**3.1. "The predictive value of immature granulocyte count and immature myeloid information in the diagnosis of neonatal sepsis"- own experience and study results**

Our study group tried to determine the predictive value of the IG# and the immature myeloid information (IMI) in neonatal early onset sepsis performing a historical cohort study (Cimenti et al., 2012).

## *3.1.1. Patients and methods*

We collected 133 blood samples of neonates admitted to the NICU of the Pediatric Department of the Medical University Graz, a tertiary care center. Based on their admission diagnosis and their clinical course patients were divided in two groups. The first group consisted of patients with blood culture verified bacteremia, clinically strongly suspected sepsis, or elevated inflammatory parameters, a history of risk factors, and antibiotic treatment ≥ 7 d. Patients in the second group were asymptomatic, healthy children without any infectious risk factors constituting the control group. They were admitted to the NICU because of low birth weight, delayed postnatal transition or prematurity.

Blood sampling was routinely performed in all neonates and repeated depending on their clinical course. Blood samples were collected into microvette tubes (Sarstedt, Nümbrecht, Germany) and analyzed using the Sysmex XE-2100 (13). In cases of suspected bacterial infection, blood samples were always taken before the initiation of antibiotic therapy. ROC curves were used for comparison of infectious indices by plotting the test sensitivity (equivalent to the true positive rate) on the y-axis and 1-specifity (equivalent to the false positive rate) on the x-axis for all possible cut off values of the diagnostic test (see Figure 1).

68 Neonatal Bacterial Infection

**results** 

study (Cimenti et al., 2012).

*3.1.1. Patients and methods* 

value of 3.2% for IG% as optimal for a normal adult population. Using a cut-off in a range between 4% and 5% of total WBC would result in a too high rate of missed cases (Bernstein & Rucinski, 2011). In a large outpatient pediatric population comprising more than 2400 samples, age dependent upper limits for reference ranges for the automated enumeration of IG were defined as 0.30% and 40/µL for IG% and IG#, respectively for children aged below 10 years (Roehrl et al., 2011). Above the age of 10 years, an upper limit of 0.90% and 70.0/µL for relative and absolute IG count was recommended (Roehrl et al., 2011). In this study blood samples were analyzed using the Sysmex XT-1800i instrument (Sysmex, Kobe, Japan). The defined upper limits showed no differences dependent on the patient sex. As expected the cause of elevated IGC differed between both groups. While respiratory or gastrointestinal infections were common associations with elevated IGC in the group < 10 years, the older children showed hematologic malignancies, drug therapy (glucocorticoids, chemotherapy), severe infections, and pregnancy (young females). In a subgroup analysis of patients < 1 year this study revealed age-stratified nonparametric estimates of upper limits of normal (95th percentiles) and associated 90% confidence intervals (CI) for IG# and IG% of 40/µL (30.0–50.0) and 0.30% (0.20–0.40). In addition, this study described an important observation: Even the most abnormal IGCs in the younger age group were quite low compared with abnormal IGCs in the older individuals. This fact highlights the importance of particular reference values appropriate for different age groups. Otherwise especially younger children with associated disease and with only small elevations of IGCs could be overlooked (Roehrl et al., 2011). As neonates represent a highly particular and often vulnerable patient

population we aimed at investigate a possible correlation between IGC and sepsis.

**3.1. "The predictive value of immature granulocyte count and immature myeloid information in the diagnosis of neonatal sepsis"- own experience and study** 

Our study group tried to determine the predictive value of the IG# and the immature myeloid information (IMI) in neonatal early onset sepsis performing a historical cohort

We collected 133 blood samples of neonates admitted to the NICU of the Pediatric Department of the Medical University Graz, a tertiary care center. Based on their admission diagnosis and their clinical course patients were divided in two groups. The first group consisted of patients with blood culture verified bacteremia, clinically strongly suspected sepsis, or elevated inflammatory parameters, a history of risk factors, and antibiotic treatment ≥ 7 d. Patients in the second group were asymptomatic, healthy children without any infectious risk factors constituting the control group. They were admitted to the NICU

Blood sampling was routinely performed in all neonates and repeated depending on their clinical course. Blood samples were collected into microvette tubes (Sarstedt, Nümbrecht, Germany) and analyzed using the Sysmex XE-2100 (13). In cases of suspected bacterial

because of low birth weight, delayed postnatal transition or prematurity.

**Figure 1.** a and b: Diff- and IMI scattergram showing graphic output of WBC differential results performed with the Sysmex XE 2100. By courtesy of © Sysmex Europe GmbH, Norderstedt, Germany.

The Youden's index (sensitivity in %)/100 + (specificity in % - 1)/100 - 1 was used for determination of optimal cut-off values. The area under the curve (AUC) was calculated using the binormal approach by McClish. ROC curve are used to assess the diagnostic accuracy of a test. The ROC curve allows analyses of the trade-offs between sensitivity and specificity at all possible cut-off points and is often used to determine optimal cut-off values and to compare the usefulness of two more diagnostic tests. The area under the curve (AUC) is another useful tool describing the discriminative ability of a test across the full range of cut-offs. A test with an AUC greater than 0.9 has high accuracy, while 0.7–0.9 describes moderate accuracy, 0.5–0.7 implies low accuracy and 0.5 displays a chance result (Akobeng, 2007; Fischer et al., 2003).

The Role of Immature Granulocyte Count

and Immature Myeloid Information in the Diagnosis of Neonatal Sepsis 71

**Figure 2.** Boxplot diagram showing the distribution of IG#, IG%, IMI#, and IMI% values in neonates with sepsis compared to the control group. The top and the bottom of the box represent the 25th and 75th percentile; the line in the box indicates the median. The whiskers display the largest data less than or equal to the 75th percentile plus 1.5 times interquartile range and the lowest data greater than or

**Figure 3.** Receiver operating characteristic (ROC) curves of IMI# (thick line), IG# (thin line) and IMI%

(dotted line).

equal to the 25th percentile minus 1.5 times interquartile range (Cimenti et al., 2012).

Of 133 blood samples of patients admitted to our neonatal intensive care unit 21 neonates were suspected and treated for sepsis (mean gestational age 34.1 weeks, mean birth weight 2287 g, 9 male, 12 female, 12 patients had a history of premature rupture of membranes (PROM)). In the control group 112 healthy neonates were analyzed (mean gestational age 34.2 weeks, mean birth weight 2128 g, 59 male, 53 female, 31 patients with a history of PROM).

### *3.1.2. Results*

The number of IMI classified cells (IMI#) was significantly elevated in patients with sepsis compared to the control group (639/µL (144; 2029) vs. 89/µL (40; 133), p=0.000065). The number of IMI/ total leucocyte count (IMI%) in patients with sepsis was significantly elevated compared to the control group (4.5 (1.3; 9.5) vs. 0.7 (0.5; 1.1), values expressed in %, p=0.000076). IG# was significantly elevated in neonates with sepsis compared to the control group (0.28x 10³/µL (0.03; 0.56) vs. 0.05x10³/µL (0.05-0.09), p=0.049). The percentage of IG% was significantly elevated in septic neonates vs. infants in the control group (1.3 (0.5; 4.5) vs. 0.5 (0.4; 0.7), values expressed in %, p=0.022) (Cimenti et al., 2012). The AUC for the IMI# was 0.76 and 0.70 for IG% and IMI%, respectively. The positive and negative predictive value, sensitivity, specificity, and the Youden's index at different cut off values are listed in Table 2 (Cimenti et al., 2012).


**Table 2.** Positive predictive value, negative predictive value, sensitivity, and specificity of IG#, IG%, IMI#, IMI% and IT ratio for optimal cut off values determined by ROC analysis using the Youden's index in 21 neonates with sepsis compared to 112 neonates with negative infectious status (Cimenti et al., 2012).

The Role of Immature Granulocyte Count

and Immature Myeloid Information in the Diagnosis of Neonatal Sepsis 71

70 Neonatal Bacterial Infection

2007; Fischer et al., 2003).

PROM).

al., 2012).

*3.1.2. Results* 

Table 2 (Cimenti et al., 2012).

The Youden's index (sensitivity in %)/100 + (specificity in % - 1)/100 - 1 was used for determination of optimal cut-off values. The area under the curve (AUC) was calculated using the binormal approach by McClish. ROC curve are used to assess the diagnostic accuracy of a test. The ROC curve allows analyses of the trade-offs between sensitivity and specificity at all possible cut-off points and is often used to determine optimal cut-off values and to compare the usefulness of two more diagnostic tests. The area under the curve (AUC) is another useful tool describing the discriminative ability of a test across the full range of cut-offs. A test with an AUC greater than 0.9 has high accuracy, while 0.7–0.9 describes moderate accuracy, 0.5–0.7 implies low accuracy and 0.5 displays a chance result (Akobeng,

Of 133 blood samples of patients admitted to our neonatal intensive care unit 21 neonates were suspected and treated for sepsis (mean gestational age 34.1 weeks, mean birth weight 2287 g, 9 male, 12 female, 12 patients had a history of premature rupture of membranes (PROM)). In the control group 112 healthy neonates were analyzed (mean gestational age 34.2 weeks, mean birth weight 2128 g, 59 male, 53 female, 31 patients with a history of

The number of IMI classified cells (IMI#) was significantly elevated in patients with sepsis compared to the control group (639/µL (144; 2029) vs. 89/µL (40; 133), p=0.000065). The number of IMI/ total leucocyte count (IMI%) in patients with sepsis was significantly elevated compared to the control group (4.5 (1.3; 9.5) vs. 0.7 (0.5; 1.1), values expressed in %, p=0.000076). IG# was significantly elevated in neonates with sepsis compared to the control group (0.28x 10³/µL (0.03; 0.56) vs. 0.05x10³/µL (0.05-0.09), p=0.049). The percentage of IG% was significantly elevated in septic neonates vs. infants in the control group (1.3 (0.5; 4.5) vs. 0.5 (0.4; 0.7), values expressed in %, p=0.022) (Cimenti et al., 2012). The AUC for the IMI# was 0.76 and 0.70 for IG% and IMI%, respectively. The positive and negative predictive value, sensitivity, specificity, and the Youden's index at different cut off values are listed in

Parameter Cut-off PPV NPV sensitivity specificity Youden'sindex IG# 0.24 0.60 0.31 0.60 0.88 0.48 IG% 1.3 0.67 0.27 0.67 0.88 0.55 IMI# 262 0.80 0.09 0.80 0.72 0.52 IMI% 0.02 0.70 0.14 0.70 0.76 0.46 IT ratio 0.06 0.75 0.20 0.75 0.94 0.69 **Table 2.** Positive predictive value, negative predictive value, sensitivity, and specificity of IG#, IG%, IMI#, IMI% and IT ratio for optimal cut off values determined by ROC analysis using the Youden's index in 21 neonates with sepsis compared to 112 neonates with negative infectious status (Cimenti et

**Figure 2.** Boxplot diagram showing the distribution of IG#, IG%, IMI#, and IMI% values in neonates with sepsis compared to the control group. The top and the bottom of the box represent the 25th and 75th percentile; the line in the box indicates the median. The whiskers display the largest data less than or equal to the 75th percentile plus 1.5 times interquartile range and the lowest data greater than or equal to the 25th percentile minus 1.5 times interquartile range (Cimenti et al., 2012).

**Figure 3.** Receiver operating characteristic (ROC) curves of IMI# (thick line), IG# (thin line) and IMI% (dotted line).

## *3.1.3. Reference values for a neonatal study population based on ROC analysis*

Preliminary data revealed a cut-off value for IG% of 1.3% based on calculations using the Youden's index and a median of 0.05x103/µL (0.05-0.09; 2.5%–97.5% confidence interval) for IG# in our control group of asymptomatic, healthy subjects. According to our data the measurement of IMI# compared to IG# seems to be more favourable as determined by ROC analysis as there seems to be a tendency that the IM# has a higher predictive value than the IG# (Figure 3). Setting a cut off value of 262/µl the measurement of IMI# revealed a positive predictive value of 0.80.

The Role of Immature Granulocyte Count

and Immature Myeloid Information in the Diagnosis of Neonatal Sepsis 73

one third of cases no cause had been detected. Except in proven early onset bacterial infection, the presence, severity and duration of low counts showed no relationship with mortality rate, whereas neutropenia within the first 3 days of life showed an association with necrotizing enterocolitis (NEC) or nosocomial infection (Christensen et al., 2006).Using IGC in a clinical context should incorporate these factors as well as the likelihood of

As for the blood cell count it has been shown that the performance characteristics in distinguishing between infants with and without infections improve significantly during the first 4 hours after birth. The AUC of the WBC, ANC and IT-ratio showed an increase from 0 hours of 0.52, 0.55 and 0.73, respectively to 0.87, 0.85 and 0.82, respectively 4 hours after birth (Newman et al., 2010). In a review article on new technologies for a diagnostic approach in neonatal sepsis Srinivasan and Harris considered the future development of computerized algorithms including these variables as possibly useful to estimate the probability of sepsis (Srinivasan & Harris, 2012). Taking this fact into account it might be advisable to perform serial IGCs, especially in cases where sepsis had to be ruled out and the result of the test will have a therapeutic consequence (i.e. discontinuing of antibiotic

The clinical applicability of automated IG counting might be limited by the relatively poor sensitivity of this method (Ansari-Lari et al., 2003; Nigro et al., 2005). Considering the satisfactory specificity and the high NPV this parameter could represent a valuable additional aid in combination with other laboratory markers or diagnostic algorithms. However, as higher values of IG% of more than 3% have been shown to predict blood culture positive results (Ansari-Lari et al., 2003), this subgroup of patients should probably

As clinical signs in preterm and term infants with severe bacterial infection are often nonspecific and scarce, automated detection of IGs and IMI seems to act as a useful adjunctive tool in the diagnosis of neonatal sepsis. Technical development of automated hematology analyzers has led to a precise, fast, accurate, and reproducible determination of IGs. The detection of all immature cells including blasts in a separate channel might be advantageous at an early stage of sepsis, when these cells are released from bone marrow and the

Although automated determination of IGs is currently carried out in the area of research, evidence exists that this method seems to be worth to be implemented in clinical practice especially as an adjunctive tool in determining early phase of bacterial sepsis. The fact that measurement of this parameter in the course of routine determination of a white blood differential count does not necessitate any additional sample volume, personal effort, or costs, might be a valuable additional argument. Further well designed prospective trials are mandatory to validate the performance characteristics of these new parameters as diagnostic tool in neonatal sepsis. In this context, the availability of an internal quality control and the

be revaluated for potential infection even in the absence of specific symptoms.

peripheral neutrophil count can still be in the normal range.

infection in every individual patient.

treatment).

**4. Conclusion** 
