Association Between Body Composition Measured by Artificial Intelligence and the Long-Term Sequelae After Acute Pancreatitis
The American Journal of Gastroenterology(2023)
Abstract
Introduction: About 10-30% of patients with acute pancreatitis (AP) have a recurrent episode. The impact of body composition on recurrent AP is unclear. We aimed to determine the associations between body composition and the long-term sequelae after AP. Methods: We performed a retrospective study of patients hospitalized for AP in 2018 at 3 tertiary centers. Patients with CT imaging of the abdomen at admission were included. A fully automated abdominal segmentation algorithm developed by Weston et al.(Radiology 2019) was used for body composition analysis. The primary endpoint was a recurrent hospitalization for AP. Secondary endpoints included a diagnosis of chronic pancreatitis (CP) or diabetes mellitus (DM) in patients evaluated at follow-up. Cox Proportional Hazards regression analysis was used. Results: 347 patients (median follow-up time: 1035 days, IQR: 697-1515) were included. The most common etiologies were alcohol (28.8%), gallstone (24.2%) and idiopathic (25.1%). 89 patients (25.6%) had a recurrent episode (median time: 219 days, IQR: 74-471). 34 of 112 patients (30.4%) developed CP (median time: 311 days, IQR: 160-797) and 22 of 88 (25.0%) developed DM (median time: 1104 days, IQR: 507-1242). After adjusting for age, male sex, first episode of AP, BUN, and severity of AP; obesity, body mass index (BMI), alcohol pancreatitis, and gallstone pancreatitis were significantly associated with a recurrent episode (Table 1). Body composition was not associated with a recurrent episode. In unadjusted analysis, BMI (HR 0.66, P=0.0052), obesity (HR 0.36, P=0.0131), history of DM (HR 2.12, P=0.0371), and subcutaneous adipose tissue (SAT) (HR 0.87 per 10 cm2, P=0.002) were associated with a diagnosis of CP. Skeletal muscle (SM) mass approached significance for CP (HR 1.33 per 10 cm2, P=0.0546). Mean corpuscular volume (HR 0.45 per 10 fL, P=0.0209) and intermuscular adipose tissue (IMAT) (HR 1.45 per 5 cm2, P=0.0264) were associated with DM at follow-up (Figure 1). Conclusion: In this multi-site retrospective study, we found body mass composition was not associated with a recurrent episode of AP after adjusting for age, male sex, having an initial episode, and AP severity. At follow-up, 30% and 25% of patients developed CP and DM at approximately one and 3 years, respectively. In unadjusted analysis, a higher SAT and IMAT were associated with a lower incidence of CP and a higher incidence of DM, respectively. Further studies are needed to corroborate these findings and to examine other long-term sequelae.Figure 1.: Cumulative incidence of recurrent acute pancreatitis by A. Etiology of initial acute pancreatitis. B. Severity of initial episode. C. Visceral adipose tissue ≥ 100 cm2. D. Intermuscular adipose tissue ≥ 15 cm2. Table 1. - Baseline Characteristics According to Recurrent Episode of Acute Pancreatitis and Multivariable Cox Proportional Hazards Regression Analysis to Predict a Recurrent Episode Demographics All patientsN=347 No recurrenceN=258 Recurrent APN=89 P value Multivariable Cox† HR (95% CI) P value Age at admission, per 10 years 57.8 (43.2-69.7) 60.9 (47.2-71.0) 51.5 (38.8-60.9) < 0.001 0.81 (0.70-0.93) 0.0040 Male sex 211 (60.8%) 158 (61.2%) 53 (59.6%) 0.802 0.99 (0.64-1.55) 0.9900 White race 312 (89.9%) 236 (91.5%) 76 (85.4%) 0.106 0.69 (0.37-1.29) 0.2394 Body surface area, per 0.25 m2 2.0 (1.8-2.2) 2.0 (1.8-2.3) 2.0 (1.8-2.2) 0.187 0.79 (0.65-0.96) 0.0167 Body Mass Index, per 5 kg/m2 28.8 (25.2-33.4) 29.3 (25.5-33.9) 28.0 (24.3-32.4) 0.105 0.85 (0.73-0.99) 0.0333 Obesity 150 (43.2%) 120 (46.5%) 30 (33.7%) 0.047 0.55 (0.35-0.87) 0.0109 Smoking, never 135 (38.9%) 105 (40.7%) 30 (33.7%) 0.259 1.00 (0.63-1.58) 0.9979 Alcohol use 146 (46.9%) 110/230 (47.8%) 36/81 (44.4%) 0.608 0.86 (0.55-1.33) 0.4912 Vital signs and labs Systolic blood pressure, per 10 mmHg 138 (124-157) 139 (124-159) 134 (123-151) 0.249 0.85 (0.85-1.04) 0.2219 Diastolic blood pressure, per 10 mmHg 82 (71-94) 81 (71-94) 83 (74-96) 0.296 0.98 (0.84-1.14) 0.7734 Mean arterial pressure, per 5 mmHg 101.3 (90-114.3) 101.7 (89.7-114.7) 99.3 (93.3-112.0) 0.927 0.97 (0.90-1.04) 0.4369 Heart rate, per 10 beats per minute 86 (72-101) 85 (72-97) 92 (77-113) 0.019 1.09 (0.95-1.25) 0.2156 SIRS positive 109/296 (55.6%) 79/147 (53.7%) 30/49 (61.2%) 0.409 0.97 (0.53-1.76) 0.9167 White blood cells, per 1 x 109/L 11.9 (8.9-14.5) 11.9 (8.9-15.3) 12.2 (9.0-14.1) 0.622 0.96 (0.92-1.01) 0.1380 Hematocrit, per 5 % 41.4 (38.2-44.2) 41.4 (38.2-44.1) 41.4 (38.0-44.8) 0.941 0.98 (0.80-1.20) 0.8236 Hemoglobin, per 1 g/dL 14 (12.9-15.3) 14.0 (12.9-15.3) 14.0 (12.9-15.2) 0.852 1.00 (0.89-1.12) 0.9674 MCV, per 10 fL 90.6 (87.6-94.9) 90.6 (87.8-94.7) 90.3 (87.2-95.2) 0.983 0.94 (0.68-1.32) 0.7363 Platelets, per 50 x 109/L 232 (176-285) 232 (177-281) 229 (171-302) 0.966 1.00 (0.98-1.02) 0.8781 Sodium, per 1 mmol/L 138 (136-141) 138 (136-141) 138 (135-140) 0.116 0.99 (0.93-1.05) 0.7913 BUN, per 1 mg/dL 13 (10-21) 15 (11-21) 13 (9-18) 0.004 0.97 (0.95-1.00) 0.0265 Total bilirubin, per 1 mg/dL 0.8 (0.5-1.4) 0.8 (0.5-1.6) 0.7 (0.4-1.1) 0.042 0.98 (0.83-1.16) 0.8301 Direct bilirubin, per 1 mg/dL 0.4 (0.2-1.5) 0.5 (0.3-1.5) 0.2 (0.2-0.7) 0.018 1.00 (0.78-1.27) 0.9702 Lipase, per 100 U/L 485 (156-600) 504 (155-600) 456 (165-600) 0.725 1.00 (0.99-1.02) 0.6589 Pancreatitis First episode 206 (59.4%) 172 (66.7%) 34 (38.2%) < 0.001 0.45 (0.29-0.70) 0.0004 Alcohol, etiology 100 (28.8%) 56 (21.7%) 44 (49.4%) < 0.001 1.72 (1.07-2.77) 0.0260 Gallstone, etiology 84 (24.2%) 76 (29.5%) 8 (9.0%) < 0.001 0.45 (0.21-0.95) 0.0371 Idiopathic, etiology 87 (25.1%) 69 (26.7%) 18 (20.2%) 0.258 0.74 (0.44-1.26) 0.2708 Other, etiology 76 (21.9%) 57 (22.1%) 19 (21.3%) 1.000 1.21 (0.71-2.03) 0.4837 Necrotizing 39 (11.2%) 24 (9.3%) 15 (16.9%) 0.078 1.00 (0.51-1.95) 0.9997 Metabolic conditions Hypertension 220/295 (75.6%) 159/216 (73.6%) 61/79 (77.2%) 0.651 1.82 (1.03-3.22) 0.0396 Diabetes 84 (24.2%) 56 (21.7%) 28 (31.5%) 0.084 1.28 (0.81-2.02) 0.2945 Hyperlipidemia 179/295 (60.7%) 138 (63.9%) 41 (51.9%) 0.080 1.10 (0.66-1.83) 0.7123 Metabolic conditions, per 1 condition Ϯ 2 (1-2) 2 (1-2) 2 (1-2.5) 0.926 1.17 (0.93-1.48) 0.1955 ≥ 1 condition 260/295 (88.1%) 192/216 (88.9%) 68/79 (86.1%) 0.543 1.27 (0.62-2.59) 0.5177 ≥ 2 conditions 175/295 (59.3%) 128/216 (59.3%) 47/89 (59.5%) 1.000 1.47 (0.90-2.38) 0.1202 ≥ 3 conditions 67/295 (22.7%) 47/216 (21.8%) 20/89 (25.3%) 0.532 1.27 (0.75-2.16) 0.3732 Body composition* SM area, per 10 cm2 73.3 (62.6-84.2) 73.5 (62.5-84.8) 72.1 (64.4-80.9) 0.762 0.90 (0.73-1.09) 0.2857 SM mean HU, per 1 HU 33.8 (36.1-40.3) 33.2 (25.3-40.2) 36.0 (30.3-40.8) 0.050 1.03 (0.99-1.08) 0.1098 SAT area, per 10 cm2 114.0 (84.6-155.5) 116.6 (86.5-157.7) 109.7 (79.8-146.4) 0.195 0.96 (0.92-1.01) 0.0850 SAT mean HU, per 1 HU -92.3 (-98.3- -83.3) -92.0 (-97.9- -83.4) -93.0 (-99.2- -82.7) 0.464 0.99 (0.96-1.01) 0.2789 VAT area, per 10 cm2 100.1 (59.4-140.4) 102.6 (61.6-144.1) 86.7 (57.0-124.8) 0.076 0.98 (0.93-1.03) 0.4399 VAT mean HU, per 1 HU -87.3 (-92.4 - -79.7) -88.1 (-92.9- -79.6) -85.4 (-90.3- -80.1) 0.163 0.96 (0.93-1.00) 0.0421 IMAT area, per 5 cm2 11.3 (8.6-15.1) 11.5 (8.8-15.8) 10.9 (7.9-14.2) 0.104 0.98 (0.78-1.22) 0.8336 IMAT mean HU, per 1 HU -59.0 (-61.2- -56.8) -58.9 (-61.3- -56.7) -59.1 (-61.1- -57.1) 0.582 0.93 (0.89-0.98) 0.0074 Hospitalization Mild, severity (reference) 251 (72.3%) 194 (75.2%) 57 (64.0%) 0.075 1.00 NA Moderate, severity 66 (19.0%) 46 (17.8%) 20 (22.5% 2.01 (1.17-3.44) 0.0110 Severe, severity 30 (8.7%) 18 (7.0%) 12 (13.5%) 2.92 (1.53-5.57) 0.0011 Length of stay ≥ 7 days 56 (16.1%) 42 (16.3%) 14 (15.7%) 1.000 0.80 (0.39-1.62) 0.5321 ICU admission 45 (13.0%) 25 (9.7%) 20 (22.5%) 0.003 1.53 (0.84-2.78) 0.1617 Cardiovascular Failure 37 (10.7% 28 (10.9%) 9 (10.1%) 1.000 0.63 (0.29-1.38) 0.2453 Respiratory failure 37 (10.7%) 43 (16.7%) 17 (19.1%) 0.627 0.70 (0.30-1.61) 0.3999 Renal failure 60 (17.3%) 22 (8.5%) 15 (16.9%) 0.044 1.06 (0.51-2.17) 0.8812 New diagnosis diabetes 9 (3.4%) 7 (3.5%) 2 (3.3%) 1.000 0.47 (0.11-1.99) 0.3079 HR, hazards ratio; CI, confidence interval; SIRS, systemic inflammatory response syndrome; MCV, mean corpuscular volume; BUN, blood urea nitrogen; SM, skeletal muscle; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue; HU, Hounsfield unit; IMAT, intermuscular adipose tissue; ICU, intensive care unit. Reference ranges: hematocrit, 38.3-48.6%, hemoglobin, 13.2-16.6 g/dL, MCV, 78.2-97.9 fL; platelets, 135 – 317 x 109 /L; sodium, 135-145 mmol/L; BUN, 8-24 mg/dL; total bilirubin, ≤ 1.2 mg/dL; direct bilirubin, 0.0-0.3 mg/dL; lipase, 13-60 U/L Ϯ Metabolic conditions included hypertension, diabetes, or prediabetes (treated mutually exclusive), hyperlipidemia, or fatty liver. Due to the high number of missing data for fatty liver, it was not included in the analysis.*Areas for skeletal muscle (SM), subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and intermuscular adipose tissue (IMAT) are corrected for body surface area (BSA), calculated using the Mosteller formula. Mean HU is not corrected as it is a relative quantitative measurement of radio density, which is independent of BSA.† Multivariable models were adjusted for age, male sex, first episode of acute pancreatitis, BUN, and pancreatitis severity.
MoreTranslated text
Key words
acute pancreatitis,body composition,artificial intelligence,long-term
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined