Patient characteristics in the derivation cohort
Altogether, 33 CD patients were identified as having IAA through radiography and underwent regular follow-up for at least 1 year. There were 27 (81.8%) males and 6 (18.26%) females with a median age at disease onset of 26.0 (20.0, 33.5) years. The median age at admission was 29.0 (23.5, 37.0) years, and the range from onset to admission was 3.0 (1.0, 7.5) years. Half of the patients were underweight, with a median body mass index (BMI) of 17.5 (15.0, 20.8) kg/m2. The most common symptoms were abdominal pain, fever, and abdominal mass, which occurred in 32 (96.7%), 29 (87.9%), and 17 (51.5%) patients, respectively. One patient was diagnosed with cytomegalovirus infection, and another 4 patients were diagnosed with Clostridium difficile infection. Median white blood cells, red blood cells and serum albumin levels were normal. Median platelet counts and inflammatory markers, including ESR and hsCRP levels, were elevated. Thirteen (39.4%) patients were on glucocorticoids and 5 patients were on immunosuppressants before IAA was detected.
Conservative medical treatment and invasive treatment
Initial therapeutic strategies included medical therapy with antibiotics. Seven patients underwent percutaneous drainage as the primary approach, and among them, 5 patients subsequently underwent colectomy. Eight patients underwent surgical intervention directly without previous percutaneous drainage. Altogether, 18 (54.5%) patients were assigned to the conservative medical treatment group, and 15 (45.5%) patients were assigned to the invasive treatment group. The percentage of male patients was higher in the conservative treatment group (94.4% vs. 66.7%, p = 0.039). There were significant differences in the range from onset of disease to admission between the two groups, with a p value of 0.049. The median range was longer in the invasive treatment group, which was 5.0 (3.0, 8.0) years. The median range was only 1.5 (1.0, 4.5) years in the conservative treatment group. There was no difference in smoking status, previous history of abdominal and/or pelvic surgery or perianal disease between the two groups. Additionally, no differences in white blood cells, hemoglobin, platelets, albumin, hsCRP or ESR were found between the two groups. Of those who underwent percutaneous drainage, the most common pathogenic microorganism was Escherichia coli, and other pathogenic microorganisms included Serratia marcescens, Enterococcus faecium, Pseudomonas aeruginosa, and Candida albicans. Of those who underwent surgical intervention, the most common surgical approach was partial bowel resection. The percentages of glucocorticoid, immunosuppressant and biologic use after IAA showed no difference between the two groups (Table 1).
Table 1
Patient charateristics in conservative and invasive treatment group in Crohn’s Disease related intra-abdominal abscess
Items | Total patients (N = 33) | Conservative treatment group (N = 20) | Invasive treatment group (N = 13) | p value |
Male, n(%) | 27(81.8%) | 17(94.4%) | 10(66.7%) | 0.039 |
Age at onset (years), IQR | 26.0(20.0,33.5) | 25.0(17.8,30.8) | 28.0(21.0,41.0) | 0.169 |
Age at admission (years), IQR | 29.0(23.5,37.0) | 27.0(21.0,34.0) | 30.0(29.0,45.0) | 0.057 |
Range from onset to admission, IQR | 3.0(1.0,7.5) | 1.5(1.0,4.5) | 5.0(3.0,8.0) | 0.049 |
Smoking status, n(%) | | | | 0.879 |
Never smoking | 22(66.7%) | 12(66.7%) | 10(66.7%) | |
Ex-smoker | 3(9.1%) | 1(5.6%) | 2(23.3%) | |
Current smoking | 8(24.2) | 5(27.8) | 3(20.0%) | |
Previous history of abdominal and/or pelvic surgery, n(%) | 11(33.3%) | 7(38.9%) | 4(26.7%) | 0.465 |
Perianal disease | 17(50.0%) | 8(42.1%) | 9(60.0%) | 0.381 |
BMI (kg/m2), mean ± SD | 17.5(15.0,20.8) | 18.9(14.9,20.8) | 17.4(15.8,21.2) | 0.914 |
WBC (×10^9/L), mean ± SD | 8.6(6.0,10.4) | 9.3(6.8,13.7) | 8.1(5.5,9.4) | 0.239 |
HGB (g/L), mean ± SD | 113.4 ± 24.4 | 117.4 ± 25.5 | 108.5 ± 22.8 | 0.299 |
PLT (×10^9/L), mean ± SD | 360.5(283.8,477.8) | 361.0(318.3,506.0) | 353.50(243.8,513.3) | 0.338 |
Serum albumin (g/L), mean ± SD | 37.1 ± 6.1 | 36.1 ± 6.1 | 38.3 ± 6.1 | 0.292 |
ESR (mm/h), mean ± SD | 38.0(26.3,49.5) | 38.0(24.0,49.3) | 37.0(26.3,54.3) | 0.856 |
hsCRP (mg/dL), mean ± SD | 54.9(25.9,83.9) | 65.4(38.3,99.4) | 33.7(7.6,60.6) | 0.096 |
Location, n (%) | | | | 0.001 |
perienteric abscess | 19(57.6) | 15(83.3) | 4(26.7) | |
non perienteric abscess | 14(42.4) | 3(16.7) | 11(73.3) | |
number of abscess, n (%) | | | | 0.808 |
single | 27(81.8) | 15(83.3) | 12(80.0) | |
multiple | 6(18.2) | 3(16.7) | 3(20.0) | |
max abscess diameter(mm), IQR | 49.0(33.0,62.5) | 39.0(25.0,54.3) | 50.0(39.0,144.0) | 0.027 |
Fistula, n (%) | | | | 0.140 |
single | 9(27.3) | 3(16.7) | 6(40.0) | |
multiple | 24(72.7) | 15(83.3) | 9(60.0) | |
length of sinus(mm, x ± s) | 15.4 ± 6.5 | 15.7 ± 6.5 | 15.1 ± 6.8 | 0.825 |
width of sinus(mm, x ± s) | 6.9 ± 2.4 | 5.8 ± 1.7 | 8.1 ± 2.5 | 0.004 |
Treatment before IAA diagnosis | | | | |
Glucocorticoid, n (%) | 13(39.4) | 5(27.8) | 8(53.3) | 0.141 |
Immunosuppresants, n (%) | 5(15.2) | 2(11.1) | 3(20.0) | 0.485 |
Treatment after IAA diagnosis | | | | |
Glucocorticoid, n (%) | 23(69.7) | 13(72.2) | 10(66.7) | 0.733 |
Immunosuppresants, n (%) | 16(48.5) | 7(38.9) | 9(60.0) | 0.234 |
Biologics, n (%) | 10(30.3) | 7(38.9) | 3(20.0) | 0.247 |
IQR interquartile range, SD standard deviation, BMI body mass index, WBC white blood cell, HGB hemoglobin, PLT platelet, ESR erythrocyte sedimentation rate, hsCRP hypersensitive C reactive protein |
Imaging Characteristics
Based on the enhanced CT scan, most lesions were segmental (24/33, 72.7%) instead of local. Mesenteric fat turbidity was found in each patient. Comb signs were found in 16 (48.5%) patients. Bowel stenosis was found in 29 (87.9%) patients, while bowel dilation was found in 9 (27.3%) patients.
Most abscesses were perienteric, accounting for 57.6% of all abscesses. Other abscesses were located in the retroperitoneal space, had access to the abdominal wall, or invaded adjacent organs, including the gallbladder or uterus and were classified as nonperienteric abscesses (Fig. 1). The invasive treatment group had more nonperienteric abscesses than the conservative medical treatment group (83.3% vs. 16.7%, p = 0.001). A total of 27 (81.8%) patients had a single abscess, while 6 (18.2%) patients had multiple abscesses. There was no difference in the number of abscesses between the two groups. The median maximum abscess diameter was 49.0 (33.0, 62.5) mm. The maximum abscess diameter was larger in the invasive treatment group (50.0 (39.0, 144.0) mm vs. 39.0 (25.0, 54.3) mm), with a p value of 0.027.
Fistulas were found in each patient. Nine (27.3%) patients had a single fistula, while 24 (72.7%) patients had 2 to more than 10 fistulas. There was no difference in the number of fistulas between the two treatment groups. The length of the sinus was almost the same in the two groups. However, the maximum width of the sinus was larger in the invasive treatment group (8.1 ± 2.5 mm vs. 5.8 ± 1.7 mm), with a p value of 0.004.
Overall survival (OS) without invasive treatment and prognostic factors
The median follow-up time was 52.0 (33.5, 60.0) months. No patients died by the time of the last follow-up. Six patients underwent percutaneous drainage within 1 month after diagnosis, while 1 patient underwent percutaneous drainage 34 months after diagnosis because of failure of medical treatment. The median time from diagnosis to surgery was 13.0 (4.5, 23.5) months. The median time from percutaneous drainage to surgery was 7.0 (3.5, 21.5) months. The 1-, 2-, and 5-year overall survival rates without invasive treatment were 69.7%, 57.3%, and 53.7%, respectively.
On univariate Cox analysis, nonperienteric abscess (HR: 6.391, 95% CI: 1.970-20.728, p = 0.002), max abscess diameter (HR: 1.015, 95% CI: 1.005–1.025, p = 0.003) and width of sinus (HR: 1.246, 95% CI: 1.037–1.498, p = 0.019) were significantly associated with worse OS without invasive treatment. Nonperienteric abscess was significantly associated with worse OS without invasive treatment on multivariate Cox analysis (HR: 5.738, 95% CI: 1.763–18.672, p = 0.004). (Table 2 and Fig. 2).
Table 2
Univariate and multivariate analyses on predictors for invasive treatment
Items | Univariate analysis | Multivariate analysis |
| HR (95% CI) | P value | HR (95% CI) | P value |
Non perienteric abscess | 6.391 (1.970, 20.728) | 0.002 | 5.738 (1.763, 18.672) | 0.004 |
Max abscess diameter | 1.015 (1.005, 1.025) | 0.003 | | |
Width of sinus | 1.246 (1.037, 1.498) | 0.019 | | |
HR hazard ratio, CI confdence interval |
Validation cohort
Altogether, 13 CD patients were identified in the validation cohort. They were 8 (61.5%) males and 5 (38.5%) females with a median age at admission of 29.0 (19.0, 40.0) years. The median follow-up time was 13.0 (12.0, 27.0) months. Three patients underwent percutaneous drainage as the primary approach, and among them, 2 patients subsequently underwent colectomy. Five patients underwent surgical intervention directly without previous percutaneous drainage. Altogether, 8 (61.5%) patients were assigned to the conservative medical treatment group, and 5 (38.5%) patients were assigned to the invasive treatment group. The 1-year overall survival without invasive treatment was 46.2%.
Nomogram construction and validation
A prognostic nomogram that integrated all the significant independent factors from the multivariate analysis in the whole cohort was constructed, as shown in Fig. 3. The nomogram was internally confirmed. The concordance index (C-index), calculated by bootstrap resampling, was 0.838 (95% CI: 0.736–0.922, p < 0.001), indicating a significant discriminative ability. Calibration curves for the probability of overall survival at 1 and 5 years showed a good correlation between the nomogram-predicted and actually observed values (Fig. 4a, 4b).
The nomogram was externally confirmed in the validation cohort. The C-index, calculated by bootstrap resampling, was 0.842 (95% CI: 0.681–0.983, p < 0.001), indicating a significant discriminative ability. A calibration curve plotted for the 1-year overall survival also showed an optimal agreement between nomogram prediction and actual observation (Fig. 4c). Discriminative demonstration with the use of the receiver operating characteristic curve in the validation cohort showed an AUC value of 0.875, a sensitivity of 80.0% and a specificity of 87.5% (Fig. 4d).