Utility of neutrophil CD64 in distinguishing bacterial infection from inflammation in severe alcoholic hepatitis fulfilling SIRS criteria

To assess utility of neutrophilCD64 (nCD64) expression in differentiating bacterial infection from inflammation in patients with severe alcoholic hepatitis (SAH) fulfilling systemic inflammatory response syndrome criteria. Patients with SAH and infection (n = 58), SAH without infection (n = 70), and healthy controls (n = 20) were included. Neutrophil CD64 expression by flowcytometry, serum Procalcitonin (ELISA) and C-reactive protein (Nephelometry) and neutrophil–lymphocyte ratio (NLR) were studied. Percentage of neutrophils with CD64 expression (nCD64%) was significantly higher in patients with SAH and infection than in those without infection and controls [76.2% (56.9–86.5) vs. 16% (12.6–23.1) vs. 7.05% (1.4–9.5), p < 0.05], as was their mean fluorescence intensity [MFI; 1431 (229–1828) vs. 853 (20–968) vs. 99.5 (54.7–140.7), p < 0.05]. Using a cut-off of 27%, the sensitivity and specificity of nCD64% to diagnose bacterial infection was 94% and 81%, respectively, with area under curve (AUC) of 0.95. At a cut-off value of 0.261 ng/ml, the sensitivity and specificity of serum procalcitonin was 83% and 72%, respectively, with AUC of 0.86. Serum CRP, total leukocyte count, NLR had AUCs of 0.78, 0.63 and 0.64, respectively. Quantitative measurement of nCD64 can better distinguish systemic bacterial infection and inflammation in SAH as compared to traditional biomarkers.


Results
Sixteen of 144 patients enrolled were excluded as they were HBsAg positive (n = 11), anti-HCV positive (n = 3) or had a space occupying lesion in the liver (n = 2). Finally, 128 patients (126 males and 2 females), and 20 healthy controls were included (Fig. 1).
Mean age of the patients was 43.8 ± 8.5 (range 27-70 years) and the mean mDF score was 105.4 ± 54.2. None of the patients were receiving glucocorticoids at the time of enrolment in the study. Infection group included 58 patients with proven or probable infection while non-infection group included 70 patients with inflammation who did not have infection. When baseline characteristics of patients in the two groups were compared (Table 1), patients in Infection group had significantly higher international normalized ratio (p < 0.001) and MELD score (p < 0.001) compared with patients in non-infection group. Fever at presentation was observed in 73% of patients in Infection group and 42% in non-infection group (p < 0.0005). Twenty-four patients in Infection group had culture proven infection (18 blood Fig. 3). There was no correlation between duration of fever and nCD64%. The positive and negative predictive values of nCD64% were 84% and 94%, respectively which were higher than any other biomarkers; serum procalcitonin, NLR (Neutrophil-Lym-  (23) UTI=7 (20.5%) (>15 pus cells in urine, symptomatic patient) (24) Lung=17 (50%) (CXR new infiltrates, Consolidation) (24) Dual Infection=12 (34%)  Validation cohort. We performed nCD64% in 38 patients with acute on chronic liver failure (ACLF) of similar severity with (n = 27) and without (n = 11) infections. The mean nCD64% was 86.5% in infection vs 16.6% in non-infection patients. However, this was a mixed cohort of ACLF patients.

Discussion
It is challenging to differentiate acute inflammation from systemic bacterial infection, particularly in patients requiring immunosuppressive medications for treatment. Bacterial infections are the most feared because they can evolve very rapidly and can be life-threatening within hours to a few days. We observed higher nCD64% in patients of SAH with infection compared to SAH without infection. It had better area under the curve than serum   22 . In an earlier study of systemic inflammatory diseases [systemic lupus erythematosus (SLE) and anti-neutrophil cytoplasmic antibody associated vasculitis], Ajmani et al. reported a sensitivity and specificity of 85% and 84%, respectively at a nCD64% cut-off of 30%, to differentiate between infection and inflammation 13 . In another study in active autoimmune inflammatory conditions and vasculitis, a high sensitivity (85%) and specificity (91%) for distinguishing between systemic infection and active inflammation was reported 23 . Similarly, in another study conducted on patients with SLE and rheumatoid arthritis,  24 . Results of the present study are in accord with these data. Severe AH is an intensely inflammatory condition on one hand and is complicated by infections in almost one-half of patients on the other. This creates a major dilemma for the clinician about the timing of offering glucocorticoid therapy, before or after antibiotic therapy. Early introduction of antibiotics at the time of admission even in absence of infection has not been reported to reduce mortality or incidence of subsequent infections in AH 21 . Both SIRS and infections are major causes of mortality and require diverging therapeutic strategies. High dose glucocorticoids may reduce short term mortality in SIRS but may predispose towards or worsen existing infection. In such a scenario, percentage nCD64 expression may provide guidance in choosing the right treatment. Additionally, nCD64 expression is not affected by steroid therapy, making it even more suitable for further evaluation for early detection of bacterial infection in patients with SAH receiving glucocorticoids 25 .
Recently, NLR has been reported to be a better diagnostic marker to detect bacterial infection in hospitalized patients with fever 26 , additionally day-4 NLR has prognostic value for 90-day mortality in patients with SAH 27,28 . STOPAH trial reported NLR data for 789 patients. NLR was associated with acute kidney injury and infection. In a cohort of patients with SAH, those with baseline NLR score between 5 and 8 were more likely to respond to glucocorticoids 28 .
For a long time, serum procalcitonin has been regarded as the best diagnostic test available for systemic infection. However, serum procalcitonin has been reported to be elevated in patients with severe liver dysfunction without infection, limiting its diagnostic utility for infection in patients with SAH 29 . In a recent meta-analysis, the sensitivity and specificity of procalcitonin (88% and 81%) was reported to be higher than that of CRP (75% and 67%) 30 . Our study demonstrated that nCD64% was more sensitive and specific than serum procalcitonin, CRP, NLR, ESR or TLC in the detection of bacterial infections in SAH patients. The AUC in our study for nCD64% at 0.95 was consistent with other studies and meta-analysis 19 on the same marker for bacterial infection in other disease conditions.
Ours being a tertiary care referral center receive patients in advanced stage of the disease with high mDF and MELD scores and are generally initiated on antibiotics at peripheral centers so the number of culture positive patients was low. However, all the patients categorized in our study as probable infection had infections based on a composite of clinician's judgement and laboratory investigations despite negative microbial cultures.
There are few limitations of our study; small sample size of the validation cohort and it being a mixed cohort of ACLF patients rather than SAH patients. The mean mDF score of 105 was extremely high in our study therefore studies with larger number of patients and mDF > 32 but < 100 and MELD > 20 should be done to validate it in future studies. Other limitation is the lacking assessment of other biomarkers; interferon-γ, lipopolysaccharides and other cytokines in plasma of these patients. Majority of patients in the present study were males, probably due to the socio-cultural milieu of our population. Due to cross-sectional design of the study, we did not assess the correlation between the nCD64% and the outcome. Despite limitations, taking into consideration the fact that nCD64% has performed almost equally well in other systemic inflammatory diseases like ANCA associated vasculitis and SLE and the data obtained from the current study, nCD64% ability to distinguish bacterial infection from SIRS makes a strong case for its utility in SAH patients as well.
In conclusion, we propose that nCD64% expression may be a useful tool to differentiate bacterial infection from SIRS in patients with SAH. It is a simple test with a short turn-around time (2-4 h) and has a cost comparable with CRP and pro-calcitonin (about 8 USD per assay in our laboratory). In a setting of SAH where 30-day www.nature.com/scientificreports/ mortality is very high and definite evidence of infection in form of blood and other body fluid cultures may take 3-5 days, a short turn-around time marker such as nCD64% may facilitate timely initiation of appropriate treatment and is likely to impact prognosis. These characteristics make it an attractive test to incorporate into routine clinical practice for the management of patients with SAH.

Material and methods
This prospective, cross sectional, observational descriptive study was conducted in the Department of Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, India from March 2017 to June 2018. Alcoholic hepatitis was diagnosed as per National Institute on Alcohol Abuse and Alcoholism (NIAAA), Alcoholic Hepatitis Consortia criteria 31 after exclusion of other liver diseases in a patient with a long history of heavy alcohol use (typically > 100 g per day for more than 5 years) with recent and rapid development or worsening of jaundice, liver related complications and serum bilirubin of > 3 mg/dl, along with elevation in aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels to more than 1.5 times the upper limit of normal but less than 400 U/l, with AST/ALT ratio of > 1.5 and with persistent alcohol use until 8 weeks before onset of symptoms 31 . Severe AH was defined as a mDF score of 32 or more 32 . The definition of SIRS used was presence of ≥ 2 of the following: (1) temperature > 38 °C or < 36 °C (2) Respiratory Rate > 20/min (3) heart rate > 90/min (4) White blood cell count > 12,000/µl/ < 4000/ or > 10% bands 33 . Liver biopsy was carried out when there was a doubt in diagnosis due to overlapping features in patients with decompensated cirrhosis with ongoing or recent alcohol abuse and severe alcoholic hepatitis.
Sepsis by definition is systemic inflammatory response (SIRS) due to infection-microbial etiology (proven or suspected) 33 . In our study, SIRS and any one of the following mentioned below was used to diagnose Sepsis: (1) positive blood/urine/ascitic fluid/sputum culture, (2) spontaneous bacterial peritonitis (SBP; as defined by infection of the ascitic fluid, as evidenced by an ascitic fluid absolute polymorphonuclear leukocyte (PMN) count of at least 250 cells/µl (0.25 × 10 9 /L), with or without a positive ascitic fluid culture, in the absence of an intraabdominal surgically treatable source of infection 20 (3) lower respiratory tract infections with new pulmonary infiltrates in the presence of respiratory symptom/s (cough, sputum production, dyspnea, pleuritic pain) with rales or crepitation on auscultation or one sign of infection (core body temperature > 38 °C or < 36 °C, shivering or leukocyte count > 10,000/mm 3 or < 4000/mm 3 ) in the absence of antibiotics or new infiltrates or consolidation/lung abscess demonstrable on radiograph/ computed tomography of chest, (4) evidence of extra hepatic biliary obstruction or hepatic abscess on abdomen ultrasound (5) skin infections; fever with cellulitis, abscess, discharging pus from the skin lesion (6) urinary tract infection (UTI) being diagnosed as positive urine dipstick result for leukocytes or nitrites, or urine WBC > 15/high-power field with either positive urine gram stain or culture in a symptomatic patient 21 . Patient. All consecutive patients suffering from SAH were evaluated at Gastroenterology department, and those with SIRS or SIRS with infection were prospectively recruited in the study after obtaining informed consent. Study was approved by Institutional ethics committee, Sanjay Gandhi Postgraduate Institute of Medical Sciences (A-18-PGI-IMP-75-2017) and was performed in accordance with the Declaration of Helsinki. All laboratory experiments were performed in accordance with relevant guidelines and regulations.
Inclusion criteria. Patients aged 18 years or more, suffering from SAH according to clinical, biochemical, imaging and liver tissue pathology criteria with mDF score of more than 32, model for end stage liver disease (MELD) score > 20 and suspected to have either SIRS or SIRS with infection were included in this study.
Exclusion criteria. Patients with absolute neutrophil count less than 1500 per microliter (μl), those who had received granulocyte-colony stimulating factor (G-CSF) therapy, who were suspected to have mycobacterial or fungal infection who were on immunosuppressants (other than glucocorticoids) or chemotherapeutic agents, who had been found to have hepatocellular carcinoma or other malignancy, who had sustained trauma or undergone surgery, or had suffered from some vital disease other than liver disease within 3 months of study entry were excluded 6 .
Patients were divided into two groups, namely those with SAH and SIRS with infection (Infection Group, 58 patients) and those with SAH with SIRS (Non-infection Group, 70 patients). Infection was considered proven when a pathogen was detected by microbial culture techniques. Infection was considered probable in patients with a composite of suspicion of infection in the form of fever and raised TLC (> 10,000/mm 3 ) or Leukopenia (< 4000/mm 3 ) plus any one of the following: Spontaneous bacterial peritonitis, Radiological findings suggesting new infiltrates, consolidation or lung abscess on chest radiograph or computerized tomography (CT), Pus discharge from skin lesions, Cellulitis which were culture negative due to prior antibiotic therapy initiated prior to referral to our center. Infection was considered unlikely if workup to rule out infection, including chest radiograph, urine examination, ascitic fluid evaluation, and appropriate body fluid cultures turned out to be negative. Patients with proven and probable infection were analyzed together and compared with those with inflammation but no infection. Pending cultures all patients fulfilling the criteria of SAH and SIRS were initiated on antibiotics.
Six milliliters of venous blood (3 ml each in a plain vial and in ethylene diamine tetra acetic acid (EDTA) was collected from each patient and healthy controls. Laboratory personnel carrying out the tests were unaware to the patients' clinical disease status. Neutrophil CD64 expression. nCD64 was analyzed via flow-cytometry as previously described 13  www.nature.com/scientificreports/ human anti-CD64 antibody (BD Pharmingen CA, USA) and 20 µl PE-conjugated matched-isotype control antibody, followed by lysis of RBCs and two steps of washing with phosphate buffer saline. Results are expressed as the percentage of neutrophils expressing CD64 (nCD64%) and as mean fluorescence intensity (MFI).
Serum procalcitonin. Blood samples were allowed to clot for 45 min, then centrifuged at 1500 rpm for 10 min and serum was aliquoted and stored at − 80 °C until analysis. Procalcitonin levels were measured in duplicate after a single freeze-thaw cycle in batched assays by sandwich enzyme-linked immunosorbent assay (ELISA), R&D systems (catalog No DY8350-05), as per manufacturer's protocol.
Lab parameters. C-reactive protein (CRP) in the serum was measured by nephelometry. ESR was measured by Westergren method.
Statistical analysis. Normality of the continuous variables was assessed and considered normally distributed when Z score of the skewness was ± 3.29. Descriptive statistics of the continuous variables were presented as mean ± standard deviation (S.D.) or median (interquartile range) whereas categorical data were presented in frequency (%). Independent samples t test was used to compare the means between two groups whereas Mann Whitney U test used to compare the non-normal continuous variables between two groups. Receiver operating characteristics (ROC) curve was used to estimate the diagnostic accuracy (area under curve "AUC", sensitivity, specificity) of the inflammatory variables with corresponding significance levels. AUROC cut-offs were chosen by taking the optimum balance between the sensitivity and specificity. The value was provided as given in the output of the software used for statistical analysis (SPSSv23). To compute the sensitivity and specificity of the two combined variables, either variable (based on identified cutoff value of nCD64% of 26.4 and Pro-calcitonin of 0.261) present was considered positive (infection present) and it was compared with clinical evidence of infection (both probable and confirmed infection) as the gold standard. A p value < 0.05 was considered statistically significant. Statistical package for social sciences, version-23 (SPSS-23, IBM, Chicago, USA) and MedCalc statistical software was used for data analysis.