Comparing Prognostic Scores and Inflammatory Markers in Predicting the Severity and Mortality of Acute Pancreatitis

Background: Acute pancreatitis is an emergency gastrointestinal condition for which severity prediction is crucial during hospitalization. This study aimed to compare the diagnostic accuracy of inflammatory markers with gold standard scoring systems in predicting pancreatitis severity. Materials and methods: A prospective, hospital-based, cohort study was conducted, including 249 patients diagnosed with acute pancreatitis via clinical examination. Laboratory investigations and radiological investigations were conducted. The diagnostic accuracy of the inflammatory markers neutrophil/lymphocyte ratio (NLR), lymphocyte/monocyte ratio (LMR), red cell distribution width (RDW), and prognostic nutritional index (PNI) was compared with gold standard prognostic scores, namely, the Acute Physiology and Chronic Health Evaluation II (APACHE II), Simplified Acute Physiology Score II (SAPS II), Bedside Index of Severity in Acute Pancreatitis (BISAP), and Systemic Inflammatory Response Syndrome (SIRS), in predicting primary and secondary outcomes. All values were analyzed using mean and standard deviation (SD). Sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristic curve for mortality prediction were calculated for NLR, LMR, RDW, and PNI. Results: Of 249 patients with acute pancreatitis (mean age: 39-43 years), 94 were classified as mild acute, 74 as moderately severe acute, and 81 as severe acute. The most common etiology was alcohol use (40.2%), followed by gallstones (29.7%), hypertriglyceridemia (6.4%), steroid use (4%), diabetic ketoacidosis (2.8%), hypercalcemia (2.8%), and complication of endoscopic retrograde cholangiopancreatography (2%). On day 1, mean values of NLR, LMR, RDW, and PNI were 8.23±5.11, 2.63±1.76, 15.93±3.64, and 32.84±8.13, respectively. Compared to APACHE II, SAPS II, BISAP, and SIRS on day 1, day 3, day 7, and day 14, the cutoff values for NLR were 4.06, 10.75, 8.75, and 13.75, respectively. Similarly, on day 1, the cutoff value of LMR was 1.95, and on day 1 and day 3, the cutoff values of RDW were 14.75% and 15%, respectively. Conclusion: The results indicate that inflammatory biomarkers NLR, LMR, RDW, and PNI are comparable with gold standard scoring systems for predicting the severity and mortality of acute pancreatitis. NLR on day 7 was significantly associated with higher severity of illness. NLR on days 3, 7, and 14, LMR on day 1, and RDW on days 1 and 3 were significantly associated with mortality.


Introduction
Acute pancreatitis is a frequent cause of gastrointestinal hospitalizations. Incidence has increased sharply in the last several decades to 20-40 per 100,000 people annually worldwide [1,2]. The mortality rate of acute pancreatitis is around 23% in the initial three days and increases to 53% by day 7, making proper management and prevention of complications essential. Numerous scoring systems such as the Acute Physiology and Chronic Health Evaluation II (APACHE II), Simplified Acute Physiology Score II (SAPS II), Bedside Index of Severity in Acute Pancreatitis (BISAP), Systemic Inflammatory Response Syndrome (SIRS), Modified Marshall, and GLASGOW are considered gold standards for predicting severity at admission in patients with acute pancreatitis. However, in rural areas with limited resources, these scoring systems are often too costly. Various clinical biomarkers (e.g., elevated levels of hematocrit, blood urea nitrogen (BUN), procalcitonin, and C-reactive protein (CRP)) have been used in the first 48 hours of hospitalization to aid in predicting illness severity [3][4][5][6]. Other cost-effective metrics include hematological parameters such as neutrophil/lymphocyte ratio (NLR), lymphocyte/monocyte ratio (LMR), red cell distribution width (RDW), and prognostic nutritional index (PNI) [7]. We thus aimed to compare the diagnostic accuracy of NLR, LMR, RDW, and PNI with the above gold standard prognostic scores in predicting severity and outcomes among patients with acute pancreatitis. 1 2 1 3 1  After obtaining written consent, 249 patients were included. A detailed history, thorough examination, and laboratory testing were conducted, including complete blood count, serum electrolyte, serum amylase, serum lipase, serum hepatic and renal function, bilirubin, C-reactive protein, procalcitonin, and blood gas analyses. A radiological evaluation was performed using a transabdominal ultrasound and contrastenhanced CT scan of the whole abdomen on days 5-7 of admission. APACHE II, SAPS II, BISAP, and SIRS scores were recorded in the first 24 hours. NLR, LMR, RDW, and PNI levels were calculated at the time of admission and on day 3, day 7, and day 14. Assessment of diagnostic accuracy included predicting acute pancreatitis as a primary cause of mortality and as a secondary cause of morbidity (duration of ICU and hospital stay) and the need for intervention (surgical, radiological, and endocrine). These parameters then were compared with the previously mentioned gold standard severity scores in terms of prognostic accuracy.

Materials And Methods
All qualitative parameters are presented in numbers and percentages. Quantitative parameters are presented as means, standard deviations, and medians with interquartile ranges, depending on the distribution. For all four scores, predictive diagnostic parameters including sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristic curve were assessed for severity and mortality prediction. The Kruskal-Wallis test was applied for the hypothesis test of multiple variables (NLR, LMR, RDW, etc.) in severity prediction. A null hypothesis has been established in all independent variables. The Mann-Whitney U test was also applied to validate the null hypothesis. A p-value of less than 0.05 was considered significant. Data analysis was done using STATA software.

Discussion
The diagnostic accuracy of inflammatory markers NLR, LMR, RDW, and PNI for predicting severity and mortality in 249 patients with acute pancreatitis was compared to gold standard prognostic scores. Clinicodemographic features were recorded. The mean age was 41.44±14.78 years, in which mostly male (>50%) patients participated, similar to studies by Li et al. [7] and Zhou et al. [9]. As far as illness severity, 32.5% of patients had severe cases, which is consistent with the observations of Zhou et al. [9]. The most common comorbidity was diabetes mellitus (23.2%), followed by hypertension (7.2%), chronic kidney disease (2.4%), and coronary artery disease (0.8%). A significant difference was not observed among severity levels, consistent with the findings of Zhou et al. [9].
SIRS was observed in 34.9% of patients in our study, consistent with the findings of Singh et al. [10], who concluded that SIRS is more reliable than APACHE II for predicting the severity of acute pancreatitis. Compared to patients with mild acute pancreatitis in this study, patients with severe acute pancreatitis had higher incidences of SIRS; higher rates of organ failure; longer hospital stays; higher mean NLR, LMR, RDW, BISAP, SAPS II, and APACHE II values; and higher mortality.
Numerous scoring systems have been established in predicting outcome severity in acute pancreatitis. Scores such as BISAP and SIRS are less complicated than APACHE II, but multifactorial scoring methods are laborious and call for considerable measures. CT scans and levels of C-reactive protein, hematocrit, and procalcitonin have also been used to predict the severity of acute pancreatitis, with varying degrees of success. LMR and NLR, which are determined from differential white blood cell counts, can indicate the magnitude of the inflammatory process in a particular disease condition. RDW, which measures the volume of red blood cell fluctuation, is also linked to inflammatory processes and is used to assess anemia and predict in-hospital mortality in patients with sepsis [11].  [7] reported a mean NLR of 8.46 in patients with mild acute pancreatitis, which increased to 19.65 as the illness worsened to severe acute pancreatitis; LMR was 1.88 in patients with mild acute pancreatitis and decreased to 1.03 as the illness progressed to severe acute pancreatitis. Zhou et al. [9] found that the LMR value increases in severity in parallel with NLR, Ranson, sequential organ failure assessment, and BISAP values, along with significant differences between mild acute and severe acute pancreatitis.
The diagnostic accuracy of NLR was analyzed based on the receiver operating characteristic curve. The cutoff value of NLR on day 7 was 4.06 with 79.7% sensitivity and 38.3% specificity for severity prediction. Dancu et al. [12] reported a day 1 cutoff value of NLR of 9.6 with 65% sensitivity and 70% specificity for acute pancreatitis severity; after 48 hours, it decreased to 6.15 with 100% sensitivity and 63% specificity. Zhou et al. [9] found a 10.31 cutoff value of NLR with 64.3% sensitivity and 77.1% specificity for 28-day mortality prediction. In our study, acute pancreatitis mortality on day 3 was associated with an NLR cutoff of 10.75 with 70% sensitivity and 64.1% specificity. Similarly, NLR on day 7 and day 14 had cutoff values of 8.75 and 13.75, sensitivities of 80% and 100%, and specificities of 68.8% and 90.6%, respectively, for mortality prediction.
Regarding predictions of mortality in our study, LMR on day 1 had a cutoff value of 1.95 with 100% sensitivity and 62.5% specificity; RDW on days 1 and 3 had cutoff values of 14.75% and 15% with sensitivities of 90% and 90% and specificities of 31.3% and 32.3%, respectively. Li et al. [7] found that at a 16.64 cutoff value, NLR was 82.4% sensitive and 75.6% specific in predicting mortality of acute pancreatitis, whereas LMR at a 1.4 cutoff value was 82.4% sensitive and 57.3% specific and RDW at a cutoff value of 13% was 94.1% sensitive and 54.3% specific. Zhou et al. [9] found that at a 12.20 cutoff value, NLR was 85.7% sensitive and 84.2% specific in predicting 28-day mortality in patients with acute pancreatitis and that RDW at a 13.55% cutoff value was 100% sensitive and 74.7% specific in predicting the severity of acute pancreatitis [13][14][15].
Comparing patients with mild acute pancreatitis to patients with severe acute pancreatitis in our study, significant differences (p<0.05) were observed in the incidence of SIRS ( [9] similarly observed that compared to patients who died during their study period, those who survived had significantly lower NLR levels, RDW levels, BISAP scores, sequential organ failure assessment scores, APACHE II scores, and Ranson scores (p<0.05). Li et al. [7] also observed that compared to non-survivors, patients who survived had significantly lower NLR levels but higher LMR and PNI levels (p<0.05) ( Table 3). The limitation of the study was the low sample size during the study period.

Conclusions
Early aggressive management and resuscitation are vital in treating acute pancreatitis, so severity prediction during hospitalization is important. Numerous scoring methods have been used for severity prediction, but they are impractical in many settings due to cost and resource limitations. Inflammatory indicators such as NLR, LMR, and RDW are easy and cost-effective methods and are comparable to multiparameter scoring systems for assessing severity and mortality risk prediction in patients with acute pancreatitis. NLR on days 1 and 7, and LMR and RDW on day 1 are comparable with APACHE II and SAP II for severity prediction. For mortality prediction, NLR on days 3 and 7, LMR on day 1, and RDW on days 1 and 3 are comparable with APACHE II, BISAP, and SAP II in acute pancreatitis.