This is a single-center retrospective obeservational study. The study protocol was approved by the Ethics Committee of Ren Ji Hospital (Project no. KY2021-048) in accordance with the Declaration of Helsinki. As a retrospective data analysis, patients’ written consent was waived by the same committee .

All participants were recruited from Ren Ji Hospital from 1 to 2020 to 1 Dec 2021. The inclusion criteria were as following: (1) aged between 18 and 80 years; (2) who received PN for one week or longer (3) whose serum level of P-AMY and LP were repeatedly measured. The exclusion criteria were those: (1) aged less than 18 years or more than 90 years; (2) with missing data; (3) with estimated glomerular filtration rate (eGFR) < 30 ml/min/1.73m2, because urinary excretion of lipase decreased in patients with chronic kidney diseases [17]; (4) with serum level of alkaline phosphatase and gamma-glutamyl transferase ≥ 1.5 ULN (upper limit of normal value) and serum level of direct bilirubin ≥ 2 ULN based on the diagnostic criteria of cholestasis [7, 8]; (5) with history of confirmed liver, gall bladder and pancreas injuries (e.g., cancer, acute or chronic inflammation, trauma, abused drugs, autoimmune disease), rheumatic disease, organ transplant and pregnancy; (6) admitted to critical care units.

The intial inclusion resulted in 477 patients receiving PN, which was trimmed to 321 after sequential inclusion. After excluding those who lost follow-up, the final number of participants were 190. (Fig. 1). There were no significant differences among the participants in gender ratio, disease history and major clinical parameters in and out of the study. (Supplementary Table 1. The comparison of baseline characteristics between participants in and out of the study).

The process of sample recruitment. Abbreviation: eGFR, estimated glomerular filtration rate; P-AMY, pancreatic amylase; LP, lipase;

A professional nutrition support team reviewed the patients and prescribed PN regimens. The indication for PN was based on the insufficient oral intake or tube feeding that cannot meet 60% of the daily requirement [12]. These patients were reviewed daily by the same nutrition support team. Other sources of energy supply, such as enteral nutrition and oral intake, were also recorded and calculated. For each patient, the duration of PN, enteral nutrition and oral intake were accounted separately from the initial to the last day of PN treatment.

Biochemical parameters were monitored and recorded at baseline and were measured repeatedly for seven consecutive days after initiation of PN. Blood samples were drawn after at least an 8 h fast. Baseline parameters were analyzed within 24 h before PN (day 0). Subsequent analyses were performed at a 7-day interval until PN was stopped (e.g., day 7, day 14). Serum levels of P-AMY and LP were measured by the enzyme colorimetry analysis (Roche 701 Bioanalyzer, Roche, UK). Serum levels of alkaline phosphatase, gamma glutamyl-transferase, direct bilirubin, total bilirubin, bile acids, alanine transferase and aspartate transferase were measured by enzyme-linked immunosorbent assay (Roche 701 Bioanalyzer, Roche, UK). Fasting blood glucose, albumin, pre-albumin, and creatinine were also measured by enzyme-linked immunosorbent assay (Roche 701 Bioanalyzer, Roche, UK). The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration 2-level race Eq. 1 [8]. White blood cells were measured using an automated hematology analyzer (DxH 690T, Beckman Coulter, USA). The concentration of C-reaction protein was measured by the immunoturbidimetric method (CardioPhase hsCRP kit, Siemens Healthcare Diagnostics Products GmbH, Germany). All the measurements above were completed in the Clinical Laboratory of the Ren Ji Hospital.

Age, sex, diagnosis, history of disease and surgery, body weight and height were recorded from the medical record. Gastrointestinal dysfunction is defined as the presence of diarrhea, vomiting, abdominal distension, or intestinal obstruction [18]. The IBW (ideal body weight) was calculated as [current body height (cm) − 80) × 0.7] in men and [current body height (cm) − 70) × 0.6] in women [19].

Pancreatic injury were confirmed by serum of P-AMY (≥53 U/L) or LP (≥63 U/L) based on the normal range recommended by the Clinical Laboratory of Ren Ji Hospital. Serum levels of P-AMY and LP were repeated measured every 7 day [14, 15].

All statistical analyses were performed using IBM SPSS Statistics 23 software (IBM, Armonk, New York, United States). For the hypothesis testing, a two-sided test with a 5% significant level was used. Data with normal distribution were presented as mean±standard and the differences between the groups were tested by non-paired student tests. Non-normally distributed data were presented in median and quartile ranges, and the differences between groups were tested using the Chi-square method.

We performed univariate and multivariable binary logistic regression modeling to evaluate the association between baseline clinical features and pancreatic injury. Potential parameters were selected by Univariable regression (p < 0.2) and refered to relatively clinical trials [20,21,22]. Nive parameters (sex, age, the proportion of daily energy supplement by PN, surgery, diabetes mellitus, cancer, infection, gastrointestinal dysfuction, fistula) entered multivariate binary logistic regression model. Risk factors for PN-related liver dysfunction were evaluated by Spearman correlation and linear regression analysis. Two-step linear regression modeling was performed to evaluate the association between the serum level of liver function and pancreatic enzymes. Univariable regression was first carried out to identify eligible factors based on p value (p < 0.2). Using the regression results, four parameters (alanine transferase, aspartate transferase, total bile acids, gamma glutamyl-transferas) were included in the multivariate regression model.