Of the total study population [n = 810], 97 patients experienced IASC after ileocaecal resection for CD [Table 4]. On univariate analysis of patients, risk factors that were associated with IASC included preoperative anti-TNF treatment (p = 0.04; OR 1.61 [1.02, 2.55]), longer disease duration (p = 0.04; OR 1.61 [1.02, 2.55]), preoperative a (p = 0.03; OR 0.88 [0.79, 0.99]), preoperative WBC (0.02; OR 1.04 [1.01, 1.08]), and preoperative albumin < 3.5 g/dl (p = 0.01; OR 1.98 [1.15, 3.41]) [Table 4]. Multivariable logistic regression models revealed that the only significant risk factor independently associated with IASC was preoperative albumin < 3.5 g/dl (p = 0.04; OR 0.57 [0.33,0.97]) when including preoperative biologic type, preoperative albumin, preoperative steroid use, and CD behaviour. Anti-TNF (p = 0.21; OR 1.39 [0.83, 2.34]), VDZ (p = 0.17; OR 2.06 [0.73, 5.83]), and UST (p = 0.52; OR 0.33 [0.01, 9.30]) were not significantly associated with IASC. Given the high rates of diverting ostomy in the VDZ and UST groups, IASC rates were assessed in patients with an anastomosis without ostomy. There was no statistical difference in rates of IASC in the VDZ [0%; p = 0.61] and UST [0%; p = 0.61] groups compared with the no biologic group. Furthermore, there was no statistical difference in any other postoperative surgical outcomes in the VDZ or UST versus no biologic groups when excluding those with ostomy.

Table 4.

Risk factors associated with intra-abdominal sepsis in patients with anastomosis and/or diverting ostomy.

	No intra-abdominal sepsis	Intra-abdominal sepsis	Univariate OR [95% CI]	p-value	*Multivariate OR [95%, CI]	p-value
	N = 713□	N = 97□			N = 585
Preoperative biologic type [N, %]
No biologic	451 [63.3%]	53 [54.6%]	Reference		Reference
Anti-TNF	216 [30.4%]	37 [38.1%]	1.61 [1.02, 2.55]	0.04	1.39 [0.83, 2.34]	0.21
VDZ	26 [3.66%]	6 [6.19%]	2.12 [0.78, 5.75]	0.14	2.06 [0.73, 5.83]	0.17
UST	20 [2.81%]	1 [1.03%]	0.21 [0, 13.5]	0.47	0.33 [0.01, 9.30]	0.52
Median age at surgery [years, IQR]	35.0 [26.0;49.0]	37.0 [30.0;48.0]	1.01 [0.99, 1.02]	0.3
Gender: female [N, %]	386 [54.1%]	55 [56.7%]
Median disease duration [years, IQR]	10.0 [3.00;19.0]	12.5 [5.00;22.0]	1.61 [1.02, 2.55]	0.04
CD location [N, %]:
Colon	28 [3.93%]	4 [4.12%]	Reference
Ileocolon	417 [58.6%]	65 [67.0%]	0.94 [0.31, 2.85]	0.91
TI	267 [37.5%]	28 [28.9%]	0.66 [0.21, 2.08]	0.48
CD behaviour [N, %]:
Inflammatory	53 [7.45%]	4 [4.12%]	Reference		Reference
Penetrating	153 [21.5%]	27 [27.8%]	2.11 [0.7, 6.42]	0.19	5.06 [0.64, 40.13]	0.13
Stricturing	291 [40.9%]	27 [27.8%]	1.15 [0.38, 3.48]	0.8	2.86 [0.36, 23.01]	0.32
Stricturing + penetrating	214 [30.1%]	39 [40.2%]	2.1 [0.71, 6.24]	0.18	4.76 [0.61, 37.41]	0.14
Prior perianal disease [N, %]	233 [32.9%]	39 [40.2%]	1.34 [0.85, 2.11]	0.21
Perianal disease at time of surgery [N, %]	121 [17.0%]	16 [16.5%]	0.93 [0.51, 1.69]	0.81
Tobacco use [N, %]:
Active	157 [22.1%]	23 [23.7%]	Reference
Former	131 [18.5%]	24 [24.7%]	1.45[0.84, 2.52]	0.18
Never	421 [59.4%]	50 [51.5%]	1.27 [0.73, 2.22]	0.39
Diabetes mellitus [N, %]	43 [6.06%]	7 [7.22%]	1.33 [0.58, 3.03]	0.50
Median BMI [kg/m2, IQR]	23.5 [20.6;27.5]	23.8 [21.1;28.4]	1.01 [0.97, 1.05]	0.51
Median number of prior biologics [N, IQR]	1.00 [0.00;2.00]	1.00 [0.00;2.00]	1.11 [0.93, 1.33]	0.23
Preoperative steroid use [N, %]:
Other/no	509 [71.5%]	65 [67.0%]	Reference		Reference
Systemic	203 [28.5%]	32 [33.0%]	1.17 [0.73, 1.89]	0.52	1.07 [0.63, 1.84]	0.80
Preop IMM use [N, %]	149 [21.0%]	15 [15.5%]	0.71 [0.39, 1.32]	0.28
Preop Hb [g/dL, IQR]	12.0 [10.4;13.3]	11.2 [10.0;12.7]	0.9 [0.8, 1.0]	0.03
Preop WBC [cells/L, IQR]	9.39 [7.05;12.70]	10.60 [7.47;14.60]	1.04 [1.01, 1.08]	0.02
Preop Plt [Plt/L, IQR]	302 [241;381]	320 [231;419]	1 [1, 1]	0.19
Preop Albumin**
<3.5 g/dL	140 [27.6%]	33 [42.3%]	1.98 [1.15, 3.41]	0.01	1.76 [1.03, 3.01]	0.04
≥3.5 g/dL	368 [72.4%]	45 [57.7%]	Reference		Reference
Preop CRP [mg/L, IQR]	2.10 [0.95;6.40]	1.70 [0.80;8.05]	1 [0.96, 1.03]	0.80
Anastomosis +/- ostomy [N, %]	802
Anastomosis with temporary ostomy	194 [27.4%]	29 [30.5%]	Reference
Anastomosis without ostomy	426 [60.3%]	46 [48.4%]	0.75 [0.45, 1.24]	0.26
No anastomosis with temporary ostomy	87 [12.3%]	20 [21.1%]	1.44 [0.74, 2.79]	0.28
Anastomosis technique and approach [N, %]:
Handsewn	137 [22.2%]	22 [29.3%]	Reference
Stapled	479 [77.8%]	53 [70.7%]	0.61 [0.35 , 1.06]	0.08
Lap	301 [47.6%]	36 [42.4%]	0.83 [0.52, 1.34]	0.45
No lap	331 [52.4%]	49 [57.6%]	Reference

OR, odds ratio; CI, confidence interval. BMI, body mass index; IMM, immunomodulator; Hb, haemoglobin; WBC, white blood cells; Alb, albumin; CRP, C-reactive protein; CD, Crohn’s disease; VDZ, vedolizumab; UST, ustekinumab; TI, terminal ileum; TNF, tumour necrosis factor; Lap, Laparoscopic; IQR interquartile range; Plt, platelets; CD, Crohn’s disease; Preop, preoperative.

^□Of 815 patients, there is a lack of data regarding IASC in 5 patients

*Multivariate analysis: 585 patients included with complete available data for all risk factors in model. Preoperative steroid use and albumin were included in the model based on clinical significance. CD behaviour was selected by the subset method and included in the model.

**Preoperative albumin as a continuous variable was also included in the analysis and was insignificant [p = 0.13].

4. Discussion

In this retrospective analysis of 810 CD patients undergoing ileocaecal resection within a large health care system, preoperative biologic exposure, including more recent biologic agents, was not associated with postoperative intra-abdominal septic complications. These data provide reassurance regarding biologic exposure in the preoperative period. Additionally, preoperative hypoalbuminaemia was independently associated with perioperative complications, potentially identifying an at-risk population, a finding which has been observed by others.^21–24

Early controversy over the use of preoperative biologics in CD patients undergoing intra-abdominal surgery raised concerns of post-surgical infectious complication rates.^25,26 These postoperative complications increase patient morbidity, mortality, and length of hospital stay.²⁷ This lack of effect of preoperative biologic exposure on post-surgical outcomes was specifically seen in patients with ileocolic resection for CD and may not generalise to other operative interventions. Future studies are required in specific surgical populations for the effect of newer preoperative biologic exposure on postsurgical outcomes.

Early data suggested anti-TNF exposure may increase postoperative complication rates including intra-abdominal septic complications. The GETAID study prospectively observed that anti-TNF use < 3 months preceding surgery was associated with a higher risk of overall and intra-abdominal septic postoperative morbidity.²⁶ Subsequent studies from numerous centres, including meta-analyses and the largest prospective study to date, PUCCINI, have countered this experience.^2,11,12 The reason for observed differences is likely multifactorial, including study design and definitions, unmeasured confounders [eg, malnutrition, corticosteroid use], differences in clinical management or surgical paradigms, and sample size. Our data support the lack of association of anti-TNFs with surgical complications for this ongoing debate. In our study, rates of complications were higher in anti-TNF exposed patients; however, after controlling for confounding variables such as corticosteroids, no association was observed.

Previous studies on surgical risk in biologic-exposed patients largely investigated anti-TNFs. For VDZ and UST, limited data exist on the postoperative infectious complication rates after major abdominal surgery for CD. Our results indicate that VDZ is not associated with IASC when factoring in corticosteroid use, preoperative albumin, and CD behaviour in a multivariable regression model of CD patients undergoing ileocaecal resection. This homogeneous study population assessment agrees with a recent meta-analysis suggesting that there is no association between preoperative biologic use and postoperative infectious complications in the broad context of IBD patients undergoing abdominal surgery.¹² Our study adds to the limited published literature focused specifically on the CD population rather than all IBD patients, creating a more homogeneous population.^12–14 Regardless of the presence of an ostomy with anastomosis formation, there were no significant differences in postoperative surgical outcomes in the VDZ group [Table 5]. Presuming that patients without ostomy formation have less complicated disease, this finding further suggests that VDZ does not contribute to significant post-surgical morbidity. Interestingly, VDZ-treated patients experienced a 34.4% 90-day ileus rate. This high rate of ileus in VDZ-exposed patients was also observed in a separate cohort, at 36%.²⁸ This finding, if confirmed in larger VDZ-exposed sample sizes, may have a unique pathophysiological mechanism, given the gut-specific nature of VDZ; and may identify a patient population for proactive ileus preventative measures such as use of alvimopan.²⁹

Table 5.

Outcomes across VDZ versus non-biologic treated patients with anastomosis [with and without ostomy].

	Anastomosis with temporary ostomy				Anastomosis, no ostomy
	[ALL]	No bio	VDZ	p	[ALL]	No bio	VDZ	p
	N = 137	N = 122	N = 15		N = 328	N = 318	N = 10
Intra-abdominal sepsis	18 [13.1%]	15 [12.3%]	3 [20.0%]	0.418	29 [8.90%]	29 [9.18%]	0 [0.00%]	0.609
Intra-abdominal abscess	11 [8.03%]	8 [6.56%]	3 [20.0%]	0.103	24 [7.32%]	24 [7.55%]	0 [0.00%]	1.000
Leak	9 [6.57%]	9 [7.38%]	0 [0.00%]	0.597	9 [2.76%]	9 [2.85%]	0 [0.00%]	1.000
Superficial wound infection	12 [8.76%]	12 [9.84%]	0 [0.00%]	0.361	19 [5.79%]	18 [5.66%]	1 [10.0%]	0.454
90-day mortality	0 [0.00%]	0 [0.00%]	0 [0.00%]	.	0 [0.00%]	0 [0.00%]	0 [0.00%]	.
90-day ileus	48 [35.0%]	42 [34.4%]	6 [40.0%]	0.888	85 [25.9%]	83 [26.1%]	2 [20.0%]	1.000
Return OR 90 days	19 [13.9%]	17 [13.9%]	2 [13.3%]	1.000	13 [3.96%]	13 [4.09%]	0 [0.00%]	1.000
Readmit 90 days	43 [31.4%]	37 [30.3%]	6 [40.0%]	0.556	42 [12.8%]	41 [12.9%]	1 [10.0%]	1.000
Any infectious complication	43 [31.4%]	39 [32.0%]	4 [26.7%]	0.776	60 [18.3%]	58 [18.2%]	2 [20.0%]	1.000
Non-surgical infectious complication	21 [15.3%]	20 [16.4%]	1 [6.67%]	0.467	25 [7.65%]	24 [7.57%]	1 [10.0%]	0.554
Surgical infectious complication	26 [19.1%]	23 [19.0%]	3 [20.0%]	1.000	44 [13.4%]	43 [13.5%]	1 [10.0%]	1.000

VDZ, vedolizumab; No bio, no biologic treatment; OR, operating room.

There are controversial reports, although limited by sample size, on the perioperative infectious rates of CD patients undergoing abdominal surgery with the use of perioperative UST.^16,18,19 In the largest cohort of 57 ustekinumab-exposed CD patients undergoing any intra-abdominal surgery, UST was associated with an increased rate of IASC even when accounting for corticosteroid exposure,¹⁷ but not with superficial surgical site infections or overall postoperative infectious complications. On the contrary, our results limited to ileocaecal resections suggest that perioperative use of UST had no association with IASC in patients with anastomosis while accounting for albumin, corticosteroid use, and CD behaviour. Only 1/21 [4.8%] of patients had IASC secondary to intra-abdominal abscess formation. One patient [4.8%] developed a superficial wound infection. It should be noted that UST-exposed patients had significantly higher rates of diversion, which may convey that this population had more repeat surgeries or more complex disease, but also perhaps a protective effect against IASC as was observed by Neary et al.³⁰ When assessing UST-treated patients without ostomy formation, albeit with a limited sample size, there were similar rates of IASC and post-surgical morbidities compared with those treated with no biologic. This finding further suggests that preoperative UST does not increase rates of IASC. Despite having a low event rate in a limited cohort, our study adds to the existing controversial data regarding perioperative UST use in CD patients. Ultimately, a prospective cohort study on the perioperative use of UST would help ascertain the true association.

In the current study, preoperative hypoalbuminaemia < 3albuminaemia.5 g/dl was an independent risk factor for IASC, which aligns with earlier reports.^15,21–24 Preoperative albumin may be reflective of the severity of systemic inflammation, protein-losing enterocolopathy, malnutrition, or concurrent liver dysfunction, and may fluctuate accordingly. Other confounding factors, such as preoperative biologic exposure or other chronic inflammatory conditions, were not accounted for as aetiologies of hypoalbuminaemia. Data regarding concomitant dietician assessments of nutrition or presence and severity of liver disease were unavailable. Although hypoalbuminaemia is not a direct marker of preoperative nutritional status, if malnutrition is present, preoperative nutritional optimisation via enteral or parenteral routes may be considered and has been shown to improve outcomes.^31,32 Future studies may assess preoperative markers and evaluations of malnutrition and optimise the patient accordingly.

In our study, we did not observe an association of preoperative corticosteroid exposure, defined as exposure within 12 weeks preceding surgery, with postoperative infectious complications. The association of preoperative corticosteroid use and postoperative complications was not seen as it was in previous studies.^15,33–37 Our findings may be confounded by the lack of specification of dosage and duration of corticosteroid use preoperatively, as corticosteroid use was dichotomously recorded and may have diluted an effect. Similarly, the preoperative time period of 12 weeks is longer than other studies that have used 4 weeks.^33–35 Additionally, our cohort was limited by the small sample size of patients using preoperative corticosteroids.

On univariate analysis, diverting ostomy use was not protective against IASC, as suggested by Neary et al.³⁰ The clinical use of this potentially protective diversion may influence the IASC rates seen in this study and potentially explain the lack of significant difference. This is suggested by unevenly distributed diverting ostomies performed in the VDZ and UST groups which had more previous biologic exposure and perianal disease, suggesting more refractory populations.

Interestingly, superficial wound infections were seen at similar rates in the non-biologic and biologic-exposed groups. There have been conflicting reports on whether preoperative VDZ or UST use is associated with postoperative superficial site infections in CD patients.^12–19 Based on our findings, biologics may not adversely affect wound healing or superficial wound infection risk and other factors may be implicated.

There are several limitations to this study. First, this is a retrospective chart review of a large tertiary system and our patient cohort may not be generalisable to other centres, patient populations, or surgical interventions. However, Cleveland Clinic comprises a large number of hospitals across both Ohio and Florida where both secondary and tertiary CD care [medical and surgical] is provided, potentially minimising a large referral bias. A significant proportion of our patients had multiple previous surgeries which may affect many of the operative approaches, limiting generalisability to other centres. There is great variability in preoperative through postoperative care among physicians, which was unaccounted for. Surgical details were documented in operative reports at the time of surgery, which may introduce reporting bias, though the operative details included were generally standardised. Selection bias is introduced by the nature of retrospective data collection. Another limitation is the low event rate, specifically in the VDZ and UST groups, and raises the possibility of Type II errors; however this was one of the largest series reported to date. Due to our homogeneous patient population limited to CD and ileocaecal resections, although fewer patients were included, the comparisons and outcomes are potentially more applicable. Furthermore, such a retrospective analysis has potential unaccounted confounders. Many patients treated with VDZ or UST may have been previously non-responsive to anti-TNF therapy, indicating more aggressive or refractory disease. Despite this, our results attempting to control for identifiable and likely confounders still did not indicate differences in IASC rates in CD patients treated with preoperative VDZ or UST. Another potential confounding factor is the duration of therapy preceding surgery, which may be further confounded by the varying half-lives of each of these biologics. Most patients did not have drug levels checked before surgery to confirm active biologic exposure; however, this is reflective of real-world clinical practice before surgery for disease complications. We did not have data regarding potential aetiologies of the hypoalbuminaemia, such as dietician assessment for malnutrition, faecal protein loss measurement, or concurrent liver disease. Although our study included multiple variables, we are unable to account for all of the variables that may affect CD and postoperative infectious outcomes.

In conclusion, preoperative exposure to biologic therapy was not associated with increased intra-abdominal septic complications in adult CD patients undergoing an a resection, whereas preoperative hypoalbuminaemia was independently associated. These data suggest that necessary surgical interventions should not be delayed based on preoperative biologic exposure. Interventions to identify, address, or mitigate risk in individuals with hypoalbuminaemia may improve perioperative outcomes. Larger, prospective studies are necessary to confirm these findings.

Data Availability Statement

The data underlying this article will be shared on reasonable request to the corresponding author.

Funding

None.

Conflict of Interest

MR receives the following financial support: research support from Abbvie, Janssen, Takeda, Pfizer; unrestricted educational grants from Abbvie, Janssen, UCB, Pfizer, Takeda, Celgene, Genentech, Gilead Advisory Boards; and as consultant for Abbvie, Janssen, UCB, Takeda, Pfizer, Miraca Labs, Amgen, Celgene, Seres, Allergan, Genentech, Gilead, Salix, Prometheus, Lilly, TARGET Pharma Solutions, ALFASIGMA, S.p.A., CME Companies: CME Outfitters, Imedex, GI Health Foundation [GiHF], Cornerstones, Remedy, MJH life sciences; royalties from Wolters.

Author Contributions

RSS: design of study, acquisition of data, interpretation of data, drafting and revising of article. SB: acquisition of data, interpretation of data, drafting and revising of article. XJ: acquisition of data, interpretation of data. SDH: patient recruitment, revising of article, interpretation of data. TLH: patient recruitment, revising of article. JPA: patient recruitment, revising of article. JP: patient recruitment, revising of article. TQ: patient recruitment, revising of article. FR: patient recruitment, revising of article. BLC: revising of article. MDR: patient recruitment, interpretation of data, revising of article. All authors: conception and design of study, patient recruitment, interpretation of data.