Circulating Mitochondrial DNA as a Diagnostic Biomarker for Predicting Disease Severity in Patients With Acute Pancreatitis

Acute pancreatitis (AP) is a leading cause of gastrointestinal hospital admissions with up to 40% mortality in moderate to severe AP.1Boxhoorn L. et al.Lancet. 2020; 396: 726-734Abstract Full Text Full Text PDF PubMed Scopus (405) Google Scholar Cell death and inflammation are key processes in the pathophysiology of AP, and mitochondria-mediated cell death plays a central role in the initiation of inflammatory pathways of AP.2Krysko D.V. et al.Trends Immunol. 2011; 32: 157-164Abstract Full Text Full Text PDF PubMed Scopus (524) Google Scholar Evidence suggests that several damage-associated molecular patterns (DAMPs) are released from dead pancreatic acinar mitochondria into the extracellular circulation. Given the need for early diagnosis of AP severity for proper treatment, we investigated whether circulating levels of mitochondrial DNA (mtDNA) fragments are a potential biomarker of early distinction for AP severity. We prospectively enrolled 20 AP patients who presented to the emergency department and 26 healthy control subjects and collected blood samples within 24 hours of hospitalization (admission) and at discharge. Circulating plasma mtDNA fragments (D-Loop [DL], NADH ubiquinone oxidoreductase chain 1 [ND, and averaged mtDNA) were measured using our newly developed, high-sensitivity, duplex droplet digital PCR (ddPCR) assay as described in Supplementary Methods. mtDNA levels (reported as copies/μL) were compared between AP patients and control subjects on admission and discharge by AP severity. Multivariable logistic regression and predictive analyses were performed to determine the association between mtDNA levels and AP severity on admission. Demographic comparisons between groups were similar, except for higher body mass indices in AP patients compared with healthy control subjects (29.9 vs 25.0 kg/m2, P = .001). Ten patients each had mild AP and moderate to severe AP based on modified Atlanta criteria (see Supplementary Methods). The etiology for AP was as follows: alcoholic (9/20), biliary (3/20), hypertriglyceridemia (3/20), and other causes (5/20). Compared with healthy control subjects, on admission median levels of plasma mtDNA fragment ND1 (5.8 copies/μL [interquartile range {IQR}, 1.7–16.3] vs 2.2 copies/μL [IQR, 1.0–3.8], P = .03) and median averaged mtDNA fragments (5.26 copies/μL [IQR, 1.47–15.7] vs 2.26 copies/μL [IQR, 0.84–3.67], P = .039) were significantly higher in AP patients on admission but did not differ on discharge. To characterize dynamic changes in circulating mtDNA levels during disease progression, we performed a longitudinal analysis for 1 AP patient (patient 7) for whom we were able to obtain additional samples throughout hospitalization and after discharge. Representative images of ddPCR amplitude plots for healthy control subjects and for multiple time points of patient 7 are shown in Supplementary Figure 1. In patient 7, circulating mtDNA levels were elevated on admission (15 hours), peaked in the next few days during hospitalization, and returned to (upper limit of) baseline at 30 days after discharge. Patients with moderate to severe AP compared with mild AP had poorer clinical outcomes including longer length of stay (7.9 ± 6.2 vs 3.5 ± 2.4 days, P = .05) and necrotizing pancreatitis (70% vs 0%, P = .003). Comparison of mtDNA levels by severity (Table 1) showed that patients with moderate to severe AP had significantly higher median DL (12.6 copies/μL [IQR, 5.73–25.7] vs 1.34 copies/μL [IQR, 1.08–2.75], respectively; P = .004) and ND1 (15.2 copies/μL [IQR, 6.21–24.9] vs 1.69 copies/μL [IQR, 1.57–3.04], respectively; P = .003) levels compared with patients with mild AP at admission. Similarly, averaged mtDNA levels were higher in moderate to severe compared with mild AP (6.59 copies/μL [IQR, 5.26–15.7] vs 1.46 copies/μL [IQR, 1.41–2.89], P = .02). However, compared with healthy control subjects, all 3 mtDNA levels (DL, ND1, and averaged mtDNA) did not differ in patients with mild AP but significantly increased in patients with moderate to severe AP at admission (P < .001). Additionally, there was no significant difference in all mtDNA levels between the 3 groups (healthy control subjects, patients with mild AP, and patients with moderate to severe AP) at discharge.Table 1mtDNA Levels by Severity of APHealthy control subjects (n = 21)All AP (n = 20)Mild AP (n = 10)Moderate to severe AP (n = 10)P (healthy vs mild AP)P (healthy vs moderate to severe AP)P (mild vs moderate to severe AP)DL fragment, copies/μL2.34 (1.08–3.88) AP (admission)—4.76 (1.36–15.0)a1.34 (1.08–2.75)12.6 (5.73–25.7).46<.001.004 AP (discharge)—3.48 (1.93–9.70)3.08 (1.56–6.42)3.65 (2.73–11.6).33.22.48ND1 fragment, copies/μL2.23 (1.04–3.82) AP (admission)—5.77 (1.74–16.3)b1.69 (1.57–3.04)15.2 (6.21–24.9).82<.001.003 AP (discharge)—3.83 (1.12–10.2)3.06 (1.09–5.38)4.04 (2.82–11.0).58.13.43Averaged mtDNA fragments, copies/μL2.26 (0.84–3.67) AP (admission)—5.26 (1.47–15.7)b1.46 (1.41–2.89)13.9 (5.97–25.6).67<.001.004 AP (discharge)—3.38 (1.07–8.18)2.87 (1.07–5.41)3.61 (2.31–9.64).51.22.73Values are median (interquartile range). Significance between AP and healthy control subjects is denoted by aP < .1 and bP < .05. Open table in a new tab Values are median (interquartile range). Significance between AP and healthy control subjects is denoted by aP < .1 and bP < .05. Multivariable logistic regression, adjusting for body mass index, showed that compared with mild AP on admission, moderate to severe AP was associated with increasing mtDNA DL (odds ratio, 1.94; 95% confidence interval, 1.25–4.41; P = .02) and ND1 (odds ratio, 1.99; 95% confidence interval, 1.25–4.63; P = .02). Full estimates are available in Supplementary Table 1. Additionally, receiver-operating characteristic curve analysis to distinguish AP severity yielded an area under the curve for mtDNA fragments of 0.91 (95% confidence interval, 0.75–1.0) at a sensitivity of 0.90 and specificity of 0.90. Youden's J statistic, which combines sensitivity and specificity to determine an optimal cutoff value, reported levels of 3.90, 5.62, and 4.76 copies/μL for DL, ND1, and averaged mtDNA fragments, respectively, at a sensitivity of 0.90 and specificity of 0.90. Further, using the cutoff value for averaged mtDNA fragments, the positive predictive value and negative predictive value for differentiating AP vs control on admission was 73.3% and 66.7%, respectively, whereas the positive predictive value and negative predictive value for moderate to severe AP vs mild AP on admission was 81.8% and 88.9%, respectively. Our study shows that circulating mtDNA in the first 24 hours of presentation could potentially be a unique marker of early prediction of AP severity on admission. Circulating mtDNA levels are elevated in several inflammatory states such as coronavirus disease 2019,3Scozzi D. et al.JCI Insight. 2021; 6: 4Google Scholar and our study similarly shows high circulating mtDNA early in the course of AP and corresponds with the clinical presentation and severity. The pathogenesis of AP is complex; it starts with an initial injury to the pancreatic acinar cells and leads to release of pancreatic enzymes and eventual local cell death, resulting in systemic inflammation.4Sah R.P. et al.Curr Opin Gastroenterol. 2012; 28: 507-515Crossref PubMed Scopus (122) Google Scholar,5Kang R. et al.Mol Med. 2014; 20: 466-477Crossref PubMed Scopus (110) Google Scholar At the cellular level, the mitochondria play an important role in regulating pancreatic cell death through adenosine triphosphate and reactive oxygen species production. Dysfunctional mitochondria lead to oxidative injury and damage of other organelles, which ultimately results in pancreatic acinar cell death.6Kubisch C.H. et al.Am J Physiol Gastrointest Liver Physiol. 2006; 291: G238-G245Crossref PubMed Scopus (100) Google Scholar,7Pandol S.J. et al.Gastroenterology. 2007; 132: 1127-1151Abstract Full Text Full Text PDF PubMed Scopus (495) Google Scholar Dead pancreatic acini release intracellular DAMPs including mtDNA, which ultimately leads to activation of various inflammatory signal pathways and subsequent systemic inflammatory response syndrome and eventual end-organ damage characteristic of moderate to severe AP.8Shimada K. et al.Immunity. 2012; 36: 401-414Abstract Full Text Full Text PDF PubMed Scopus (1515) Google Scholar,9Huang H. et al.Gastroenterology. 2013; 144: 202-210Abstract Full Text Full Text PDF PubMed Scopus (178) Google Scholar Because DAMPs may be the connecting link between local cell death and systemic inflammatory response syndrome, measurement of their circulating levels may be a unique tool for accurate prediction of AP severity in an effort to improve outcomes in patients with moderate to severe AP. Moreover our study shows that in absence of cell death or severe acinar injury the mtDNA levels do not rise, hence the nonsignificant difference between mtDNA levels in mild AP and healthy controls. One possible limitation of the study may be the disproportionately higher number of moderate/severe AP patients (50% vs typically 25%). We do not think this represents a selection bias as severity of AP was not known on admission. Enrollment was also random and dependent on patient consent and obtaining blood samples within 24 hours, and availability of resources to enroll the patient. Wu et al10Wu L. et al.Dig Dis Sci. 2018; 63: 2975-2982Crossref PubMed Scopus (15) Google Scholar reported that circulating mtDNA, measured in plasma with conventional quantitative PCR (using SYBR green–based detection of a single ND1 fragment), appeared to detect pancreatic necrosis. In their study, most patients had blood levels drawn 2.4 days after presentation, by which time it is clinically apparent if the patient has mild or moderate to severe AP. In contrast, our study used a highly sensitive and more precise ddPCR method for better quantification of mtDNA copy number and duplex detection of different mtDNA fragments, and patients were enrolled within 24 hours of presentation. Additionally, collection of serial and/or discharge samples and presence of a well-defined healthy control group are added strengths of our study. Finally, the faster turnaround time of the ddPCR assay (ie, approximately 8 hours from blood draw to results becoming available and potentially as fast as 2 hours with assay modifications for commercial use) will allow clinicians to predict AP severity. In conclusion, this pilot study highlights the potential of circulating mtDNA as a novel biomarker for severity of AP. Further multicenter studies of AP patients with variable etiology and severity are warranted to rigorously validate our results. Members of the BIDMC Acute Pancreatitis Working Group are as follows: Awais Ahmed (Division of Gastroenterology, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts), Pinzhu Huang (Division of Gastroenterology, Hepatology and Nutrition, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts), Ankit Chhoda (Division of Gastroenterology, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts), and Konstantin Khrapko (Laboratory of Ageing and Infertility Research, Department of Biology, Northeastern University, Boston, Massachusetts). William Yakah, BS (Data curation: Equal; Formal analysis: Lead; Writing – original draft: Equal; Writing – review & editing: Equal). Ishani Shah, MD (Data curation: Equal; Formal analysis: Supporting; Writing – original draft: Equal; Writing – review & editing: Equal). Disha Skelton-Badlani, BS (Data curation: Equal; Formal analysis: Supporting; Methodology: Equal; Writing – original draft: Supporting). Steven D. Freedman, MD, PhD (Funding acquisition: Equal; Resources: Equal; Writing – review & editing: Equal). Yury V. Popov, MD, PhD (Conceptualization: Equal; Data curation: Equal; Formal analysis: Supporting; Funding acquisition: Equal; Methodology: Lead; Resources: Equal; Writing – original draft: Equal; Writing – review & editing: Equal). Sunil G. Sheth, MD (Conceptualization: Equal; Data curation: Supporting; Formal analysis: Supporting; Funding acquisition: Equal; Investigation: Equal; Project administration: Lead; Supervision: Lead; Writing – original draft: Equal; Writing – review & editing: Equal). This single-center study was approved by the Institutional Review Board. We prospectively enrolled patients with AP who presented to the emergency department at our tertiary care medical center from July 2020 to June 2021. The diagnosis of AP was made based on meeting 2 of 3 revised Atlanta criteria: abdominal pain consistent with AP, serum lipase activity at least 3 times greater than the upper limit of normal, and characteristic findings of AP on cross-sectional imaging such as computed tomography or magnetic resonance imaging. Patients were excluded from enrollment if they were younger than 18 years or older than 85 years of age or had a prior history of undergoing pancreatic surgery or pancreatic cancer. Eligible AP patients were approached by our group at their presentation to the hospital. Written informed consent was obtained for participation in the study. We then collected blood samples from each patient at different time points of their hospitalization: The first blood sample was collected at the time of enrollment (first admission day) and the last blood sample on the day of discharge from the hospital. Plasma mtDNA levels were measured in AP patients at admission, during treatment, at discharge, and after discharge in 1 patient using an in-house duplex ddPCR mtDNA assay. Briefly, total cell-free DNA was isolated from 53 μL of plasma after additional centrifugation at 13,000g for 10 minutes at 4oC using the Gentra Puregene kit (Qiagen). The absolute concentration of the mtDNA template (copy numbers/μL of plasma) was quantified using the TaqMan principle on the QX200 ddPCR System (Bio-Rad) via simultaneous amplification of 2 different, highly conserved sequences within DL and ND1 regions of the mtDNA genome. The number of copies of the mtDNA template for DL and ND1 fragments and the mathematical average of the 2 fragments (termed averaged mtDNA) were then analyzed in relation to clinical parameters. Corresponding primers and hexachlorofluorescein- or 6-carboxy-fluorescein–labeled TaqMan probes sequences are available on request. Quantitative TaqMan PCR amplification of DL and ND1 regions using the same primer/probe sequences was performed according to a previously established method in parallel reactions on Light Cycler 1.5 instrument (Roche) for validation and direct comparison with ddPCR assay. Relative mtDNA abundance was calculated based on corresponding cycle threshold values using a second derivative maximum option and internal calibration curve. We collected the following information for all enrolled AP patients: demographic and clinical characteristics, etiology and severity of AP, laboratory and radiologic findings, and data on complications and procedures performed during the hospital course. Individual variables included age, sex, race, body mass index, alcohol use, smoking status, diabetes status, presence of prior AP, etiology of AP, severity of AP based on revised Atlanta Classification, presence of pancreatic necrosis, and other local pancreatic complications. Severity of AP was graded as mild, no local pancreatic complication or organ dysfunction; moderate, local complication (pancreatic fluid collection or necrosis) and/or transient (<48 hours) end-organ damage; or severe, local complication and/or persistent (>48 hours) end-organ damage. Other variables included splanchnic vein thrombosis, systemic complications (such as acute respiratory failure requiring mechanical ventilation, ascites, bacteremia, Clostridoides difficile infection, gastrointestinal bleeding, acute liver injury, acute encephalopathy, acute kidney injury), imaging studies (computed tomography and magnetic resonance imaging), and invasive procedures (endoscopic ultrasound, endoscopic retrograde cholangiopancreatography, and cholecystectomy). We also collected data regarding inpatient mortality, time taken to initiate refeeding, admission to the intensive care unit, and length of stay. We also enrolled volunteers as healthy control subjects in our study. Eligibility criteria for healthy control subjects were age > 18 years; absence of known pancreatic disease; and absence of other chronic comorbidities such as diabetes, hypertension, heart disease, chronic liver disease, chronic kidney disease, chronic lung disease, or malignancy. After obtaining informed consent, we collected 1 fasting blood sample from the control subjects. The same protocols were followed for both sample collection and processing in healthy control subjects as was for the enrolled AP patients. Data collection included demographic data (age, sex, race, body mass index) and mtDNA levels. We divided our study population into 2 groups: AP patients and healthy control subjects. Our primary outcome of interest was comparison of mtDNA levels between the 2 groups. We compared the admission and discharge mtDNA levels of the AP group with mtDNA levels of the control group. In a subgroup analysis we compared mtDNA levels between healthy control subjects and all AP patients as well as within AP patients based on severity (ie, mild vs moderate to severe). We also compared other clinically relevant outcomes such as necrotizing pancreatitis, length of stay, and intensive care unit admission. We also analyzed the effectiveness of mtDNA levels in predicting severity of AP. Categorical variables are presented as proportions and continuous variables as mean with SD. Hypothesis testing was performed using Pearson χ2 test for categorical variables and the Student t test for continuous variables. Logistic regression following multivariable linear regression model was used to determine the association of several clinical outcomes of AP (intensive care unit stay, necrosis, severity, etc) with mtDNA levels (categorized as deciles). Unless otherwise specified, all analyses were performed using R software (version 3.6.1; R Core Team 2018a) within RStudio (version 1.1463; RStudio, Inc) using the tidyverse package.e1Wickham H. et al.Journal of Open Source Software. 2019; : 1686Crossref Google Scholar Predictive receiver-operating characteristic curve analysis (specificity, sensitivity, area under the roc curve, Youden's J) was performed using Epitools platform. Two-tailed P < .05 was considered statistically significant.Supplementary Table 1Association of mtDNA Levels (Categorized as Deciles) and Severity of Acute Pancreatitis on AdmissionaLogistic regression following generalized linear models is adjusted for body mass index. CI, confidence interval; OR, odds ratio.DL FragmentND1 FragmentAveraged mtDNA FragmentsOR (95% CI)POR (95% CI)POR (95% CI)PMild AP vs control (ref)0.86 (0.61–1.17).350.94 (0.68–1.28).700.93 (0.67–1.26).64Moderate to severe vs control (ref)2.13 (1.23–4.91).032.12 (1.25–4.87).022.11 (1.25–4.86).03Moderate to severe vs mild AP (ref)1.94 (1.25–4.41).021.99 (1.25–4.63).021.98 (1.25–4.57).02a Logistic regression following generalized linear models is adjusted for body mass index. CI, confidence interval; OR, odds ratio. Open table in a new tab