To the Editor:
Progress in the understanding of coronavirus 2019 (COVID-19) has been substantial but incomplete. Infection of the gastrointestinal system with the SARS-CoV-2 virus may result in systemic inflammation and increased bowel permeability. Serum amylase released from the pancreas into the gut may exit the bowel and enter the systemic circulation in critical illness.1 Increases in serum amylase may thus serve as a surrogate marker for impaired bowel integrity and severity of illness in COVID-19.2,3 We sought to test the hypothesis that pancreatic inflammation as assessed by serum amylase is associated with poor outcomes in critically ill patients with COVID-19.
MATERIAL AND METHODS
The University of California, San Diego institutional review board approved waiver of consent for this prospective observational cohort study. All adult patients with COVID-19 admitted to the intensive care unit (ICU) for more than 1 day between May 1, 2020, and October 31, 2020, were enrolled. A random sample of COVID-19 “rule-out” patients admitted to the ICU and subsequently found to be COVID-19 negative were enrolled as comparators.
Serum amylase levels were recorded on all patients as part of usual care on ICU days 1, 4, 7, 10, and 20 as available. Maximum amylase levels in survivor and nonsurvivor cohorts during ICU stay were recorded as the primary outcome, with amylase 150 U/L or greater defined as “elevated.” Demographics, laboratory values, and outcomes were gathered as part of routine (Table 1).
TABLE 1.
Characteristics of the Amylase Cohorts
| Covariate | Maximum Amylase ≤150 U/L, n (%) | Maximum Amylase >150 U/L, n (%) | P |
|---|---|---|---|
| Total, n | 108 | 29 | |
| Inpatient mortality | 25 (23.1) | 13 (44.8) | 0.037 |
| Age, mean (SD), y | 58.9 (15.05) | 57.0 (14.3) | 0.56 |
| Sex, male, n | 70 | 12 | 0.04 |
| Ethnicity | 0.99 | ||
| Non-Hispanic | 28 (25.9) | 5 (17.2) | |
| Hispanic | 79 (73.1) | 24 (82.8) | |
| Unknown | 1 (0.9) | 0 | |
| Required endotracheal intubation | 74 (68.5) | 25 (86.2) | 0.098 |
| ECMO | 16 (14.8) | 4 (13.8) | 0.99 |
| CRRT | 15 (13.9) | 10 (34.5) | 0.02 |
| Required pressors | 54 (50.0) | 19 (65.5) | 0.2 |
| Had another infection | 47 (43.5) | 20 (69.0) | 0.03 |
| Received tracheostomy | 20 (18.5) | 7 (24.1) | 0.68 |
| Comorbidities | |||
| COPD | 6 (5.6) | 0 | 0.43 |
| Chronic kidney disease | 13 (12.0) | 2 (6.9) | 0.65 |
| End-stage renal disease | 4 (3.7) | 0 | 0.67 |
| Coronary artery disease | 10 (9.3) | 2 (6.9) | 0.98 |
| Hypertension | 58 (53.7) | 11 (37.9) | 0.19 |
| Congestive heart failure | 11 (10.2) | 2 (6.9) | 0.86 |
| Cardiac arrhythmia | 7 (6.5) | 2 (6.9) | 0.99 |
| Diabetes mellitus | 44 (40.7) | 10 (34.5) | 0.69 |
| Obesity | 21 (19.4) | 7 (24.1) | 0.77 |
| Liver cirrhosis | 3 (2.8) | 0 | 0.85 |
| Maximum laboratory values, mean (SD) | |||
| White blood count, 109/L | 16.4 (8.7) | 19.7 (10.8) | 0.14 |
| International normalized ratio | 1.5 (0.58) | 1.3 (0.19) | 0.03 |
| D-dimer, ng/mL | 6812 (12,370) | 4407 (8766) | 0.35 |
| Fibrinogen, mg/dL | 656.3 (241.3) | 645.5 (237.4) | 0.88 |
| Ferritin, mcg/L | 3052.1 (6702.8) | 4784 (14,784.6) | 0.66 |
| C-reactive protein, mg/L | 20.4 (13.4) | 17.3 (12.6) | 0.37 |
| Lactate dehydrogenase, U/L | 584.5 (353.1) | 599.9 (515.8) | 0.91 |
| Procalcitonin, ng/mL | 6.5 (33.3) | 4.0 (11.0) | 0.56 |
| Lactate, mmol/L | 3.7 (5.0) | 4.4 (4.6) | 0.49 |
P was calculated by Welch 2-sample t test and Pearson χ2 test for continuous and categorical variables, respectively.
COPD indicates chronic obstructive pulmonary disease; ECMO, extracorporeal membrane oxygenation; CRRT, continuous renal replacement therapy; SD, standard deviation.
Amylase values, demographics, interventions, laboratory values, and comorbidities between survivor and nonsurvivor cohorts were compared using Welch 2-sample t test and Pearson χ2 test for continuous and categorical variables, respectively. We performed univariate logistic regression for each covariate to determine its association with mortality. Covariates with P < 0.2 on univariate logistic regression were included in the initial model for the multivariable logistic regression. The final model was built via a combination of forward selection and backward elimination based on the Aikake information criterion. From the final model, the association of elevated amylase to mortality was reported. Odds ratios (OR) and their 95% confidence intervals (CI) were reported for covariates. P < 0.05 was considered to be significant.
We assessed discrimination of a predictive model using variables available upon ICU admission and reported the area under the receiver operating characteristics curve (AUC). The model was tested via 10-fold cross validation, and the mean AUC with 95% CI was reported.
RESULTS
Of the 137 patients with COVID-19 enrolled, there were 29 (21.2%) patients with a maximum amylase level during their ICU stay of 150 U/L or greater. Overall, 38 of the 137 patients with COVID-19 (27.8%) died. There was a statistically significantly greater proportion of patients who died in the elevated amylase cohort (44.8%) compared with the nonelevated amylase cohort (23.1%, P = 0.037) (Table 1). In contrast, mortality in COVID-19 rule-out patients (n = 23) was significantly less (17% vs 27.8%, P = 0.015) than for COVID-19–positive patients.
On univariate logistic regression analysis modeling covariates to mortality, maximum amylase 150 U/L or greater (OR, 2.70; 95% CI, 1.14–6.36; P = 0.02), endotracheal intubation (OR, 6.38; 95% CI, 1.83–22.24; P = 0.004), use of extracorporeal membrane oxygenation (OR, 3.18; 95% CI, 1.20–8.42; P = 0.02), use of continuous renal replacement therapy (OR, 15.5; 95% CI, 5.47–43.94; P < 0.0001), pressor requirement (OR, 9.74; 95% CI, 3.50–27.1; P < 0.0001), and presence of a concomitant infection (OR, 2.61; 95% CI, 1.20–5.69; P = 0.02) were associated with increased mortality.
We performed multivariable logistic regression to determine the association of maximum amylase values during ICU stay with mortality. The final model included maximum amylase, age, extracorporeal membrane oxygenation, continuous renal replacement therapy, need for pressors, tracheostomy, elevated white blood cell count, elevated D-dimer, and elevated C-reactive protein. When controlling for confounders, patients with COVID-19 with elevated maximum amylase levels had a significantly increased odds of mortality (OR, 4.64; 95% CI, 1.07–20.23; P = 0.04) (Fig. 1).
FIGURE 1.
Forest plot of association of variables with mortality while controlling for various confounders. Results of the multivariable logistic regression modeling the association of maximum amylase during intensive care unit stay with mortality in patients with COVID-19. The final logistic regression model was built via backward elimination and forward selection. All covariates listed in the figure were included in the final multivariable model. CRP, C-reactive protein; CRRT, continuous renal replacement therapy; ECMO, extracorporeal membrane oxygenation; OR, adjusted odds ratio; WBC, white blood cell count.
From our predictive model for mortality using only covariates known at time of ICU admission, an initial amylase of 150 U/L or greater was predictive of mortality (OR, 6.17; 95% CI, 1.61–23.63; P = 0.008). The mean AUC was 0.769 (95% CI, 0.676–0.854) calculated on 10-fold cross-validation.
DISCUSSION
Results from this study demonstrate a strong association between elevated serum amylase and mortality in ICU patients with COVID-19. This finding is consistent with the notion that bowel permeability may be causally important in the pathophysiology of COVID-19. In theory, amylase elevations in ICU patients with COVID-19 reflect pancreatic inflammation (ie, amylase production) rather than changes in small bowel permeability, and elevated amylase has not consistently been associated with worsened outcomes in retrospective analyses of general (non–critically ill) hospitalized populations.4–6 Nonetheless, results from this study implicate bowel and perhaps pancreatic dysfunction in the propagation of severe disease in patients with COVID-19. Our design does not allow definitive mechanistic conclusions, and we welcome further data to corroborate or refute our findings. We believe our findings may be clinically important and hope that they stimulate further research.
Mark Kowalczyk, MD
Department of Medicine
University of California San Diego
La Jolla, CA
m1kowalczyk@health.ucsd.edu
Rodney A. Gabriel, MD
Department of Anesthesiology
University of California San Diego
La Jolla, CA
Division of Biomedical Informatics
University of California San Diego
La Jolla, CA
Atul Malhotra, MD
Department of Medicine
Division of Pulmonary, Critical Care
and Sleep Medicine
University of California San Diego
La Jolla, CA
Erik B. Kistler, MD, PhD
Division of Critical Care
Department of Anesthesiology
University of California San Diego
La Jolla, CA
VA San Diego Healthcare System
San Diego, CA
Footnotes
This study was supported by the California Breast Cancer Research Program of the University of California, RGPO Grant R01RG3766 (R00RG2541) (to E.B.K.), and USARMY MEDICAL RESEARCH ACQUISITION ACTIVITY (USAMRAA) Award #W81XWH-17-2-0047 (to E.B.K.).
The authors declare no conflict of interest.
The views and results presented here are entirely those of the authors and do not necessarily represent those of the Department of Defense or its components.
Ethics approval and consent to participate—included in text. The University of California, San Diego Institutional Review Board (IRB) provided approval of the study conduct (IRB approval #200722X); subject informed consent was waived by the IRB because of the observational nature of the study and deidentified use of the data.
The datasets used and/or analyzed during this study are available from the corresponding author on reasonable request.
M.K. helped with data collection and analysis, manuscript review, and approval; R.A.G. helped with statistical analysis and interpretation, manuscript preparation, review, and approval; A.M. helped to prepare, review, and approve the manuscript; and E.B.K. helped to design, prepare, review, and approve the manuscript.
Contributor Information
Rodney A. Gabriel, Email: ragabriel@health.ucsd.edu.
Atul Malhotra, Email: amalhotra@health.ucsd.edu.
Erik B. Kistler, Email: ekistler@health.ucsd.edu.
REFERENCES
- 1.Mayer AD Airey M Hodgson J, et al. Enzyme transfer from pancreas to plasma during acute pancreatitis. The contribution of ascitic fluid and lymphatic drainage of the pancreas. Gut. 1985;26:876–881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Zhang J Liu P Wang M, et al. The clinical data from 19 critically ill patients with coronavirus disease 2019: a single-centered, retrospective, observational study. Z Gesundh Wiss. 2022;30:361–364. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Alsaigh T Chang M Richter M, et al. In vivo analysis of intestinal permeability following hemorrhagic shock. World J Crit Care Med. 2015;4:287–295. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Pribadi RR, Simadibrata M. Increased serum amylase and/or lipase in coronavirus disease 2019 (COVID-19) patients: is it really pancreatic injury? JGH Open. 2020;5:190–192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Akarsu C Karabulut M Aydin H, et al. Association between acute pancreatitis and COVID-19: could pancreatitis be the missing piece of the puzzle about increased mortality rates? J Invest Surg. 2022;35:119–125. [DOI] [PubMed] [Google Scholar]
- 6.Stephens JR Wong JLC Broomhead R, et al. Raised serum amylase in patients with COVID-19 may not be associated with pancreatitis. Br J Surg. 2021;108:e152–e153. [DOI] [PMC free article] [PubMed] [Google Scholar]