British Journal of Haematology, 2000, 111, 1093±1102

Procalcitonin in paediatric cancer patients: its diagnostic relevance is superior to that of C-reactive protein, interleukin 6, interleukin 8, soluble interleukin 2 receptor and soluble tumour necrosis factor receptor II Gudrun Fleischhack, Isabell Kambeck, Dubravka Cipic, Carola Hasan and Udo Bode Department of Paediatric Haematology/Oncology, University of Bonn, Germany Received 3 June 2000; accepted for publication 28 July 2000

Summary. Sensitive parameters of inflammation are rare in neutropenic cancer patients. In this study, procalcitonin (PCT), C-reactive protein (CRP), interleukin 6 (IL-6), IL-8, the soluble IL-2 receptor (sIL-2R) and the soluble tumour necrosis factor receptor II (sTNFRII) were evaluated for their diagnostic relevance in febrile episodes of cancer patients. Plasma or serum levels of these parameters were determined in neutropenic children with febrile episodes (n ˆ 122) classified according to both the kind of infection [60 cases of fever of unknown origin (FUO), 28 cases of localized infection, 13 cases of pneumonia, 20 cases of bacteraemia, one case of fungaemia] and the World Health Organization (WHO) score of chemotherapy-induced mucositis. At baseline and during the febrile episodes, the highest levels of all

parameters were observed in cases of gram-negative bacteraemia. However, in FUO and localized infections, low or only slightly elevated median levels of all parameters were documented. The degree of chemotherapy-induced mucositis did not influence the value of any parameter. In comparison with the other inflammatory parameters, PCT (optimum cut-off level 0´5 mg/l) was a more sensitive and more specific parameter in the diagnosis of high-risk (gramnegative bacteraemia) and low-risk (FUO) episodes, as well as in the sequential assessment of all febrile neutropenic episodes.

Infections are the major cause of treatment-related morbidity and mortality in cancer patients undergoing intensive chemotherapy. The early diagnosis of severe infections and the prompt initiation of adequate anti-microbial therapy are essential for the successful treatment of febrile episodes. In addition, the sequential assessment of febrile episodes, the early diagnosis of patients at high risk of deterioration or secondary infectious complications and the recognition of patients at low risk of complications are essential factors in decisions regarding risk-adapted anti-microbial therapy and the need for hospitalization. The acute-phase reactant C-reactive protein (CRP) is the best known biochemical inflammatory marker in cancer patients. However, its use in this setting has disadvantages: it does not increase significantly until 24±48 h after onset

of inflammation and its serum concentration correlates with the grade of tissue damage and the activity of the underlying malignancy (Heney et al, 1992; Katz et al, 1992; Jaye & Waites, 1997; Engel et al, 1998; Lehrnbecher et al, 1999). In response to inflammatory stimuli, a number of macrophage-/monocyte-derived cytokines mediate a number of metabolic changes that are known as acutephase reactions. Two of the most important mediators are interleukin (IL)-6 and IL-8 (Bauer, 1989; Baggiolini & Clark Lewis, 1992; Lotz, 1995; Suffredini et al, 1999). Both cytokines and the soluble IL-2 receptors (sIL-2R), the soluble tumour necrosis factor receptor (sTNFR)I (sTNFRp55) and sTNFRII (sTNFR-p75) have been reported to be increased in viral and/or bacterial infections. In recent studies, IL-6, IL-8, TNF-a and the sTNFRs were identified as early and sensitive markers of severe infections in both non-neutropenic patients and immunosuppressed cancer patients (Heney et al, 1992; Waage et al, 1994; Steinshamn et al, 1995; De Werra et al, 1997; Engel et al, 1998). The first report on the most important precursor protein

Correspondence: Gudrun Fleischhack, Department of Paediatric Haematology/Oncology, Children's Medical Hospital, University of Bonn, Adenauerallee 119, D-53113 Bonn, Germany. E-mail: [email protected] q 2000 Blackwell Science Ltd

Keywords: procalcitonin, CRP, febrile neutropenia, inflammatory cytokines, inflammatory soluble cytokine receptors.

1093

1094 G. Fleischhack et al of calcitonin, procalcitonin (PCT), as an inflammatory parameter in bacterial infections was published by Assicot et al (1993). To date, various authors have confirmed that PCT is a specific and sensitive marker of severe bacterial infections in non-immunocompromised patients (De Werra et al, 1997; Karzai et al, 1997; Chiesa et al, 1998; Oberhoffer et al, 1999). In these patients, PCT serum and plasma levels were helpful in both the early diagnosis and the assessment of severe infections (Al-Nawas et al, 1996; Karzai et al, 1997). With respect to the diagnostic significance of PCT in immunosuppressed patients, only few data are available on patients after transplantation and on neutropenic cancer patients undergoing conventional intensive chemotherapy (Al-Nawas & Shah, 1996; Staehler et al, 1997; Engel et al, 1999; Ruokonen et al, 1999; Fleischhack et al, 2000; Sauer et al, 2000). To date, no serial measurements comparing PCT with inflammatory cytokines in neutropenic children have been published. This study was initiated in order to evaluate the diagnostic relevance of PCT in febrile neutropenia of paediatric cancer patients in comparison with (i) the conventional inflammatory parameter CRP, (ii) the inflammatory cytokines IL-6 and IL-8, and (iii) the soluble cytokine receptors sIL-2R and sTNFRII. In particular, we investigated whether all parameters could reliably predict severe systemic infections (high-risk episodes), i.e. gram-negative bacteraemia, and mild infections (low-risk episodes), i.e. fever of unknown origin (FUO), at the time of admission, and whether they are useful in assessing all febrile courses and in monitoring the response to anti-microbial therapy. A group of afebrile patients undergoing intensive chemotherapy and another group of newly diagnosed cancer patients served as control groups for evaluation of the influence of the cytostatic therapy and of the underlying malignancy. PATIENTS AND METHODS Patients. In a retrospective study, inflammatory parameters were analysed in 122 febrile episodes (defined as fever $ 38´58C) of 51 paediatric cancer patients undergoing intensive chemotherapy at the Department of Paediatric Haematology/Oncology of the University of Bonn, Germany. All patients were treated intravenously with broadspectrum antibiotics according to a sequential anti-microbial regimen for at least 24 h after defervescence, up to a maximum of 72 h. In cases of bacteraemia, antibiotics were given for a minimum of 7 d. Two control groups comprising 23 newly diagnosed cancer patients and 12 afebrile immunosuppressed patients undergoing intensive chemotherapy were investigated. The age, gender and underlying diseases in the different groups are listed in Table I. Depending on the kind of infection, febrile episodes were classified as FUO, localized microbiologically or clinically documented infection (upper airway, urinary tract, gastrointestinal tract infection), pneumonia, gram-positive or gram-negative bacteraemia and fungaemia. In addition, all episodes were classified according to the World Health Organization (WHO) scaling score of chemotherapy-induced

mucositis (Table II) (World Health Organization, 1979). The diagnosis of microbiologically and clinically documented infections was based on positive cultures of urine, faeces, throat swabs or bronchoalveolar lavage and/or fever accompanied by clinical symptoms of urinary, gastrointestinal or respiratory tract infection. Bacteraemia or fungaemia was defined as fever with positive blood cultures for bacteria or fungi (peripheral blood or from the central venous indwelling catheter), with or without septic symptoms and with no other clinical signs of infection. Common skin contaminants and mucosal colonization with no clinical symptoms were not considered in the classification of febrile episodes. Methods. During the febrile episodes, blood samples were obtained at admission (before starting intravenous antibiotics) and thereafter every 2 d until defervescence and completion of antibiotic treatment. In both control groups, blood samples were taken during routine blood checks in afebrile patients and at admission of newly diagnosed cancer patients before they started intensive chemotherapy. Plasma samples from EDTA blood obtained by venepuncture or from an indwelling central venous catheter were frozen immediately and stored at 2208C pending determination of PCT, IL-6, IL-8, sIL-2R and sTNFRII. PCT was determined by an immunoluminometric assay (LUMItestPCT, B´R´A´H´M´S-Diagnostica GmbH, Berlin, Germany), and IL-6, IL-8, sIL-2R and sTNFRII were determined with commercially available enzyme-linked immunosorbent assay (ELISA) kits (Laboserv GmbH, Staufenberg, Germany). The detection limits for PCT, IL-6, IL-8, sIL-2R and sTNFRII were 0´1 mg/l, 2´0 pg/ml, 10´0 pg/ml, 16´0 pg/ml and 0´05 ng/ml respectively. CRP was determined in serum samples immediately after specimen collection and was measured by an immunonephelometric method (BNA analyser, Behring-Werke AG, Marburg, Germany). The detection limit for CRP was 3´0 mg/l. At the time of sampling for the inflammatory parameters the leucocyte count with a differential count was also recorded. Statistical analysis. The data of this study were evaluated using descriptive statistical methods [mean ^ SD (standard deviation), median, ranges, frequencies and percentages]. The Mann±Whitney U-test was used for comparison of independent variables, the Wilcoxon test for comparison of dependent variables and Pearson's chi-square test for comparison of percentages. The correlation analysis was carried out with Spearman's correlation coefficient as a two-tailed analysis. In all statistical tests, values of P , 0´05 at a , 0´05 were considered significant. The diagnostic relevance was estimated as sensitivity (true positives, patients with gram-negative bacteraemia or FUO respectively) and as specificity (true negatives, all patients with febrile episodes exclusive of patients with gramnegative bacteraemia or FUO respectively). Statistical calculations were done with the SPSS software package (Statistical Program for Social Science, version 9.0.1, Chicago, USA). Ethics. The study was conducted in accordance with the

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

Procalcitonin and Inflammatory Cytokines in Cancer Patients 1095 Table I. Patients' characteristics in the different diagnostic groups. Diagnostic group

Febrile patients

Afebrile patients

Newly diagnosed cancer patients

Number Age (median/range; years) Gender (female/male) Malignancies Leukaemia/lymphoma Solid tumours Peripheral blood cell counts at admission (median/range) WBC (109/l) ANC (109/l) ALC (109/l) AMC (109/l)

51 9´0/0´7±31´8 20/31 febrile episodes 122 68 54

12 8´4/1´2±25´2 4/8

23 7´8/3´2±28´8 9/14

0´80 0´04 0´10 0´00

0´1±88´7 0´0±18´2 0´0±6´9 0´0±2´0

6 6 2´20 1´00 1´00 0´11

16 7 1´0±6´7 0´1±2´0 0´5±5´9 0´0±0´7

8´20 2´71 2´25 0´22

1´4±98´5 0´3±18´5 0´2±17´4 0´0±2´7

WBC, white blood cell count; ANC, absolute neutrophil count; ALC, absolute lymphocyte count; AMC, absolute monocyte count.

updated declaration of Helsinki. Prior to enrolment in the study, the patients' parents and/or the patients themselves were informed of the retrospective and investigational character of the study and gave their informed consent. RESULTS Comparison of different patient groups In comparison to the newly diagnosed cancer patients, the majority of both febrile patients and afebrile patients showed marked leucopenia and neutropenia (P # 0´002) (Table I). The degree of cytopenia, i.e. neutropenia, lymphopenia and monocytopenia, respectively, was more severe in febrile patients than in afebrile immunosuppressed patients (P # 0´001). Ninety-one percent of the febrile patients were neutropenic [absolute neutrophil count (ANC) # 1´0  109/l] at admission or during the febrile course. In these patients, the median duration of neutropenia following admission was 7 d (range, 1±26 d). The peak values of the different inflammatory markers measured in patients with febrile episodes compared with the median levels of the two control groups are shown in Fig 1 (median, maximum range). Significant differences were documented between all values of the afebrile patients and the peak values of the febrile patients (PCT, P ˆ 0´004; CRP, P ˆ 0´000; IL-6, P ˆ 0´009; IL-8, P ˆ 0´000; sIL-2R, P ˆ 0´000; sTNFRII, P ˆ 0´001). The peak levels of PCT, CRP, IL-6 and IL-8 in febrile patients were also significantly higher than the levels measured in newly diagnosed cancer patients (P # 0´001). Markedly elevated levels of sIL-2R and sTNFRII were observed in the newly diagnosed cancer patients. These were comparable with the median peak levels in febrile episodes and did not correlate with body temperature or with the kind of underlying disease. With the exception of sIL-2R (P ˆ 0´045), the differences between the two control groups were not significant. Comparison of different kinds of febrile episodes Figure 2 (median, maximum range) and Fig 3A±F (median) show the levels of the inflammatory parameters depending on the different kinds of infection at admission and during

the first 5 d of the febrile episode respectively. Owing to the small number of samples, the time-points later than day 5 episodes and fungal infections were excluded from the analysis. Both at admission and during the course of the episodes, the highest median levels of all parameters were observed in cases of gram-negative bacteraemia (except for IL-8 at admission). Second in line were the levels recorded in episodes caused by gram-positive bacteraemia or pneumonia. Patients with FUO or localized infections showed low or only slightly elevated median levels of all parameters throughout the observation period. PCT levels discriminated best between more severe infections (pneumonia, gram-positive and gram-negative bacteraemia) and mild infections (FUO and localized infections), both at admission and in the assessment of the febrile course. In more severe infections, the median PCT levels at admission and during the episodes were significantly elevated compared with mild infections (P # 0´05). Although the median peak level in cases of gram-negative bacteraemia was higher than in cases of gram-positive bacteraemia, it did not differ significantly between these subgroups, possibly because of the small number of episodes. Median CRP levels elevated above 10´0 mg/l were Table II. Classification of febrile episodes (n ˆ 122) according to the kind of infections and the degree of chemotherapy-induced mucositis Kind of infection Fever of unknown origin (FUO) Localized Urogenital Gastrointestinal Upper airways Pneumonia Bacteraemia Gram-positive Gram-negative Fungaemia

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

Mucositis 60 28 11 8 9 13 20 7 13 1

WHO score 0 WHO score I/II WHO score III/IV

57 22 43

1096 G. Fleischhack et al

Fig 1. Concentrations of PCT (mg/l), CRP (mg/l), IL-6 (pg/ml), IL-8 (pg/ml), sIL-2R (pg/ ml) and sTNFRII (ng/ml) (median, maximum range) in febrile patients at peak during the febrile course compared with two control groups (afebrile cancer patients undergoing intensive chemotherapy and newly diagnosed cancer patients prior to intensive chemotherapy).

observed in all subgroups; however, the highest CRP levels at admission and as peak levels were also seen in patients with gram-negative bacteraemia and were significantly different only in comparison with FUO and localized infections (P # 0´014). For TNFRII at admission and for IL-6, IL-8, sIL-2R and sTNFRII during the entire course, the highest median levels were also seen in gram-negative infections. The peak values of these parameters were significantly different from those of FUO and localized infections (P # 0´046), gram-positive bacteraemia (only for IL-6 and sTNFRII, P # 0´035) and pneumonia (only for sTNFRII, P ˆ 0´011). Regardless of the kind of infection, the median peak levels occurred on day 1 (i.e. 24 h after admission) for PCT, IL-6 and IL-8, on day 2 for CRP and sTNFRII, and on day 3 for sIL-2R. To evaluate the influence of mucositis on the different parameters, only the largest subgroup (n ˆ 60), i.e. FUO (WHO mucositis score 0, n ˆ 37; score I/II, n ˆ 12; score III/IV, n ˆ 11) was analysed. No significant differences

between admission and peak values were documented between the episodes with WHO mucositis scores 0, I/II and III/IV. Nevertheless, the time of the peak values differed significantly between the episodes with mucositis score 0 and score III/IV for PCT (day 1 vs. day 3, P ˆ 0´014) and CRP (day 1 vs. day 4, P ˆ 0´008) respectively. Relationships between the inflammatory parameters At admission, no strong correlations were observed between the values of all inflammatory parameters, apart from IL-8 and the peripheral blood counts of neutrophils, lymphocytes or monocytes (for IL-8, r ˆ 20´597/20´484/20´484, P ˆ 0´000). Only weak correlations were recorded between the different inflammatory parameters themselves (r # 0´300, P # 0´05). Strong correlations (P ˆ 0´000) were found between the peak values of PCT and IL-8 (r ˆ 0´686), PCT and CRP (r ˆ 0´603), and CRP and IL-8 (r ˆ 0´700). Only weak or

Fig 2. Concentrations of PCT (mg/l), CRP (mg/l), IL-6 (pg/ml), IL-8 (pg/ml), sIL-2R (pg/ ml) and sTNFRII (ng/ml) (median, maximum range) in febrile patients at admission depending on the kind of infection. q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

Procalcitonin and Inflammatory Cytokines in Cancer Patients 1097

Fig 3. Courses of median concentrations of PCT (A), CRP (B), IL-6 (C), IL-8 (D), sIL-2R (E) and sTNFRII (F) in febrile patients from admission until defervescence or at least during the first five febrile days depending on the kind of infection.

moderate correlations (r # 0´500, P # 0´015) observed between all other peak values.

were

Diagnostic relevance of the inflammatory parameters At admission. The sensitivity and specificity of each parameter at different cut-off levels were analysed to predict gram-negative bacteraemia and FUO, respectively, among all febrile episodes at the time of admission (Table III). In predicting gram-negative bacteraemia, the optimum PCT cut-off level was 0´5 mg/l with a moderate sensitivity (60´0%) and high specificity (85´0%) that was similar to CRP, which had an optimum cut-off level of 50´0 mg/l with a moderate sensitivity (75´0%) and specificity (73´2%). The other parameters at the different cut-off levels showed either a low sensitivity (IL-6, IL-8 and sTNFRII) or a low specificity together with a high sensitivity (sIL-2R). In predicting FUO in all other episodes, the best sensitivity and specificity were observed with PCT at a cut-off level of 0´3 mg/l (64´3% and 69´1%), followed by IL-8 at a cut-off level of 50 pg/ml (67´2% and 55´6%).

Sequential analysis. In 116±121 of 122 febrile episodes, at least three values of each parameter were determined. These episodes were evaluable for analysis of the relationships between the sequential values and the clinical response. Depending on the parameter, the cut-off levels selected showed at least 60% specificity in predicting gram-negative bacteraemia (Table III): PCT 0´5 mg/l, CRP 50´0 mg/l, IL-6 20 pg/ ml, IL-8 100 pg/ml, sIL-2R 200 pg/ml and sTNFRII 5´0 ng/ ml respectively. The values were classified as either (i) not exceeding the cut-off value throughout the observation period, (ii) exceeding the cut-off value and decreasing rapidly with defervescence, (iii) exceeding the cut-off value and decreasing gradually with clinical improvement irrespective of the timepoint of final defervescence or (iv) persisting above the cut-off value or increasing during the observation period despite clinical improvement or a lack of clinical deterioration. For the evaluable episodes, the results of the analysis are listed in Table IV. Considering FUO episodes or all episodes together with the exception of gram-negative bacteraemia, more than 70% to 85% of patients showed a rapid decrease

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

1098 G. Fleischhack et al Table III. Diagnostic relevance of inflammatory markers in predicting Gram-negative bacteraemia and FUO among all febrile episodes at admission.

Cut-off level PCT (mg/l) 0´3 0´5 1´0 5´0 CRP (mg/l) 10´0 50´0 100´0 IL-6 (pg/ml) 20 50 100 IL-8 (pg/ml) 50 100 200 SIL-2R (pg/ml) 100 200 500 sTNFRII (ng/ml) 5´0 7´5 10´0

Prediction of Gram-negative bacteraemia*

Prediction of FUO²

Sensitivity %

Specificity %

Sensitivity %

Specificity %

80´0 60´0 50´0 40´0

44´0 85´0 97´0 99´0

64´3 94´6 100´0 100´0

69´1 34´5 14´5 9´1

100´0 75´0 25´0

20´7 73´2 95´1

14´0 76´0 96´0

80´5 39´0 9´8

44´4 22´2 22´2

80´2 85´6 88´3

86´2 94´8 96´6

16´7 11´1 7´4

44´4 33´3 22´2

55´9 79´4 92´2

67´2 89´7 98´3

55´6 32´3 16´7

100´0 44´4 11´1

32´6 67´3 90´1

35´1 66´7 89´5

75´9 35´2 9´3

55´6 55´6 55´6

65´7 89´2 98´0

67´2 91´4 98´3

40´7 20´4 11´1

*True positives, values above the cut-off level. ²True positives, values below the cut-off level.

with defervescence or persistently low levels of PCT, CRP, IL6 and IL-8. In contrast, approximately half the patients in these groups had persistently high levels or a gradual decrease in sIL-2R and sTNFRII. The majority of patients with gram-negative bacteraemia showed gradual decreases or persistently high levels of all parameters despite clinical improvement, in particular, of CRP, IL-6, sIL-2R and sTNFRII. Irrespective of the kind of infection, there were only a few persistently high levels of IL-8 and PCT, and most patients showed a decrease in these levels with clinical improvement. The courses of PCT and CRP correlated moderately (r ˆ 0´553, P ˆ 0´000). DISCUSSION In neutropenic cancer patients, early markers are needed that are regulated or released independently of the leucocyte count and the activity of the underlying disease. They have to reflect the severity of infection and distinguish episodes of high risk and low risk for septic complications. The value of the inflammatory cytokines IL-6 and IL-8, as well as PCT, in the diagnosis and assessment of infection in non-neutropenic patients has been well established in clinical studies during the past decade. Although there was a considerable interindividual variation of these

parameters, the point of time and plasma level reflected the onset and severity of infection (Hack et al, 1992; Assicot et al, 1993; Marty et al, 1994; Hammond & Potgieter, 1996; Gendrel et al, 1999). Comparison of PCT with IL-6, IL-8, sTNFRs and CRP in non-neutropenic patients revealed that it had a similar or higher diagnostic efficiency (De Werra et al, 1997; Rau et al, 1997; Gendrel et al, 1999; Fleischhack et al, 2000; Oberhoffer et al, 2000). High plasma or serum levels of PCT were also reported in systemic bacterial and parasitic infections in non-neutropenic children (RichardLenoble et al, 1997; Gendrel et al, 1999; Hatherill et al, 1999). In localized and viral infections, only normal or moderately elevated PCT plasma levels were observed (Assicot et al, 1993; Gendrel et al, 1999). Recent studies involving immunocompromised patients showed that these patients are also capable of producing high PCT serum levels in severe systemic bacterial or fungal infections (Al-Nawas & Shah, 1996; Beaune et al, 1998; Bernard et al, 1998; Engel et al, 1999; Ruokonen et al, 1999; Sauer et al, 2000). Only a few studies reported PCT plasma levels in infected and non-infected cancer patients (Bernard et al, 1998; Engel et al, 1999; Ruokonen et al, 1999; Fleischhack et al, 2000; Sauer et al, 2000). This study and our previously published study showed that the activity of the underlying malignant disease itself, the

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

Procalcitonin and Inflammatory Cytokines in Cancer Patients 1099 Table IV. Relationship between the course of the parameters and the clinical response to anti-microbial therapy. All febrile episodes without Gram-negative bacteraemias

Gram-negative bacteraemias

FUO

n

%

n

%

n

%

PCT (0´5 mg/l) 1 2 3 4

62 15 23 7

57´9 14´0 21´5 6´5

1 2 7 0

10´0 20´0 70´0 0´0

44 7 7 2

73´3 11´7 11´7 3´3

CRP (50´0 mg/l) 1 2 3 4

48 18 21 18

45´7 17´1 20´0 17´1

2 1 4 4

18´2 9´1 36´4 36´4

32 8 11 5

57´1 14´3 19´6 8´9

IL-6 (20 pg/ml) 1 2 3 4

87 9 9 4

79´8 8´3 8´3 3´7

4 0 5 3

33´3 0´0 41´7 25´0

47 7 3 3

78´3 11´7 5´0 5´0

IL-8 (100 pg/ml) 1 2 3 4

76 11 19 1

71´0 10´3 17´8 0´9

6 0 5 0

54´5 0´0 45´5 0´0

51 4 3 0

87´9 6´9 5´2 0´0

3 sIL-2R (200 pg/ml) 1 2 3 4

46 11 15 37

42´2 10´1 13´8 33´9

0 0 6 5

0´0 0´0 54´5 45´5

27 6 8 19

45´0 10´0 13´3 31´7

STNFRII (5´0 ng/ml) 1 2 3 4

39 0 26 31

35´8 0´0 23´9 40´4

1 0 5 4

10´0 0´0 50´0 40´0

25 0 14 21

41´7 0´0 23´3 35´0

Parameter (cut-off level)

P ˆ 0´004*

P ˆ 0´150* (n.s.)

P ˆ 0´000*

P ˆ 0´146* (n.s.)

P ˆ 0´001*

P ˆ 0´122* (n.s.)

P ˆ 0´000*

P ˆ 0´026*

P ˆ 0´000*

P ˆ 0´001*

P ˆ 0´002*

P ˆ 0´099 (n.s.)

*Pearson's chi-square test 1. Values did not exceed the cut-off value during the whole observation time. 2. Values exceeded the cut-off value during the observation time and decreased quickly with the defervescence. 3. Values exceeded the cut-off value during the observation time and decreased slowly with the clinical improvement independently from the time-point of the final defervescence. 4. Values persisted above the cut-off value or increased during the observation time despite clinical improvement or absent clinical deterioration. n.s., not significant

chemotherapy-induced tissue damage (such as severe mucositis) and the severity of neutropenia, lymphopenia or monocytopenia did not cause considerable increases in PCT plasma levels (Fleischhack et al, 2000). A similar behaviour was observed with CRP, IL-6 and IL-8 that mostly showed slightly increased levels or levels within the normal range in the control groups. In contrast, the levels of both sIL2-R and sTNFR were influenced by the activity of the cancer and its cytostatic treatment (Bodey et al, 1996;

Carter et al, 1996; Rathore et al, 1999). Despite this, infectious stimuli led to a higher increase of these parameters than did the underlying disease. Therefore, the individual initial levels of these parameters have to be known if they are to be used as inflammatory markers and serial measurements independent of the occurrence of febrile episodes would be necessary. In febrile episodes quite different courses were observed, although the highest levels of all parameters were seen in

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

1100 G. Fleischhack et al cases of gram-negative bacteraemia. At admission, PCT offered the best discrimination of all parameters between mild (FUO and localized infections) and serious infections (pneumonia and bacteraemia). During treatment, decreasing PCT levels reflected defervescence, rapid or gradual clinical improvement and successful anti-microbial therapy, irrespective of the kind of infection. An extremely high PCT level (. 500 mg/l) was seen in only one critically ill child with gram-negative bacteraemia, who died of septic shock and multiorgan failure. None of the other patients in this study developed septic complications such as severe sepsis syndrome and septic shock. No PCT decrease or transient PCT increase was observed without clinical improvement or deterioration respectively. Whether persistently elevated PCT levels represent an active, but controlled, infection and whether PCT is of relevance in fungal infections are questions that cannot be answered by our data. The PCT levels of our study at admission and during the course of the different kinds of infection were comparable with results reported for mild, moderate and severe infections of nonneutropenic and neutropenic patients (De Werra et al, 1997; Karzai et al, 1997; Bernard et al, 1998; Chiesa et al, 1998; Engel et al, 1999; Gendrel et al, 1999; Ruokonen et al, 1999; Sauer et al, 2000). In comparison with PCT, the CRP levels showed a similar course characterized by a more delayed increase and decrease. However, there was a wide overlap of CRP levels between the different kinds of infections at admission, as well as during the entire febrile episode, as already reported in previous studies (Heney et al, 1992; Katz et al, 1992; Engervall et al, 1995; Engel et al, 1998; Lehrnbecher et al, 1999). As in other studies, we recorded similar values of IL-6 and IL-8. However, there is a large variation in the ranges reported in the different studies, possibly owing to a varying sensitivity of the assays used, i.e. ELISA or immunoradiometric assay, or to a different structure of the study groups (Engervall et al, 1995; Steinmetz et al, 1995; De Werra et al, 1997; Engel et al, 1998; De Bont et al, 1999). Published data on the diagnostic value of sIL-2R and sTNFRII in systemic infections of non-neutropenic and neutropenic patients are sparse. Elevated sIL-2R serum levels were reported in viral infections, tuberculosis and infections following liver and bone marrow transplantation (Ito et al, 1990; Rossi et al, 1994, 1996; Izzo et al, 1996; Liem et al, 1998). In some of those studies (Rossi et al, 1994, 1996; Izzo et al, 1996), the levels detected in viral infections were similar to those documented in febrile episodes in our study. However, in our study, the highest median sIL-2R peak levels were observed in cases of gram-negative bacteraemia, and we believe that the T-cell activation and subsequent release of IL-2 and sIL-2R induced by viral infections might also be triggered by other infectious stimuli, e.g. gram-negative bacteria and their endotoxins. Increased sTNFRII serum levels as a result of bacterial infections such as sepsis and pneumonia were observed previously and showed a clear correlation with the severity of the infection (Van der Pol et al, 1993; Ertel et al, 1994; De Werra et al, 1997). Confirming data reported by de Werra et al (1997),

this study documented increased sTNFRII plasma levels throughout the febrile episodes even during the period of clinical improvement. To evaluate the diagnostic relevance of inflammatory markers in neutropenic children, an end-point such as septic shock, or non-survivors, is not practicable because of the low incidence of life-threatening or fatal septic complications in these patients (Hann et al, 1997). In order to find a risk-adapted anti-microbial therapy in neutropenic children, an inflammatory laboratory test should be sensitive and specific in the early recognition of febrile patients both at high risk and low risk of septic complications. Therefore, cases of gram-negative bacteraemia as a high-risk group and FUO as a low-risk group were chosen as end-points in this study. In contrast to previously published data (Heney et al, 1992; Engel et al, 1998; De Bont et al, 1999; Lehrnbecher et al, 1999), IL-6 and IL-8 were not the best predictors either of gram-negative bactaeremia (low sensitivity) or of FUO (low specificity) at the onset of fever; they had a diagnostic value similar to that of sIL-2R and sTNFR in the present study. PCT (optimum cut-off level 0´5 mg/l) and CRP (optimum cut-off level of 50´0 mg/l) were the best parameters in predicting gram-negative bacteraemia, and PCT was the best parameter (optimimum cut off level 0´3 mg/l) in predicting FUO. In the following sequential assessment of febrile episodes, PCT and IL-8 were the best parameters because the successful anti-microbial therapy and clinical improvement were reflected in persistently low levels or a decrease in levels, irrespective of the kind of infection. In contrast, sIL-2R and sTRNFRII were unsuitable for the assessment of the febrile course because of a high percentage of persistently high levels. Our results and those of recent studies involving neutropenic patients (Bernard et al, 1998; Ruokonen et al, 1999; Sauer et al, 2000) suggest that PCT is a more useful diagnostic inflammatory parameter in febrile cancer patients than IL-6, IL-8 and CRP. Although SIL-2R and sTNFRII might be helpful in the early detection of gramnegative bacteraemia, they are not suitable for the assessment of febrile courses. The overall diagnostic efficiency of PCT was superior to that of all other parameters in the early detection of gram-negative bacteraemia (high-risk episodes) and of FUO (low-risk episodes), as well as in the sequential assessment of febrile episodes. Larger prospective investigations are needed to determine the prognostic value of PCT as an inflammatory marker in the context of other laboratory and clinical findings, and to characterize its diagnostic relevance in a risk-adapted anti-microbial therapy. REFERENCES Al-Nawas, B. & Shah, P.M. (1996) Procalcitonin in patients with and without immunosuppression and sepsis. Infection, 24, 434± 436. Al-Nawas, B., Krammer, I. & Shah, P.M. (1996) Procalcitonin in diagnosis of severe infections. European Journal of Medical Research, 1, 331±333. Assicot, M., Gendrel, D., Carsin, H., Raymond, J., Guilbaud, J. &

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

Procalcitonin and Inflammatory Cytokines in Cancer Patients 1101 Bohuon, C. (1993) High serum procalcitonin concentrations in patients with sepsis and infection. Lancet, 341, 515±518. Baggiolini, M. & Clark Lewis, I. (1992) Interleukin-8, a chemotactic and inflammatory cytokine. FEBS Letters, 307, 97±101. Bauer, J. (1989) Interleukin-6 and its receptor during homeostasis, inflammation, and tumor growth. Klinische Wochenschrift, 67, 697±706. Beaune, G., Bienvenu, F., Pondarre, C., Monneret, G., Bienvenu, J. & Souillet, G. (1998) Serum procalcitonin rise is only slight in two cases of disseminated aspergillosis. Infection, 26, 168±169. Bernard, L., Ferriere, F., Casassus, P., Malas, F., Leveque, S., Guillevin, L. & Lortholary, O. (1998) Procalcitonin as an early marker of bacterial infection in severely neutropenic febrile adults. Clinical Infectious Diseases, 27, 914±915. Bodey, B., Psenko, V., Lipsey, A.L. & Kaiser, H.E. (1996) Soluble interleukin-2 receptors in sera of children with primary malignant neoplasms. Anticancer Research, 16, 219±224. Carter, A., Haddad, N., Draxler, I., Israeli, E., Raz, B. & Rowe, J.M. (1996) Expression and role in growth regulation of tumour necrosis factor receptors p55 and p75 in acute myeloblastic leukaemia cells. British Journal of Haematology, 92, 116±126. Chiesa, C., Pacifico, L., Mancuso, G. & Panero, A. (1998) Procalcitonin in pediatrics: overview and challenge. Infection, 26, 236±241. De Bont, E.S., Vellenga, E., Swaanenburg, J.C., Fidler, V., Visser-van Brummen, P.J. & Kamps, W.A. (1999) Plasma IL-8 and IL-6 levels can be used to define a group with low risk of septicaemia among cancer patients with fever and neutropenia. British Journal of Haematology, 107, 375±380. De Werra, I., Jaccard, C., Corradin, S.B., Chiolero, R., Yersin, B., Gallati, H., Assicot, M., Bohuon, C., Baumgartner, J.D., Glauser, M.P. & Heumann, D. (1997) Cytokines, nitrite/nitrate, soluble tumor necrosis factor receptors, and procalcitonin concentrations: comparisons in patients with septic shock, cardiogenic shock, and bacterial pneumonia. Critical Care Medicine, 25, 607± 613. Engel, A., Mack, E., Kern, P. & Kern, W.V. (1998) An analysis of interleukin-8, interleukin-6 and C-reactive protein serum concentrations to predict fever, gram-negative bacteraemia and complicated infection in neutropenic cancer patients. Infection, 26, 213±221. Engel, A., Steinbach, G., Kern, P. & Kern, W.V. (1999) Diagnostic value of procalcitonin serum levels in neutropenic patients with fever: comparison with interleukin-8. Scandinavian Journal of Infectious Diseases, 31, 185±189. Engervall, P., Granstrom, M., Andersson, B. & Bjorkholm, M. (1995) Monitoring of endotoxin, interleukin-6 and C-reactive protein serum concentrations in neutropenic patients with fever. European Journal of Haematology, 54, 226±234. Ertel, W., Scholl, F.A., Gallati, H., Bonaccio, M., Schildberg, F.W. & Trentz, O. (1994) Increased release of soluble tumor necrosis factor receptors into blood during clinical sepsis. Archives of Surgery, 129, 1330±1336. Fleischhack, G., Cipic, D., Juettner, J., Hasan, C. & Bode, U. (2000) Procalcitonin ± a sensitive inflammation marker of febrile episodes in neutropenic chidren with cancer. Intensive Care Medicine, 26, S202±S211. Gendrel, D., Raymond, J., Coste, J., Moulin, F., Lorrot, M., Guerin, S., Ravilly, S., Lefevre, H., Royer, C., Lacombe, C., Palmer, P. & Bohuon, C. (1999) Comparison of procalcitonin with C-reactive protein, interleukin 6 and interferon-alpha for differentiation of bacterial vs. viral infections. Pediatric Infectious Disease Journal, 18, 875±881. Hack, C.E., Hart, M., van Schijndel, R.J., Eerenberg, A.J., Nuijens,

J.H., Thijs, L.G. & Aarden, L.A. (1992) Interleukin-8 in sepsis: relation to shock and inflammatory mediators. Infection and Immunity, 60, 2835±2842. Hammond, J.M. & Potgieter, P.D. (1996) The influence of surgery on cytokines in patients with intraabdominal sepsis. Anaesthesia and Intensive Care, 24, 430±434. Hann, I., Viscoli, C., Paesmans, M., Gaya, H. & Glauser, M. (1997) A comparison of outcome from febrile neutropenic episodes in children compared with adults: results from four EORTC studies. International Antimicrobial Therapy Cooperative Group (IATCG) of the European Organization for Research and Treatment of Cancer (EORTC). British Journal of Haematology, 99, 580±588. Hatherill, M., Tibby, S.M., Sykes, K., Turner, C. & Murdoch, I.A. (1999) Diagnostic markers of infection: comparison of procalcitonin with C reactive protein and leucocyte count. Archives of Disease in Childhood, 81, 417±421. Heney, D., Lewis, I.J., Evans, S.W., Banks, R., Bailey, C.C. & Whicher, J.T. (1992) Interleukin-6 and its relationship to C-reactive protein and fever in children with febrile neutropenia. Journal of Infectious Diseases, 165, 886±890. Ito, M., Kojiro, N., Shirasaka, T., Ikeda, T., Tachibana, I., Mima, T. & Kokubu, T. (1990) Soluble interleukin-2 receptors in pulmonary tuberculosis. Japanese Journal of Medicine, 29, 583±586. Izzo, F., Curley, S., Maio, P., Leonardi, E., Imparato, L., Giglio, S., Cremona, F. & Castello, G. (1996) Correlation of soluble interleukin-2 receptor levels with severity of chronic hepatitis C virus liver injury and development of hepatocellular cancer. Surgery, 120, 100±105. Jaye, D.L. & Waites, K.B. (1997) Clinical applications of C-reactive protein in pediatrics. Pediatric Infectious Disease Journal, 16, 735± 747. Karzai, W., Oberhoffer, M., Meier, H.A. & Reinhart, K. (1997) Procalcitonin ± a new indicator of the systemic response to severe infections. Infection, 25, 329±334. Katz, J.A., Mustafa, M.M., Bash, R.O., Cash, J.V. & Buchanan, G.R. (1992) Value of C-reactive protein determination in the initial diagnostic evaluation of the febrile, neutropenic child with cancer. Pediatric Infectious Disease Journal, 11, 708±712. Lehrnbecher, T., Venzon, D., de Haas, M., Chanock, S.J. & Kuehl, J. (1999) Assessment of measuring circulating levels of interleukin6, interleukin-8, C-reactive protein, soluble Fc gamma receptor type III, and mannose-binding protein in febrile children with cancer and neutropenia. Clinical Infectious Diseases, 29, 414±419. Liem, L.M., van Houwelingen, H.C. & Goulmy, E. (1998) Serum cytokine levels after HLA-identical bone marrow transplantation. Transplantation, 66, 863±871. Lotz, M. (1995) Interleukin-6: a comprehensive review. Cancer Treatment Research, 80, 209±233. Marty, C., Misset, B., Tamion, F., Fitting, C., Carlet, J. & Cavaillon, J.M. (1994) Circulating interleukin-8 concentrations in patients with multiple organ failure of septic and nonseptic origin. Critical Care Medicine, 22, 673±679. Oberhoffer, M., Vogelsang, H., Russwurm, S., Hartung, T. & Reinhart, K. (1999) Outcome prediction by traditional and new markers of inflammation in patients with sepsis. Clinical Chemistry and Laboratory Medicine, 37, 363±368. Oberhoffer, M., Russwurm, S., Bredle, D., Chatzinicolaou, K. & Reinhart, K. (2000) Discriminative power of inflammatory markers for prediction of tumor necrosis factor-alpha and interleukin-6 in ICU patients with systemic inflammatory response syndrome (SIRS) or sepsis at arbitrary time points. Intensive Care Medicine, 26, S170±S174. Rathore, V.B., Advani, S.H. & Nadkarni, J.J. (1999) Altered release of tumor necrosis factor and its soluble receptor in non-Hodgkin's

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102

1102 G. Fleischhack et al lymphoma patients. Cancer Detection and Prevention, 23, 226± 231. Rau, B., Steinbach, G., Gansauge, F., Mayer, J.M., Grunert, A. & Beger, H.G. (1997) The potential role of procalcitonin and interleukin 8 in the prediction of infected necrosis in acute pancreatitis. Gut, 41, 832±840. Richard-Lenoble, D., Duong, T.H., Ferrer, A., Lacombe, C., Assicot, M., Gendrel, D., Bohuon, C. & Kombila, M. (1997) Changes in procalcitonin and interleukin 6 levels among treated African patients with different clinical forms of malaria. Transactions of the Royal Society of Tropical Medicine and Hygiene, 91, 305±306. Rossi, S.J., Schroeder, T.J., Muth, K.L., Hanto, D.W., Balistreri, W.F. & Ryckman, F.C. (1994) Soluble interleukin-2 receptor monitoring during bacterial and viral infections in liver transplant recipients: a comparative evaluation. Clinical Transplantation, 8, 479±484. Rossi, S.J., Schroeder, T.J., Muth, K.L., Hanto, D.W., Munda, R., Hariharan, S., First, M.R., Ryckman, F.C. & Balistreri, W.F. (1996) Serial monitoring of soluble interleukin-2 receptor as a rapid marker of cytomegalovirus infection and response to antiviral therapy. Clinical Transplantion, 10, 45±50. Ruokonen, E., Nousiainen, T., Pulkki, K. & Takala, J. (1999) Procalcitonin concentrations in patients with neutropenic fever. European Journal of Clinical Microbiology and Infectious Diseases, 18, 283±285. Sauer, M., Tiede, K., Volland, R., Fuchs, D. & Zintl, F. (2000) Procalcitonin and C-reactive protein: comparison of two markers for sepsis-syndrome in severely immunocompromised children

after bone marrow transplantation. Klinische Paediatrics, 212, 10±15. Staehler, M., Hammer, C., Meiser, B. & Reichart, B. (1997) Procalcitonin: a new marker for differential diagnosis of acute rejection and bacterial infection in heart transplantation. Transplantation Proceedings, 29, 584±585. Steinmetz, H.T., Herbertz, A., Bertram, M. & Diehl, V. (1995) Increase in interleukin-6 serum level preceding fever in granulocytopenia and correlation with death from sepsis. Journal of Infectious Diseases, 171, 225±228. Steinshamn, S., Brekke, O.L. & Waage, A. (1995) Soluble tumour necrosis factor receptors, tumour necrosis factor and interleukin6 in serum in granulocytopenic patients with fever. British Journal of Haematology, 89, 719±724. Suffredini, A.F., Fantuzzi, G., Badolato, R., Oppenheim, J.J. & O'Grady, N.P. (1999) New insights into the biology of the acute phase response. Journal of Clinical Immunology, 19, 203±214. Van der Pol, T., Jansen, J., van Leenen, D., von der Mohlen, M., Levi, M., ten Cate, H., Gallati, H., ten Cate, J.W. & van Deventer, S.J. (1993) Release of soluble receptors for tumor necrosis factor in clinical sepsis and experimental endotoxemia. Journal of Infectious Diseases, 168, 955±960. Waage, A., Remick, D., Steinshamn, S., Deforge, L. & Lamvik, J. (1994) Interleukin 8 in serum in granulocytopenic patients with infections. British Journal of Haematology, 86, 36±40. World Health Organization (1979) WHO Handbook for reporting results of cancer treatment. WHO Offset Publication, 48, 1±41.

q 2000 Blackwell Science Ltd, British Journal of Haematology 111: 1093±1102