- Email: [email protected]
Neutrophil-Lymphocyte Ratio as a Predictor of Adverse outcomes of Acute Pancreatitis
Original Paper Received: December 29, 2010 Accepted after revision: August 4, 2011 Published online: September 28, 2011
Pancreatology 2011;11:445–452 DOI: 10.1159/000331494
Neutrophil-Lymphocyte Ratio as a Predictor of Adverse Outcomes of Acute Pancreatitis Basem Azab a Neil Jaglall a Jean Paul Atallah a Ari Lamet a Venkat Raja-Surya a Bachir Farah a Martin Lesser c Warren D. Widmann b, d
Departments of a Internal Medicine and b Surgery, Staten Island University Hospital, Staten Island, N.Y., c Feinstein Institute for Medical Research and Hofstra School of Medicine, Manhasset, N.Y., and d State University of New York Downstate Medical Center, Brooklyn, N.Y., USA
Key Words Neutrophil-lymphocyte ratio Acute pancreatitis Severity Scoring risk system
Abstract Background: Most acute pancreatitis risk scoring systems use total white blood cell counts (WBC) as one of the risk factors. The value of the neutrophil-lymphocyte ratio (NLR) to predict the severity of acute pancreatitis has not been previously evaluated. Methods: This observational study included 283 patients admitted to a tertiary center between 2004 and 2007. The patients were arranged into tertiles according to NLR and WBC values. The primary outcomes were intensive care unit (ICU) admission and length of stay (LOS) in the hospital. Results: According to NLR tertiles, patients in the 3rd tertile (NLR 67.6) had significantly more ICU admissions (17 vs. 2.2%, p ! 0.0001) and longer average LOS (6.2 vs. 4.2 days, p ! 0.002) compared with those in the 1st tertile (NLR !3.6). According to WBC tertiles, patients in the 3rd tertile had more ICU admissions (12.6 vs. 6.2%, p = 0.12) and a longer average LOS (5.8 vs. 4.4 days, p = 0.059) compared to patients in the 1st WBC tertile, but this did not reach statistical significance. In the multivariate model including NLR, WBC and other predictors, only
© 2011 S. Karger AG, Basel and IAP 1424–3903/11/0114–0445$38.00/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com
Accessible online at: www.karger.com/pan
NLR tertiles (p ! 0.0262) and modified early warning scores (p ! 0.0025) were significant predictors of ICU admission. Likewise, in the multivariate model of LOS, only NLR and glucose level were significant predictors of longer LOS (p ! 0.0161 and p ! 0.0053, respectively). Conclusion: NLR is superior to total WBC in predicting adverse outcomes of acute pancreatitis. According to our data, we suggest using the NLR cutoff value of 14.7 as a simple indicator of severity in patients presenting with acute pancreatitis. Copyright © 2011 S. Karger AG, Basel and IAP
Background
The incidence of acute pancreatitis in industrialized countries varies from 4.9 to 35 per 100,000 persons [1, 2]. In the United States, acute pancreatitis accounts for 327,474 hospitalizations and 2,834 deaths annually [3, 4]. The clinical signs of acute pancreatitis may be nonspecific and they exhibit a low sensitivity (!40%) for the prediction of an adverse outcome [5]. Also, there is a significant chance of underdiagnosis, as demonstrated by the finding of previously undiagnosed acute pancreatitis as a cause of death in some autopsy series [6]. Because of the
Basem Azab, MD Department of Internal Medicine Staten Island University Hospital, 475 Seaview Avenue Staten Island, NY 10305 (USA) Tel. +1 347 607 8727, E-Mail basemnady2000 @ yahoo.com
339 patients were assessed for eligibility 56 patients were excluded Age >80, n = 27 Hemoproliferative disease, n = 4 Steroid or chemotherapy use, n = 6 Negative enzymes and/or CT findings, n = 5 Unavailable records, n = 14 283 patients included in study
1st NLR tertile: 93 patients
2nd NLR tertile: 96 patients
3rd NLR tertile: 94 patients
Fig. 1. Schematic illustration of study.
potential deceptive presentation of acute pancreatitis, multiple severity scoring systems have been developed to guide clinicians in triaging patients who are likely to require vigorous treatment in an intensive care unit (ICU). Ideally, these scoring systems should use predictive tools that are rapid, reproducible, inexpensive, minimally invasive and highly accurate [7]. Among the available scoring systems currently used, Balthazar [8, 9] and the early warning score (EWS) [10] use CT and physical exam findings, respectively. The other scoring systems (Ranson criteria, APACHE score, Glasgow and Imrie scores) [11–15] use combined clinical and laboratory variables. However, there is no universally adopted scoring system in clinical practice, due to their low sensitivity and the complexity of the needed calculations [5, 16]. Furthermore, some (i.e. Ranson, Glasgow and Imrie) need nonroutine blood work and have not been validated beyond 48 h. Although the Balthazar score is superior to other scoring systems in the prediction of the extent of pancreatic necrosis, it was found inferior to the Ranson and APACHE systems in the prediction of organ failure and other adverse outcomes during hospitalization [9, 17]. Garcea et al. [18] found that the APACHE II score and EWS were superior to other scoring systems (i.e. modified organ dysfunction scores, Imrie scores, Balthazar and Ranson criteria) in predicting severity and survival. In a recent systemic review, APACHE II had the highest positive predictive value (69%) for mortality among scoring systems [19]. 446
Pancreatology 2011;11:445–452
Although the Ranson, APACHE, Glasgow and Imrie scoring systems use the total white blood cell count (WBC), it has been demonstrated that the neutrophillymphocyte ratio (NLR) is superior to WBC in the prediction of adverse outcomes in a variety of medical and surgical conditions (i.e. appendicitis, critical care illness and acute coronary syndrome) [20–26]. Recognizing the need for improving acute pancreatitis scoring systems’ ability to predict adverse outcomes, we studied whether NLR is superior to WBC in the prediction of adverse outcomes. In our literature search, we found no previous study that investigated the value of NLR as a predictor of adverse outcomes in acute pancreatitis. Methods This observational study explored the value of NLR as a predictor of defined adverse outcomes among 339 patients admitted to a tertiary care center between January 2005 and January 2008 with a diagnosis of acute pancreatitis. The study inclusion criteria required the clinical diagnosis of acute pancreatitis with an elevated serum amylase and lipase and/or positive CT scan findings of acute pancreatitis. The exclusion criteria included any of the following: age greater than 80 years, a diagnosis of cancer or hematological proliferative disease under treatment, current steroid or chemotherapy for any reason, normal findings on amylase and lipase testing and failure of finding changes of pancreatitis on CT examination, and unavailable complete blood counts or medical records (fig. 1). Of the 339 cases identified, 283 patients were eligible for inclusion in the study.
Azab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann
Table 1. Baseline characteristics of patients with acute pancreatitis sorted by tertiles of the admission NLR
Demographic characteristics Male Age, years Race (Caucasian) BMI Obesity (BMI >30) History of risk factors Prior acute pancreatitis Dyslipidemia Hypertension Gallstones Alcohol abuse Smoking Diabetes mellitus Congestive heart failure Obstructive airway disease Coronary artery disease Clinical presentation Days in pain before admission* Respiratory rate, breaths/min Systolic blood pressure, mm Hg Diastolic blood pressure, mm Hg Heart rate, beats/min Temperature, °C Modified early warning score In-hospital laboratory data Peak amylase, U/l Peak lipase, U/l AST, U/l BUN, mg/dl Calcium, mg/dl Hematocrit, % Serum creatinine, mg/dl Serum glucose, mg/dl White blood cells, !103/l Lymphocytes, !103/l Monocytes, !103/l Neutrophils, !103/l
NLR <3.6 (n = 93)
3.6 ≤ NLR < 7.6 (n = 96)
NLR ≥7.6 (n = 94)
p
50 (53.8) 46.3812.10 62 (67.7) 29.586.30 39 (43.8)
54 (56.3) 48.5814.50 75 (78.1) 29.987.00 44 (47.3)
52 (55.3) 50.0814.70 79 (84.0) 30.487.00 42 (46.7)
0.9430 0.1004 0.0284 0.2761 0.8963
35 (37.6) 21 (22.6) 40 (43.0) 20 (21.5) 37 (39.8) 57 (61.3) 20 (21.5) 2 (2.2) 6 (6.5) 14 (15.1)
38 (39.6) 26 (27.0) 35 (36.5) 18 (18.6) 44 (45.8) 47 (49.0) 18 (19.8) 3 (3.1) 3 (3.1) 15 (15.6)
32 (34.0) 31 (33.0) 47 (50.0) 22 (23.4) 43 (45.7) 49 (52.1) 19 (20.4) 2 (2.1) 2 (2.1) 17 (18.1)
0.7326 0.2819 0.1747 0.7449 0.6461 0.2131 0.9820 1.0000 0.2596 0.8441
3.986.20 18.082.20 183.0821.00 82.0811.80 84.0814.60 36.680.40 1.180.70
2.583.20 18.382.20 139.0824.70 80.0812.90 85.0813.00 36.780.60 1.280.50
719.081,222.00 788.081,680.00 76.0893.00 14.3814.80 9.080.60 39.984.70 1.080.30 126.0841.00 8.383.30 2.381.00 0.680.30 5.082.60
879.081,219.00 972.081,662.00 108.08174.00 14.089.20 9.280.70 42.285.40 1.080.70 163.08101.00 10.983.90 1.580.60 0.780.30 5.581.20
2.784.28 18.382.90 137.0823.90 78.0814.40 85.0817.60 36.880.70 1.481.00
0.0951 0.8614 0.9392 0.1999 0.8538 0.2834 0.0780
1,338.081,354.00 1,738.082,338.00 148.08240.00 17.6825.50 9.480.50 42.786.00 1.180.80 163.0891.00 13.484.50 0.980.40 0.780.40 11.883.90
0.0001 0.0004 0.0615 0.4932 0.0049 <0.0001 0.1355 0.0003 <0.0001 <0.0001 0.2240 <0.0001
Data are presented in either means 8 SD or n (%). AST = Aspartate transaminase; BUN = blood urea nitrogen. * The medians of the number of days in pain before admission were 2, 1 and 1 for the 1st, 2nd, and 3rd NLR tertiles, respectively.
White blood cell, neutrophil, lymphocyte and monocyte absolute counts, as well as hematocrit, were obtained from the initial complete blood count performed within 1 h of admission. Differential leukocyte counts were obtained by the Coulter Counter technique (Coulter Gen-S Hematology Analyzer; Beckman Coulter Corp., Hialeah, Fla., USA). Two physicians independently reviewed the electronic medical records to determine the possible etiology of acute pancreatitis and for potential confounders, in-
cluding demographic variables, history of comorbidities, acute pancreatitis risk factors and clinical prognostic risk factors at presentation; admission and peak serum amylase and lipase; admission serum creatinine, glucose, calcium, aminotransferases and blood urea nitrogen; and radiological procedures and interventions (table 1). Modified EWS (MEWS; i.e. the four vital signs and the level of consciousness of the patients) was calculated for each eligible index case [27]. MEWS was used because of its applicabil-
NLR in Acute Pancreatitis
Pancreatology 2011;11:445–452
447
Frequency (%)
Results 18 16 14 12 10 8 6 4 2 0
Frequency (%)
a
b
17.0 TPN
ICU 9.4 5.2
2.2
2.2
NLR <3.6 14 12 10 8 6 4 2 0
7.4
3.6 ≤ NLR < 7.6
NLR ≥7.6 12.6
9.9 6.2
7.7
7.4
0 WBC <8.7
8.7 ≤ WBC < 12.3
WBC ≥12.3
Fig. 2. a Incidence of TPN and ICU according to NLR tertiles. b TPN and ICU according to WBC tertiles.
ity when there was missing data that were needed to calculate other severity systems scores. Patients were stratified into tertile models according to each leukocyte parameter (WBC, neutrophil, lymphocyte and NLR). The primary outcomes for this study were the length of stay (LOS) and the need for admission to the ICU. Total parenteral nutrition (TPN) management was initiated by a nutritionist for the patients who could not tolerate enteral feeding after the 5th day of hospitalization. The emergency room attending, intensivist, gastroenterologist and nutritionist who made the decisions of ICU admission, TPN management and hospital discharge were not aware of the study design. The primary objective was to determine the value of admission NLR as a predictor of adverse outcomes of acute pancreatitis at the time of presentation of the patient to the hospital. The study protocol was approved by our institutional review board. Statistical Methods The distributions of continuous variables were presented as means and standard deviations; the distributions of categorical variables were presented as frequencies and percentages. Group comparisons used the Kruskal-Wallis test for continuous variables and Fisher’s exact test for categorical data. Odds ratio point estimates with confidence intervals were used to compare the individual NLR tertiles in predicting adverse outcomes. Logistic regression was used to determine the degree of association between the ICU admission and LOS outcomes and each of the leukocyte parameters. The logistic model was also run with potential confounding covariates: age, diabetes mellitus, MEWS and glucose at admission. Prior to constructing these models, univariate screening of potential confounding variables was carried out and only the aforementioned variables were significantly associated with tertile; accordingly, these four variables were entered into the model as covariates. Areas under the receiver operating characteristic (ROC) curves were computed for both NLR and WBC tertiles and compared. All analyses were carried out using SAS Version 9.2 (SAS Institute, Cary, N.C., USA).
448
Pancreatology 2011;11:445–452
The baseline characteristics of the included patients were stratified into equal tertiles according to their admission NLR as presented in table 1. The following variables increased as NLR tertile increased: hematocrit (p ! 0.0001), serum calcium (p ! 0.0049), serum glucose (p ! 0.0003), serum peak amylase (p ! 0.0001) and lipase (p ! 0.0004) levels. The higher NLR tertiles were associated with a higher prevalence of Caucasian race (p ! 0.0284). The remaining baseline variables were not significantly associated with NLR tertile. Figure 2a, b illustrates the frequency of ICU admission and TPN management in patients according to NLR and WBC tertiles, respectively. We observed a negative correlation between NLR and the number of days in pain before presentation (Spearman’s correlation coefficient, Rho = –0.0150, p = 0.0133). Out of 283 patients, there were 22 severe cases of acute pancreatitis with MEWS 63. NLR was significantly higher in severe cases (mean = 11.4, SD = 7.11) compared to those with mild-to-moderate acute pancreatitis (mean = 7.3, SD = 7.34; p = 0.0123 by two-tailed unpaired t test). In contrast, WBC was not significantly different between the severe cases (mean = 11.5, SD = 3.96) compared to mildto-moderate cases (mean = 10.8, SD = 4.47; p = 0.4647). Adverse Outcomes Out of 283 patients, 4 patients died in-hospital (WBC: 9.0, 13.3, 15.5 and 18.7; and NLR: 5, 12, 7 and 9, respectively). There were 36 patients who required ICU (n = 27), TPN (n = 14), mechanical ventilation (n = 4) or died inhospital (n = 4). Figure 2 illustrates the increasing incidence of ICU and TPN in the NLR and WBC tertiles. Table 2 shows the admission rates to ICU according to NLR, WBC, lymphocyte and neutrophil tertiles. The admission rates for ICU increased significantly as the NLR tertile increased (p ! 0.0011). There was no association between ICU admission and either WBC, lymphocyte or neutrophil tertiles. According to NLR tertiles, patients in the 3rd tertile (NLR 67.6) had statistically significant longer average LOS (6.2 vs. 4.2 days, p ! 0.002) compared with those in the 1st tertile (NLR !3.6). However, according to the total WBC tertiles, patients in the 3rd tertile had a nonsignificant difference in average LOS (5.8 vs. 4.4 days, p = 0.059) compared to patients in the 1st WBC tertile. Multivariate Analysis for ICU Admission In order to adjust for the effects of potentially confounding variables, multiple logistic regression was perAzab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann
Table 2. Univariate analysis of adverse outcomes in each leukocyte parameter tertile 1.00
Predictor
Frequency of ICU admission 1st tertile
2nd tertile
3rd tertile
2.2 6.2 10.9 5.2
9.4 9.9 11.7 10.8
17.0 12.6 5.6 12.8
p
<0.0011 0.1322 0.2253 0.0799
Sensitivity
0.75
NLR WBC Lymphocyte Neutrophil
0.50
All the frequencies are presented as percentages. 0.25 ROC curve (area) NLR (0.6495) WBC (0.6001)
Table 3. Multivariate analysis model for adverse outcomes in acute pancreatitis
Predictor
NLR tertiles WBC tertiles Age (per year) Diabetes MEWS Serum glucose
0 0
0.25
Frequency of ICU admission OR
95% CI
p
2.1551 1.0481 1.034 0.539 2.132 1.003
1.095–4.240 0.576–1.909 1.000–1.069 0.171–1.698 1.305–3.484 0.998–1.007
0.0262 0.8779 0.0471 0.2911 0.0025 0.2338
1 OR
is the risk of ICU admission when moving from one tertile to the next.
formed to predict ICU admission as a function of NLR tertile, WBC tertile and four clinical variables: age, diabetes mellitus, MEWS and glucose at admission. These four variables were selected by investigating the association of all variables in table 1 with ICU admission and selecting those variables with significant associations as potential confounders. Using this multivariable model, only NLR tertiles (p ! 0.0262) and MEWS (p ! 0.0025) were significant predictors of ICU admission. By using NLR as a continuous variable with the same aforementioned confounding variables, only MEWS (p ! 0.0004), glucose (p ! 0.0122) and age (p ! 0.0155) were significant predictors of ICU admission (table 3). Multivariate Analysis for LOS In a model including the same confounding predictors (age, MEWS, glucose, NLR and WBC), only NLR (p ! 0.0161) and serum glucose (p ! 0.0053) on admission were NLR in Acute Pancreatitis
0.50 1 – Specificity
0.75
1.00
Fig. 3. ROC curve of NLR versus WBC in predicting ICU admis-
sion for acute pancreatitis.
significant predictors of LOS. In this model, NLR had a HR of 0.88 (95% CI: 0.81–0.97, p ! 0.007). This is interpreted to mean that, at any given time, for every log2 NLR, the likelihood of being discharged decreases by a relative 22%. ROC Curves of NLR and WBC in Predicting ICU and LOS Figure 3 illustrates the superiority of NLR (area under the curve 0.6495) to predict ICU admission in comparison to WBC (area under the curve 0.6001). When LOS is dichotomized to 17 days versus ^7 days, the ROC curve for the above NLR-glucose model has an area under the curve of 0.7041 (fig. 4). We extrapolated NLR 64.7 as the left uppermost point in both the ICU and LOS ROC curves. NLR 64.7 predicted ICU admission and LOS 17 days with a sensitivity of 85.2 and 75.0%, specificity of 47.7 and 48.1%, positive predictive value of 14.6 and 24.5%, and negative predictive value of 96.8 and 89.5%, respectively. According to the ROC curve, NLR values were superior in predicting ICU admission compared to WBC values.
Pancreatology 2011;11:445–452
449
1.00
Sensitivity
0.75
0.50
0.25 ROC curve (area) Model (0.7041) NLRgluc (0.7041) 0 0
0.25
0.50 1 – Specificity
0.75
1.00
Fig. 4. ROC curve of NLR and serum glucose (NLRgluc) model in
predicting hospitalization 17 days for acute pancreatitis.
Discussion
The main finding in our study was that NLR was superior to total leukocyte count and individual absolute neutrophil and lymphocyte counts in prediction of the severity of acute pancreatitis upon presentation to the hospital. We hypothesize that the predictive superiority of NLR is attributed to the alterations of WBC by various physiological and pathological conditions (e.g. hydration status, stress, pregnancy) [28], as well as the inconsistency associated with drawing and handling of blood specimens. While these variables can change the absolute WBC and its individual subtypes, NLR should remain more stable. Furthermore, NLR represents the contrast of two complimentary, yet contradictory, immune pathways. The neutrophil is the main cell responsible for the active nonspecific inflammatory response and the main initiator of destructive tissue cascades by secretion of myeloperoxidase, elastase, interleukin-1 and interleukin-6 [29]. These inflammatory mediators have been shown to play a significant role in the systemic response to acute pancreatitis [30]. In contrast, the lymphocytic immune response occurs later and its main function is to mediate and resolve the preceding nonspecific inflammatory process. Prior studies have shown an asso450
Pancreatology 2011;11:445–452
ciation between lower peripheral lymphocyte counts and the severity of cases of acute pancreatitis [31–33]. Such decline in lymphocyte count in acute pancreatitis was attributed to apoptosis and was also associated with lymphocyte dysfunction [34, 35]. Additionally, prior studies have shown that a low lymphocyte count, together with a high neutrophil count, is associated with adverse outcomes in various other medical and surgical conditions [18, 24, 25]. In our study, we noticed a trend of more adverse outcomes with low lymphocyte and high neutrophil counts. However, neither has a significant p value in contrast to NLR. Previous studies have shown higher neutrophil and WBC in the Caucasian population compared to AfricanAmericans [36, 37]. This reflects the known racial difference regardless of the presence of pancreatitis. Similarly, we noticed that the higher NLR tertiles were significantly associated with Caucasian-Americans in comparison to the lower tertile. Among the rest of the baseline characteristics, we also noticed significantly higher serum glucose, peak amylase, peak lipase and hematocrit values in the higher NLR tertiles. In our study, only NLR and MEWS remained statistically significant predictors of the severity of acute pancreatitis upon presentation of the patient to the hospital. Accordingly, we suggest that if larger studies are confirmatory, NLR should be considered for inclusion into the established scoring systems. We found that the inclusion of NLR increased the predictability of illness severity in acute pancreatitis when added to the multivariate model with MEWS, age and diabetes regarding LOS and ICU admission. NLR is a simple, cheap and routine part of a CBC done during initial ER evaluation of patients. In an effort to define a suggested NLR cutoff value, we believe that the sensitivity of such a value is of more clinical importance than its specificity, as conservative and supportive management is the initial standard of care for patients presenting with acute pancreatitis. According to our data, we suggest using the cutoff value of 14.7 as a warning sign for patients presenting with acute pancreatitis because this value demonstrated satisfactory sensitivity although it had a low specificity of about 50% to predict the surrogates we used for the severity of acute pancreatitis. Although we found no normal range of NLR in the published literature, we used the available studies that evaluated WBC and differential counts in different populations to extrapolate the highest possible NLR by dividing the highest measured neutrophil count by the lowest measured lymphocyte count in the American popAzab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann
ulation [36, 37]. NLR of 4.7 was found to be the average highest extrapolated NLR in Caucasian- and AfricanAmerican populations. Limitations Because of the low number of adverse outcomes, we were only able to assess the NLR association with ICU admission and LOS rather than mortality and organ failure. The value of NLR after admission and throughout the hospital course was not evaluated in our study. In addition, we had a mostly Caucasian-American population. The causality role of NLR in acute pancreatitis could not be evaluated as this was an observational study.
Conclusion
NLR is superior to total WBC in predicting the severity of the acute pancreatitis. We recommend considering including NLR rather than total WBC in the risk scoring systems. NLR may improve the sensitivity of these scoring systems for predicting adverse outcomes in cases of acute pancreatitis that require hospitalization. NLR rather than WBC could be used in triaging patients. We suggest that a larger prospective multicenter study be conducted.
Disclosure Statement All authors have no conflict of interest or relationship to industry to disclose, and no grants were received. Some of the data in the report were presented at Digestive Disease Week 2010, New Orleans, La., USA, earning a ‘Poster of Distinction Award’.
References 1 Neoptolemos JP, Raraty M, Finch M, Sutton R: Acute pancreatitis: the substantial human and financial costs. Gut 1998; 42:886–891. 2 Vege SS YD, Chari ST: pancreatitis; in Talley NJ, Locke GR, Saito YA (eds): GI Epidemiology, ed 1. Malden, Blackwell Publishing, 2007, p 221. 3 Russo MW, Wei JT, Thiny MT, Gangarosa LM, Brown A, Ringel Y, Shaheen NJ, Sandler RS: Digestive and liver diseases statistics, 2004. Gastroenterology 2004; 126: 1448– 1453. 4 Sandler RS, Everhart JE, Donowitz M, Adams E, Cronin K, Goodman C, Gemmen E, Shah S, Avdic A, Rubin R: The burden of selected digestive diseases in the United States. Gastroenterology 2002; 122:1500–1511. 5 Steinberg WM: Predictors of severity of acute pancreatitis. Gastroenterol Clin North Am 1990;19:849–861. 6 Lankisch PG, Schirren CA, Kunze E: Undetected fatal acute pancreatitis: why is the disease so frequently overlooked? Am J Gastroenterol 1991; 86:322–326. 7 Windsor JA: Search for prognostic markers for acute pancreatitis. Lancet 2000; 355: 1924–1925. 8 Balthazar EJ, Robinson DL, Megibow AJ, Ranson JH: Acute pancreatitis: value of CT in establishing prognosis. Radiology 1990; 174:331–336. 9 Chatzicostas C, Roussomoustakaki M, Vardas E, Romanos J, Kouroumalis EA: Balthazar computed tomography severity index is superior to Ranson criteria and APACHE II and III scoring systems in predicting acute pancreatitis outcome. J Clin Gastroenterol 2003;36:253–260.
NLR in Acute Pancreatitis
10 Garcea G, Jackson B, Pattenden CJ, Sutton CD, Neal CP, Dennison AR, Berry DP: Early warning scores predict outcome in acute pancreatitis. J Gastrointest Surg 2006; 10: 1008–1015. 11 Blamey SL, Imrie CW, O’Neill J, Gilmour WH, Carter DC: Prognostic factors in acute pancreatitis. Gut 1984;25:1340–1346. 12 Corfield AP, Cooper MJ, Williamson RC, Mayer AD, McMahon MJ, Dickson AP, Shearer MG, Imrie CW: Prediction of severity in acute pancreatitis: prospective comparison of three prognostic indices. Lancet 1985;2:403–407. 13 Dominguez-Munoz JE, Carballo F, Garcia MJ, de Diego JM, Campos R, Yangüela J, de la Morena J: Evaluation of the clinical usefulness of APACHE II and SAPS systems in the initial prognostic classification of acute pancreatitis: a multicenter study. Pancreas 1993; 8:682–686. 14 Ranson JH: Etiological and prognostic factors in human acute pancreatitis: a review. Am J Gastroenterol 1982; 77:633–638. 15 Wilson C, Heath DI, Imrie CW: Prediction of outcome in acute pancreatitis: a comparative study of APACHE II, clinical assessment and multiple factor scoring systems. Br J Surg 1990;77:1260–1264. 16 Malfertheiner P, Dominguez-Munoz JE: Prognostic factors in acute pancreatitis. Int J Pancreatol 1993; 14:1–8. 17 Triantopoulou C, Lytras D, Maniatis P, Chrysovergis D, Manes K, Siafas I, Papailiou J, Dervenis C: Computed tomography versus Acute Physiology and Chronic Health Evaluation II score in predicting severity of acute pancreatitis: a prospective, comparative study with statistical evaluation. Pancreas 2007;35:238–242.
18 Garcea G, Gouda M, Hebbes C, Ong SL, Neal CP, Dennison AR, Berry DP: Predictors of severity and survival in acute pancreatitis: validation of the efficacy of early warning scores. Pancreas 2008;37:e54–e61. 19 Gravante G, Garcea G, Ong SL, Metcalfe MS, Berry DP, Lloyd DM, Dennison AR: Prediction of mortality in acute pancreatitis: a systematic review of the published evidence. Pancreatology 2009; 9:601–614. 20 Gibson PH, Croal BL, Cuthbertson BH, Small GR, Ifezulike AI, Gibson G, Jeffrey RR, Buchan KG, El-Shafei H, Hillis GS: Preoperative neutrophil-lymphocyte ratio and outcome from coronary artery bypass grafting. Am Heart J 2007;154:995–1002. 21 Gillum RF, Mussolino ME, Madans JH: Counts of neutrophils, lymphocytes, and monocytes, cause-specific mortality and coronary heart disease: the NHANES-I epidemiologic follow-up study. Ann Epidemiol 2005;15:266–271. 22 Halazun KJ, Hardy MA, Rana AA, Woodland DC 4th, Luyten EJ, Mahadev S, Witkowski P, Siegel AB, Brown RS Jr, Emond JC: Negative impact of neutrophil-lymphocyte ratio on outcome after liver transplantation for hepatocellular carcinoma. Ann Surg 2009;250:141–151. 23 Sarraf KM, Belcher E, Raevsky E, Nicholson AG, Goldstraw P, Lim E: Neutrophil/lymphocyte ratio and its association with survival after complete resection in non-small cell lung cancer. J Thorac Cardiovasc Surg 2009;137:425–428.
Pancreatology 2011;11:445–452
451
24 Tamhane UU, Aneja S, Montgomery D, Rogers EK, Eagle KA, Gurm HS: Association between admission neutrophil to lymphocyte ratio and outcomes in patients with acute coronary syndrome. Am J Cardiol 2008;102: 653–657. 25 Zahorec R: Ratio of neutrophil to lymphocyte counts – rapid and simple parameter of systemic inflammation and stress in critically ill. Bratisl Lek Listy 2001;102:5–14. 26 Goodman DA, Goodman CB, Monk JS: Use of the neutrophil:lymphocyte ratio in the diagnosis of appendicitis. Am Surg 1995; 61: 257–259. 27 Subbe CP, Kruger M, Rutherford P, Gemmel L: Validation of a modified early warning score in medical admissions. QJM 2001; 94: 521–526. 28 Garrey W: Variation in white blood cell counts. Physiol Rev 1935; 15:597–638.
452
29 Skubitz K: Neutrophilic leukocytes; in Greer J, Wintrobe MM (eds): Wintrobe’s Clinical Hematology, ed 12. Philadelphia, Wolters Kluwer/Lippincott Williams & Wilkins Health, 2009, pp 170–193. 30 Wittel UA, Rau B, Gansauge F, Gansauge S, Nussler AK, Beger HG, Poch B: Influence of PMN leukocyte-mediated pancreatic damage on the systemic immune response in severe acute pancreatitis in rats. Dig Dis Sci 2004;49:1348–1357. 31 Pavlov P, Uchikov P, Murdzheva M, Tuleva S, Tsvetkova T: Main lymphocyte populations and their subpopulations in patients with acute pancreatitis studied in the course of disease (in Bulgarian). Khirurgiia (Sofiia) 2001;57:4–11. 32 Pezzilli R, Billi P, Beltrandi E, Casadei Maldini M, Mancini R: Impaired lymphocyte proliferation in human acute pancreatitis. Digestion 1997;58:431–436. 33 Pezzilli R, Billi P, Beltrandi E, Maldini M, Mancini R, Morselli Labate AM, Miglioli M: Circulating lymphocyte subsets in human acute pancreatitis. Pancreas 1995;11:95–100.
Pancreatology 2011;11:445–452
34 Pezzilli R, Maldini M, Morselli-Labate AM, Barakat B, Romboli E, Beltrandi E, Migliori M, Tomassetti P, Corinaldesi R: Early activation of peripheral lymphocytes in human acute pancreatitis. J Clin Gastroenterol 2003; 36:360–363. 35 Takeyama Y, Takas K, Ueda T, Hori Y, Goshima M, Kuroda Y: Peripheral lymphocyte reduction in severe acute pancreatitis is caused by apoptotic cell death. J Gastrointest Surg 2000;4:379–387. 36 Bain BJ: Ethnic and sex differences in the total and differential white cell count and platelet count. J Clin Pathol 1996;49:664– 666. 37 Lichtman MA: Cellular deformability during maturation of the myeloblast. Possible role in marrow egress. N Engl J Med 1970; 283:943–948.
Azab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann
Pancreatology 2011;11:445–452 DOI: 10.1159/000331494
Neutrophil-Lymphocyte Ratio as a Predictor of Adverse Outcomes of Acute Pancreatitis Basem Azab a Neil Jaglall a Jean Paul Atallah a Ari Lamet a Venkat Raja-Surya a Bachir Farah a Martin Lesser c Warren D. Widmann b, d
Departments of a Internal Medicine and b Surgery, Staten Island University Hospital, Staten Island, N.Y., c Feinstein Institute for Medical Research and Hofstra School of Medicine, Manhasset, N.Y., and d State University of New York Downstate Medical Center, Brooklyn, N.Y., USA
Key Words Neutrophil-lymphocyte ratio Acute pancreatitis Severity Scoring risk system
Abstract Background: Most acute pancreatitis risk scoring systems use total white blood cell counts (WBC) as one of the risk factors. The value of the neutrophil-lymphocyte ratio (NLR) to predict the severity of acute pancreatitis has not been previously evaluated. Methods: This observational study included 283 patients admitted to a tertiary center between 2004 and 2007. The patients were arranged into tertiles according to NLR and WBC values. The primary outcomes were intensive care unit (ICU) admission and length of stay (LOS) in the hospital. Results: According to NLR tertiles, patients in the 3rd tertile (NLR 67.6) had significantly more ICU admissions (17 vs. 2.2%, p ! 0.0001) and longer average LOS (6.2 vs. 4.2 days, p ! 0.002) compared with those in the 1st tertile (NLR !3.6). According to WBC tertiles, patients in the 3rd tertile had more ICU admissions (12.6 vs. 6.2%, p = 0.12) and a longer average LOS (5.8 vs. 4.4 days, p = 0.059) compared to patients in the 1st WBC tertile, but this did not reach statistical significance. In the multivariate model including NLR, WBC and other predictors, only
© 2011 S. Karger AG, Basel and IAP 1424–3903/11/0114–0445$38.00/0 Fax +41 61 306 12 34 E-Mail [email protected] www.karger.com
Accessible online at: www.karger.com/pan
NLR tertiles (p ! 0.0262) and modified early warning scores (p ! 0.0025) were significant predictors of ICU admission. Likewise, in the multivariate model of LOS, only NLR and glucose level were significant predictors of longer LOS (p ! 0.0161 and p ! 0.0053, respectively). Conclusion: NLR is superior to total WBC in predicting adverse outcomes of acute pancreatitis. According to our data, we suggest using the NLR cutoff value of 14.7 as a simple indicator of severity in patients presenting with acute pancreatitis. Copyright © 2011 S. Karger AG, Basel and IAP
Background
The incidence of acute pancreatitis in industrialized countries varies from 4.9 to 35 per 100,000 persons [1, 2]. In the United States, acute pancreatitis accounts for 327,474 hospitalizations and 2,834 deaths annually [3, 4]. The clinical signs of acute pancreatitis may be nonspecific and they exhibit a low sensitivity (!40%) for the prediction of an adverse outcome [5]. Also, there is a significant chance of underdiagnosis, as demonstrated by the finding of previously undiagnosed acute pancreatitis as a cause of death in some autopsy series [6]. Because of the
Basem Azab, MD Department of Internal Medicine Staten Island University Hospital, 475 Seaview Avenue Staten Island, NY 10305 (USA) Tel. +1 347 607 8727, E-Mail basemnady2000 @ yahoo.com
339 patients were assessed for eligibility 56 patients were excluded Age >80, n = 27 Hemoproliferative disease, n = 4 Steroid or chemotherapy use, n = 6 Negative enzymes and/or CT findings, n = 5 Unavailable records, n = 14 283 patients included in study
1st NLR tertile: 93 patients
2nd NLR tertile: 96 patients
3rd NLR tertile: 94 patients
Fig. 1. Schematic illustration of study.
potential deceptive presentation of acute pancreatitis, multiple severity scoring systems have been developed to guide clinicians in triaging patients who are likely to require vigorous treatment in an intensive care unit (ICU). Ideally, these scoring systems should use predictive tools that are rapid, reproducible, inexpensive, minimally invasive and highly accurate [7]. Among the available scoring systems currently used, Balthazar [8, 9] and the early warning score (EWS) [10] use CT and physical exam findings, respectively. The other scoring systems (Ranson criteria, APACHE score, Glasgow and Imrie scores) [11–15] use combined clinical and laboratory variables. However, there is no universally adopted scoring system in clinical practice, due to their low sensitivity and the complexity of the needed calculations [5, 16]. Furthermore, some (i.e. Ranson, Glasgow and Imrie) need nonroutine blood work and have not been validated beyond 48 h. Although the Balthazar score is superior to other scoring systems in the prediction of the extent of pancreatic necrosis, it was found inferior to the Ranson and APACHE systems in the prediction of organ failure and other adverse outcomes during hospitalization [9, 17]. Garcea et al. [18] found that the APACHE II score and EWS were superior to other scoring systems (i.e. modified organ dysfunction scores, Imrie scores, Balthazar and Ranson criteria) in predicting severity and survival. In a recent systemic review, APACHE II had the highest positive predictive value (69%) for mortality among scoring systems [19]. 446
Pancreatology 2011;11:445–452
Although the Ranson, APACHE, Glasgow and Imrie scoring systems use the total white blood cell count (WBC), it has been demonstrated that the neutrophillymphocyte ratio (NLR) is superior to WBC in the prediction of adverse outcomes in a variety of medical and surgical conditions (i.e. appendicitis, critical care illness and acute coronary syndrome) [20–26]. Recognizing the need for improving acute pancreatitis scoring systems’ ability to predict adverse outcomes, we studied whether NLR is superior to WBC in the prediction of adverse outcomes. In our literature search, we found no previous study that investigated the value of NLR as a predictor of adverse outcomes in acute pancreatitis. Methods This observational study explored the value of NLR as a predictor of defined adverse outcomes among 339 patients admitted to a tertiary care center between January 2005 and January 2008 with a diagnosis of acute pancreatitis. The study inclusion criteria required the clinical diagnosis of acute pancreatitis with an elevated serum amylase and lipase and/or positive CT scan findings of acute pancreatitis. The exclusion criteria included any of the following: age greater than 80 years, a diagnosis of cancer or hematological proliferative disease under treatment, current steroid or chemotherapy for any reason, normal findings on amylase and lipase testing and failure of finding changes of pancreatitis on CT examination, and unavailable complete blood counts or medical records (fig. 1). Of the 339 cases identified, 283 patients were eligible for inclusion in the study.
Azab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann
Table 1. Baseline characteristics of patients with acute pancreatitis sorted by tertiles of the admission NLR
Demographic characteristics Male Age, years Race (Caucasian) BMI Obesity (BMI >30) History of risk factors Prior acute pancreatitis Dyslipidemia Hypertension Gallstones Alcohol abuse Smoking Diabetes mellitus Congestive heart failure Obstructive airway disease Coronary artery disease Clinical presentation Days in pain before admission* Respiratory rate, breaths/min Systolic blood pressure, mm Hg Diastolic blood pressure, mm Hg Heart rate, beats/min Temperature, °C Modified early warning score In-hospital laboratory data Peak amylase, U/l Peak lipase, U/l AST, U/l BUN, mg/dl Calcium, mg/dl Hematocrit, % Serum creatinine, mg/dl Serum glucose, mg/dl White blood cells, !103/l Lymphocytes, !103/l Monocytes, !103/l Neutrophils, !103/l
NLR <3.6 (n = 93)
3.6 ≤ NLR < 7.6 (n = 96)
NLR ≥7.6 (n = 94)
p
50 (53.8) 46.3812.10 62 (67.7) 29.586.30 39 (43.8)
54 (56.3) 48.5814.50 75 (78.1) 29.987.00 44 (47.3)
52 (55.3) 50.0814.70 79 (84.0) 30.487.00 42 (46.7)
0.9430 0.1004 0.0284 0.2761 0.8963
35 (37.6) 21 (22.6) 40 (43.0) 20 (21.5) 37 (39.8) 57 (61.3) 20 (21.5) 2 (2.2) 6 (6.5) 14 (15.1)
38 (39.6) 26 (27.0) 35 (36.5) 18 (18.6) 44 (45.8) 47 (49.0) 18 (19.8) 3 (3.1) 3 (3.1) 15 (15.6)
32 (34.0) 31 (33.0) 47 (50.0) 22 (23.4) 43 (45.7) 49 (52.1) 19 (20.4) 2 (2.1) 2 (2.1) 17 (18.1)
0.7326 0.2819 0.1747 0.7449 0.6461 0.2131 0.9820 1.0000 0.2596 0.8441
3.986.20 18.082.20 183.0821.00 82.0811.80 84.0814.60 36.680.40 1.180.70
2.583.20 18.382.20 139.0824.70 80.0812.90 85.0813.00 36.780.60 1.280.50
719.081,222.00 788.081,680.00 76.0893.00 14.3814.80 9.080.60 39.984.70 1.080.30 126.0841.00 8.383.30 2.381.00 0.680.30 5.082.60
879.081,219.00 972.081,662.00 108.08174.00 14.089.20 9.280.70 42.285.40 1.080.70 163.08101.00 10.983.90 1.580.60 0.780.30 5.581.20
2.784.28 18.382.90 137.0823.90 78.0814.40 85.0817.60 36.880.70 1.481.00
0.0951 0.8614 0.9392 0.1999 0.8538 0.2834 0.0780
1,338.081,354.00 1,738.082,338.00 148.08240.00 17.6825.50 9.480.50 42.786.00 1.180.80 163.0891.00 13.484.50 0.980.40 0.780.40 11.883.90
0.0001 0.0004 0.0615 0.4932 0.0049 <0.0001 0.1355 0.0003 <0.0001 <0.0001 0.2240 <0.0001
Data are presented in either means 8 SD or n (%). AST = Aspartate transaminase; BUN = blood urea nitrogen. * The medians of the number of days in pain before admission were 2, 1 and 1 for the 1st, 2nd, and 3rd NLR tertiles, respectively.
White blood cell, neutrophil, lymphocyte and monocyte absolute counts, as well as hematocrit, were obtained from the initial complete blood count performed within 1 h of admission. Differential leukocyte counts were obtained by the Coulter Counter technique (Coulter Gen-S Hematology Analyzer; Beckman Coulter Corp., Hialeah, Fla., USA). Two physicians independently reviewed the electronic medical records to determine the possible etiology of acute pancreatitis and for potential confounders, in-
cluding demographic variables, history of comorbidities, acute pancreatitis risk factors and clinical prognostic risk factors at presentation; admission and peak serum amylase and lipase; admission serum creatinine, glucose, calcium, aminotransferases and blood urea nitrogen; and radiological procedures and interventions (table 1). Modified EWS (MEWS; i.e. the four vital signs and the level of consciousness of the patients) was calculated for each eligible index case [27]. MEWS was used because of its applicabil-
NLR in Acute Pancreatitis
Pancreatology 2011;11:445–452
447
Frequency (%)
Results 18 16 14 12 10 8 6 4 2 0
Frequency (%)
a
b
17.0 TPN
ICU 9.4 5.2
2.2
2.2
NLR <3.6 14 12 10 8 6 4 2 0
7.4
3.6 ≤ NLR < 7.6
NLR ≥7.6 12.6
9.9 6.2
7.7
7.4
0 WBC <8.7
8.7 ≤ WBC < 12.3
WBC ≥12.3
Fig. 2. a Incidence of TPN and ICU according to NLR tertiles. b TPN and ICU according to WBC tertiles.
ity when there was missing data that were needed to calculate other severity systems scores. Patients were stratified into tertile models according to each leukocyte parameter (WBC, neutrophil, lymphocyte and NLR). The primary outcomes for this study were the length of stay (LOS) and the need for admission to the ICU. Total parenteral nutrition (TPN) management was initiated by a nutritionist for the patients who could not tolerate enteral feeding after the 5th day of hospitalization. The emergency room attending, intensivist, gastroenterologist and nutritionist who made the decisions of ICU admission, TPN management and hospital discharge were not aware of the study design. The primary objective was to determine the value of admission NLR as a predictor of adverse outcomes of acute pancreatitis at the time of presentation of the patient to the hospital. The study protocol was approved by our institutional review board. Statistical Methods The distributions of continuous variables were presented as means and standard deviations; the distributions of categorical variables were presented as frequencies and percentages. Group comparisons used the Kruskal-Wallis test for continuous variables and Fisher’s exact test for categorical data. Odds ratio point estimates with confidence intervals were used to compare the individual NLR tertiles in predicting adverse outcomes. Logistic regression was used to determine the degree of association between the ICU admission and LOS outcomes and each of the leukocyte parameters. The logistic model was also run with potential confounding covariates: age, diabetes mellitus, MEWS and glucose at admission. Prior to constructing these models, univariate screening of potential confounding variables was carried out and only the aforementioned variables were significantly associated with tertile; accordingly, these four variables were entered into the model as covariates. Areas under the receiver operating characteristic (ROC) curves were computed for both NLR and WBC tertiles and compared. All analyses were carried out using SAS Version 9.2 (SAS Institute, Cary, N.C., USA).
448
Pancreatology 2011;11:445–452
The baseline characteristics of the included patients were stratified into equal tertiles according to their admission NLR as presented in table 1. The following variables increased as NLR tertile increased: hematocrit (p ! 0.0001), serum calcium (p ! 0.0049), serum glucose (p ! 0.0003), serum peak amylase (p ! 0.0001) and lipase (p ! 0.0004) levels. The higher NLR tertiles were associated with a higher prevalence of Caucasian race (p ! 0.0284). The remaining baseline variables were not significantly associated with NLR tertile. Figure 2a, b illustrates the frequency of ICU admission and TPN management in patients according to NLR and WBC tertiles, respectively. We observed a negative correlation between NLR and the number of days in pain before presentation (Spearman’s correlation coefficient, Rho = –0.0150, p = 0.0133). Out of 283 patients, there were 22 severe cases of acute pancreatitis with MEWS 63. NLR was significantly higher in severe cases (mean = 11.4, SD = 7.11) compared to those with mild-to-moderate acute pancreatitis (mean = 7.3, SD = 7.34; p = 0.0123 by two-tailed unpaired t test). In contrast, WBC was not significantly different between the severe cases (mean = 11.5, SD = 3.96) compared to mildto-moderate cases (mean = 10.8, SD = 4.47; p = 0.4647). Adverse Outcomes Out of 283 patients, 4 patients died in-hospital (WBC: 9.0, 13.3, 15.5 and 18.7; and NLR: 5, 12, 7 and 9, respectively). There were 36 patients who required ICU (n = 27), TPN (n = 14), mechanical ventilation (n = 4) or died inhospital (n = 4). Figure 2 illustrates the increasing incidence of ICU and TPN in the NLR and WBC tertiles. Table 2 shows the admission rates to ICU according to NLR, WBC, lymphocyte and neutrophil tertiles. The admission rates for ICU increased significantly as the NLR tertile increased (p ! 0.0011). There was no association between ICU admission and either WBC, lymphocyte or neutrophil tertiles. According to NLR tertiles, patients in the 3rd tertile (NLR 67.6) had statistically significant longer average LOS (6.2 vs. 4.2 days, p ! 0.002) compared with those in the 1st tertile (NLR !3.6). However, according to the total WBC tertiles, patients in the 3rd tertile had a nonsignificant difference in average LOS (5.8 vs. 4.4 days, p = 0.059) compared to patients in the 1st WBC tertile. Multivariate Analysis for ICU Admission In order to adjust for the effects of potentially confounding variables, multiple logistic regression was perAzab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann
Table 2. Univariate analysis of adverse outcomes in each leukocyte parameter tertile 1.00
Predictor
Frequency of ICU admission 1st tertile
2nd tertile
3rd tertile
2.2 6.2 10.9 5.2
9.4 9.9 11.7 10.8
17.0 12.6 5.6 12.8
p
<0.0011 0.1322 0.2253 0.0799
Sensitivity
0.75
NLR WBC Lymphocyte Neutrophil
0.50
All the frequencies are presented as percentages. 0.25 ROC curve (area) NLR (0.6495) WBC (0.6001)
Table 3. Multivariate analysis model for adverse outcomes in acute pancreatitis
Predictor
NLR tertiles WBC tertiles Age (per year) Diabetes MEWS Serum glucose
0 0
0.25
Frequency of ICU admission OR
95% CI
p
2.1551 1.0481 1.034 0.539 2.132 1.003
1.095–4.240 0.576–1.909 1.000–1.069 0.171–1.698 1.305–3.484 0.998–1.007
0.0262 0.8779 0.0471 0.2911 0.0025 0.2338
1 OR
is the risk of ICU admission when moving from one tertile to the next.
formed to predict ICU admission as a function of NLR tertile, WBC tertile and four clinical variables: age, diabetes mellitus, MEWS and glucose at admission. These four variables were selected by investigating the association of all variables in table 1 with ICU admission and selecting those variables with significant associations as potential confounders. Using this multivariable model, only NLR tertiles (p ! 0.0262) and MEWS (p ! 0.0025) were significant predictors of ICU admission. By using NLR as a continuous variable with the same aforementioned confounding variables, only MEWS (p ! 0.0004), glucose (p ! 0.0122) and age (p ! 0.0155) were significant predictors of ICU admission (table 3). Multivariate Analysis for LOS In a model including the same confounding predictors (age, MEWS, glucose, NLR and WBC), only NLR (p ! 0.0161) and serum glucose (p ! 0.0053) on admission were NLR in Acute Pancreatitis
0.50 1 – Specificity
0.75
1.00
Fig. 3. ROC curve of NLR versus WBC in predicting ICU admis-
sion for acute pancreatitis.
significant predictors of LOS. In this model, NLR had a HR of 0.88 (95% CI: 0.81–0.97, p ! 0.007). This is interpreted to mean that, at any given time, for every log2 NLR, the likelihood of being discharged decreases by a relative 22%. ROC Curves of NLR and WBC in Predicting ICU and LOS Figure 3 illustrates the superiority of NLR (area under the curve 0.6495) to predict ICU admission in comparison to WBC (area under the curve 0.6001). When LOS is dichotomized to 17 days versus ^7 days, the ROC curve for the above NLR-glucose model has an area under the curve of 0.7041 (fig. 4). We extrapolated NLR 64.7 as the left uppermost point in both the ICU and LOS ROC curves. NLR 64.7 predicted ICU admission and LOS 17 days with a sensitivity of 85.2 and 75.0%, specificity of 47.7 and 48.1%, positive predictive value of 14.6 and 24.5%, and negative predictive value of 96.8 and 89.5%, respectively. According to the ROC curve, NLR values were superior in predicting ICU admission compared to WBC values.
Pancreatology 2011;11:445–452
449
1.00
Sensitivity
0.75
0.50
0.25 ROC curve (area) Model (0.7041) NLRgluc (0.7041) 0 0
0.25
0.50 1 – Specificity
0.75
1.00
Fig. 4. ROC curve of NLR and serum glucose (NLRgluc) model in
predicting hospitalization 17 days for acute pancreatitis.
Discussion
The main finding in our study was that NLR was superior to total leukocyte count and individual absolute neutrophil and lymphocyte counts in prediction of the severity of acute pancreatitis upon presentation to the hospital. We hypothesize that the predictive superiority of NLR is attributed to the alterations of WBC by various physiological and pathological conditions (e.g. hydration status, stress, pregnancy) [28], as well as the inconsistency associated with drawing and handling of blood specimens. While these variables can change the absolute WBC and its individual subtypes, NLR should remain more stable. Furthermore, NLR represents the contrast of two complimentary, yet contradictory, immune pathways. The neutrophil is the main cell responsible for the active nonspecific inflammatory response and the main initiator of destructive tissue cascades by secretion of myeloperoxidase, elastase, interleukin-1 and interleukin-6 [29]. These inflammatory mediators have been shown to play a significant role in the systemic response to acute pancreatitis [30]. In contrast, the lymphocytic immune response occurs later and its main function is to mediate and resolve the preceding nonspecific inflammatory process. Prior studies have shown an asso450
Pancreatology 2011;11:445–452
ciation between lower peripheral lymphocyte counts and the severity of cases of acute pancreatitis [31–33]. Such decline in lymphocyte count in acute pancreatitis was attributed to apoptosis and was also associated with lymphocyte dysfunction [34, 35]. Additionally, prior studies have shown that a low lymphocyte count, together with a high neutrophil count, is associated with adverse outcomes in various other medical and surgical conditions [18, 24, 25]. In our study, we noticed a trend of more adverse outcomes with low lymphocyte and high neutrophil counts. However, neither has a significant p value in contrast to NLR. Previous studies have shown higher neutrophil and WBC in the Caucasian population compared to AfricanAmericans [36, 37]. This reflects the known racial difference regardless of the presence of pancreatitis. Similarly, we noticed that the higher NLR tertiles were significantly associated with Caucasian-Americans in comparison to the lower tertile. Among the rest of the baseline characteristics, we also noticed significantly higher serum glucose, peak amylase, peak lipase and hematocrit values in the higher NLR tertiles. In our study, only NLR and MEWS remained statistically significant predictors of the severity of acute pancreatitis upon presentation of the patient to the hospital. Accordingly, we suggest that if larger studies are confirmatory, NLR should be considered for inclusion into the established scoring systems. We found that the inclusion of NLR increased the predictability of illness severity in acute pancreatitis when added to the multivariate model with MEWS, age and diabetes regarding LOS and ICU admission. NLR is a simple, cheap and routine part of a CBC done during initial ER evaluation of patients. In an effort to define a suggested NLR cutoff value, we believe that the sensitivity of such a value is of more clinical importance than its specificity, as conservative and supportive management is the initial standard of care for patients presenting with acute pancreatitis. According to our data, we suggest using the cutoff value of 14.7 as a warning sign for patients presenting with acute pancreatitis because this value demonstrated satisfactory sensitivity although it had a low specificity of about 50% to predict the surrogates we used for the severity of acute pancreatitis. Although we found no normal range of NLR in the published literature, we used the available studies that evaluated WBC and differential counts in different populations to extrapolate the highest possible NLR by dividing the highest measured neutrophil count by the lowest measured lymphocyte count in the American popAzab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann
ulation [36, 37]. NLR of 4.7 was found to be the average highest extrapolated NLR in Caucasian- and AfricanAmerican populations. Limitations Because of the low number of adverse outcomes, we were only able to assess the NLR association with ICU admission and LOS rather than mortality and organ failure. The value of NLR after admission and throughout the hospital course was not evaluated in our study. In addition, we had a mostly Caucasian-American population. The causality role of NLR in acute pancreatitis could not be evaluated as this was an observational study.
Conclusion
NLR is superior to total WBC in predicting the severity of the acute pancreatitis. We recommend considering including NLR rather than total WBC in the risk scoring systems. NLR may improve the sensitivity of these scoring systems for predicting adverse outcomes in cases of acute pancreatitis that require hospitalization. NLR rather than WBC could be used in triaging patients. We suggest that a larger prospective multicenter study be conducted.
Disclosure Statement All authors have no conflict of interest or relationship to industry to disclose, and no grants were received. Some of the data in the report were presented at Digestive Disease Week 2010, New Orleans, La., USA, earning a ‘Poster of Distinction Award’.
References 1 Neoptolemos JP, Raraty M, Finch M, Sutton R: Acute pancreatitis: the substantial human and financial costs. Gut 1998; 42:886–891. 2 Vege SS YD, Chari ST: pancreatitis; in Talley NJ, Locke GR, Saito YA (eds): GI Epidemiology, ed 1. Malden, Blackwell Publishing, 2007, p 221. 3 Russo MW, Wei JT, Thiny MT, Gangarosa LM, Brown A, Ringel Y, Shaheen NJ, Sandler RS: Digestive and liver diseases statistics, 2004. Gastroenterology 2004; 126: 1448– 1453. 4 Sandler RS, Everhart JE, Donowitz M, Adams E, Cronin K, Goodman C, Gemmen E, Shah S, Avdic A, Rubin R: The burden of selected digestive diseases in the United States. Gastroenterology 2002; 122:1500–1511. 5 Steinberg WM: Predictors of severity of acute pancreatitis. Gastroenterol Clin North Am 1990;19:849–861. 6 Lankisch PG, Schirren CA, Kunze E: Undetected fatal acute pancreatitis: why is the disease so frequently overlooked? Am J Gastroenterol 1991; 86:322–326. 7 Windsor JA: Search for prognostic markers for acute pancreatitis. Lancet 2000; 355: 1924–1925. 8 Balthazar EJ, Robinson DL, Megibow AJ, Ranson JH: Acute pancreatitis: value of CT in establishing prognosis. Radiology 1990; 174:331–336. 9 Chatzicostas C, Roussomoustakaki M, Vardas E, Romanos J, Kouroumalis EA: Balthazar computed tomography severity index is superior to Ranson criteria and APACHE II and III scoring systems in predicting acute pancreatitis outcome. J Clin Gastroenterol 2003;36:253–260.
NLR in Acute Pancreatitis
10 Garcea G, Jackson B, Pattenden CJ, Sutton CD, Neal CP, Dennison AR, Berry DP: Early warning scores predict outcome in acute pancreatitis. J Gastrointest Surg 2006; 10: 1008–1015. 11 Blamey SL, Imrie CW, O’Neill J, Gilmour WH, Carter DC: Prognostic factors in acute pancreatitis. Gut 1984;25:1340–1346. 12 Corfield AP, Cooper MJ, Williamson RC, Mayer AD, McMahon MJ, Dickson AP, Shearer MG, Imrie CW: Prediction of severity in acute pancreatitis: prospective comparison of three prognostic indices. Lancet 1985;2:403–407. 13 Dominguez-Munoz JE, Carballo F, Garcia MJ, de Diego JM, Campos R, Yangüela J, de la Morena J: Evaluation of the clinical usefulness of APACHE II and SAPS systems in the initial prognostic classification of acute pancreatitis: a multicenter study. Pancreas 1993; 8:682–686. 14 Ranson JH: Etiological and prognostic factors in human acute pancreatitis: a review. Am J Gastroenterol 1982; 77:633–638. 15 Wilson C, Heath DI, Imrie CW: Prediction of outcome in acute pancreatitis: a comparative study of APACHE II, clinical assessment and multiple factor scoring systems. Br J Surg 1990;77:1260–1264. 16 Malfertheiner P, Dominguez-Munoz JE: Prognostic factors in acute pancreatitis. Int J Pancreatol 1993; 14:1–8. 17 Triantopoulou C, Lytras D, Maniatis P, Chrysovergis D, Manes K, Siafas I, Papailiou J, Dervenis C: Computed tomography versus Acute Physiology and Chronic Health Evaluation II score in predicting severity of acute pancreatitis: a prospective, comparative study with statistical evaluation. Pancreas 2007;35:238–242.
18 Garcea G, Gouda M, Hebbes C, Ong SL, Neal CP, Dennison AR, Berry DP: Predictors of severity and survival in acute pancreatitis: validation of the efficacy of early warning scores. Pancreas 2008;37:e54–e61. 19 Gravante G, Garcea G, Ong SL, Metcalfe MS, Berry DP, Lloyd DM, Dennison AR: Prediction of mortality in acute pancreatitis: a systematic review of the published evidence. Pancreatology 2009; 9:601–614. 20 Gibson PH, Croal BL, Cuthbertson BH, Small GR, Ifezulike AI, Gibson G, Jeffrey RR, Buchan KG, El-Shafei H, Hillis GS: Preoperative neutrophil-lymphocyte ratio and outcome from coronary artery bypass grafting. Am Heart J 2007;154:995–1002. 21 Gillum RF, Mussolino ME, Madans JH: Counts of neutrophils, lymphocytes, and monocytes, cause-specific mortality and coronary heart disease: the NHANES-I epidemiologic follow-up study. Ann Epidemiol 2005;15:266–271. 22 Halazun KJ, Hardy MA, Rana AA, Woodland DC 4th, Luyten EJ, Mahadev S, Witkowski P, Siegel AB, Brown RS Jr, Emond JC: Negative impact of neutrophil-lymphocyte ratio on outcome after liver transplantation for hepatocellular carcinoma. Ann Surg 2009;250:141–151. 23 Sarraf KM, Belcher E, Raevsky E, Nicholson AG, Goldstraw P, Lim E: Neutrophil/lymphocyte ratio and its association with survival after complete resection in non-small cell lung cancer. J Thorac Cardiovasc Surg 2009;137:425–428.
Pancreatology 2011;11:445–452
451
24 Tamhane UU, Aneja S, Montgomery D, Rogers EK, Eagle KA, Gurm HS: Association between admission neutrophil to lymphocyte ratio and outcomes in patients with acute coronary syndrome. Am J Cardiol 2008;102: 653–657. 25 Zahorec R: Ratio of neutrophil to lymphocyte counts – rapid and simple parameter of systemic inflammation and stress in critically ill. Bratisl Lek Listy 2001;102:5–14. 26 Goodman DA, Goodman CB, Monk JS: Use of the neutrophil:lymphocyte ratio in the diagnosis of appendicitis. Am Surg 1995; 61: 257–259. 27 Subbe CP, Kruger M, Rutherford P, Gemmel L: Validation of a modified early warning score in medical admissions. QJM 2001; 94: 521–526. 28 Garrey W: Variation in white blood cell counts. Physiol Rev 1935; 15:597–638.
452
29 Skubitz K: Neutrophilic leukocytes; in Greer J, Wintrobe MM (eds): Wintrobe’s Clinical Hematology, ed 12. Philadelphia, Wolters Kluwer/Lippincott Williams & Wilkins Health, 2009, pp 170–193. 30 Wittel UA, Rau B, Gansauge F, Gansauge S, Nussler AK, Beger HG, Poch B: Influence of PMN leukocyte-mediated pancreatic damage on the systemic immune response in severe acute pancreatitis in rats. Dig Dis Sci 2004;49:1348–1357. 31 Pavlov P, Uchikov P, Murdzheva M, Tuleva S, Tsvetkova T: Main lymphocyte populations and their subpopulations in patients with acute pancreatitis studied in the course of disease (in Bulgarian). Khirurgiia (Sofiia) 2001;57:4–11. 32 Pezzilli R, Billi P, Beltrandi E, Casadei Maldini M, Mancini R: Impaired lymphocyte proliferation in human acute pancreatitis. Digestion 1997;58:431–436. 33 Pezzilli R, Billi P, Beltrandi E, Maldini M, Mancini R, Morselli Labate AM, Miglioli M: Circulating lymphocyte subsets in human acute pancreatitis. Pancreas 1995;11:95–100.
Pancreatology 2011;11:445–452
34 Pezzilli R, Maldini M, Morselli-Labate AM, Barakat B, Romboli E, Beltrandi E, Migliori M, Tomassetti P, Corinaldesi R: Early activation of peripheral lymphocytes in human acute pancreatitis. J Clin Gastroenterol 2003; 36:360–363. 35 Takeyama Y, Takas K, Ueda T, Hori Y, Goshima M, Kuroda Y: Peripheral lymphocyte reduction in severe acute pancreatitis is caused by apoptotic cell death. J Gastrointest Surg 2000;4:379–387. 36 Bain BJ: Ethnic and sex differences in the total and differential white cell count and platelet count. J Clin Pathol 1996;49:664– 666. 37 Lichtman MA: Cellular deformability during maturation of the myeloblast. Possible role in marrow egress. N Engl J Med 1970; 283:943–948.
Azab /Jaglall /Atallah /Lamet /Raja-Surya / Farah /Lesser /Widmann