Bone Marrow Transplantation (2003) 32, 941–945 & 2003 Nature Publishing Group All rights reserved 0268-3369/03 $25.00 www.nature.com/bmt

Immunobiology Procalcitonin and C-reactive protein do not discriminate between febrile reaction to anti-T-lymphocyte antibodies and Gram-negative

HJ Dornbusch, V Strenger, R Kerbl, H Lackner, W Schwinger, P Sovinz and C Urban

Division of Paediatric Haematology/Oncology, Karl-Franzens-University, Graz, Austria

Summary: antibody infusion following kidney transplantation with no evidence of .9–11 Treatment with antibodies against T-lymphocytes usually To further investigate these observations, we analysed triggers a febrile response potentially mimicking or the characteristics of PCT and CRP serum levels in masking infection. Procalcitonin (PCT) is considered a paediatric patients with haematological or oncological sensitive and specific marker of systemic bacterial and diseases during treatment with T-cell antibodies and fungal infection. It was the aim of this study to investigate compared these with those occurring with episodes of the characteristics of PCT and C-reactive protein (CRP) Gram-negative sepsis. Furthermore, the roles of cytotoxic during treatment with polyclonal or monoclonal anti-T- chemotherapy, irradiation and products were exam- cell antibodies, in order to examine the ability of these ined in the context of other potential stimuli for the release parameters to distinguish between systemic bacterial of these parameters. infection and reaction to antibody treatment. Thus, 15 consecutive febrile episodes after T-cell antibody infusion without clinical signs of infection were compared with nine Patients and methods episodes of Gram-negative sepsis. After T-cell antibody infusion PCT and CRP serum levels increased to a similar In all, 24 episodes of either Gram-negative sepsis or T-cell extent as in Gram-negative sepsis. Therefore, during antibody treatment in 21 paediatric patients with haema- T-cell antibody treatment neither PCT nor CRP are tological or oncological disorders were included in this adequate for differentiating between fever due to infection study. Patient characteristics and underlying diseases are or to unspecific cytokine release. shown in Table 1. Bone Marrow Transplantation (2003) 32, 941–945. Group A consisted of 15 consecutive febrile episodes in doi:10.1038/sj.bmt.1704265 15 patients treated with T-cell antibodies between October Keywords: antilymphocyte antibodies; fever; procalcito- 1998 and July 2001. Four patients received Thymoglobu- nin; CRP; sepsis line Merieuxs (0.75 ml/kg/day), for 2, 4 (two patients) or 8 days, six ATG-Freseniuss (5 mg/kg/day) for 4 days and five OKT 3 (daily dose escalation from 0.01 mg/kg to a final Treatment with polyclonal (ATG) or monoclonal (OKT 3) dose of 0.1 mg/kg/day on day 4) for 14–21 days either as T-cell antibodies is an integral part of numerous condition- part of the conditioning regimen prior to allogeneic SCT or ing regimens for allogeneic stem cell transplantation (SCT) as conventional treatment for SAA (n ¼ 1). T-cell anti- and of conventional therapy for severe aplastic anaemia bodies were administered as a 4–6 h infusion. In addition, (SAA). Febrile responses after the first doses of T-cell cyclophosphamide (n ¼ 12), busulphan (n ¼ 6), thiotepa antibodies pose a diagnostic dilemma in these patients who (n ¼ 4), etoposide (n ¼ 3), fludarabine (n ¼ 1) and total are frequently neutropenic.1 CRP is known as a nonspecific body/lymphoid irradiation (n ¼ 4) were used as condition- marker unable to distinguish between bacterial or fungal ing therapy for SCT starting before (n ¼ 12), simulta- infection and other causes of fever during , neously with (n ¼ 1) or after (n ¼ 1) the first antibody including certain viral , fever related to the administration. Five patients received blood products underlying malignancy or to engraftment.2–4 Recent pub- (platelet transfusions/n ¼ 4, packed erythrocyte transfu- lications suggest an even higher diagnostic value of sions/n ¼ 3) on the first day of T-cell antibody treatment. In procalcitonin (PCT).4–8 However, there are sporadic all cases, T-cell antibody infusion was accompanied by reports on high PCT levels in adult patients after T-cell steroid administration. Seven patients received cyclosporin A as GvHD prophylaxis, starting on the first (n ¼ 2), fourth (n ¼ 4) or sixth (n ¼ 1) day of antibody treatment. Collection of blood samples for determination of Correspondence: Dr HJ Dornbusch, Department of Paediatrics and inflammatory markers was commenced at the beginning Adolescence Medicine, Division of Paediatric Haematology/Oncology, Auenbruggerplatz 30, A-8036 Graz, Austria; of conditioning therapy and was performed daily for at E-mail: [email protected] least 3 weeks. Baseline PCT levels measured prior to the Received 07 November 2002; accepted 29 March 2003 first T-cell antibody infusion were available for 14/15 PCT and CRP under anti-T-cell antibody treatment HJ Dornbusch et al 942 Table 1 Patient characteristics and underlying haematological and levels vs baseline was calculated for each component of oncological diseases conditioning therapy using the following formula: (level of Group A Group B the day after first administrationÀbaseline level before administration)/baseline level. Differences were assessed Age 7.5 a (11 months–15.6 a) 14.4 a (5.6–22.9) using the Mann–Whitney U-test. N (male/female) 15 (7/8) 9 (3/6) For assessment of diagnostic values receiver-operating SAA 6 0 ALL 2 4 characteristics (ROC) curves were calculated, and AUC AML 2 1 was compared with a reference value of 0.5, indicating no CML 2 1 diagnostic value. Lymphoma 0 2 Since the diagnosis of SAA occurred in group A only, Others 3 1 additional statistical analysis was performed disregarding the six SAA patients, in order to exclude a possible bias Group A: T-cell antibody treatment. Group B: Gram-negative sepsis. caused by an uneven distribution of patient characteristics. All computations were performed by means of SPSS for Windows 10.0. patients. Blood for the first determination of inflammatory markers after starting T-cell antibody treatment was drawn 13–22 (median 18) h after the start of antibody infusion. Results Prior to T-cell antibody treatment, the white blood count (WBC) was 400–34.600/ml (median 2200/ml). A total of 269 PCT (group A: 228, group B: 41) and 396 No physical signs or symptoms of infection other than CRP (group A: 296, group B: 100) measurements of 24 fever were observed. Microbiological findings were un- episodes in 21 patients were evaluated. remarkable. Despite additional steroid administration, all 15 patients Group B comprised nine consecutive episodes of Gram- treated with T-cell antibodies became febrile 2–12 (median negative sepsis in nine patients between January 1999 and 5) h after the start of infusion (peak temperatures 37.7– February 2001. Sepsis was defined according to the ACCP/ 40.71C/median 38.91C). In 11 episodes (78.5%), the body SCCM Consensus Conference Committee guidelines12 temperature exceeded 37.91C. In 10 episodes, the peak adapted for children13 (infection þ two or more of the temperature was reached during the first 24 h. Deferves- following: temperature 438 or o361C, heart rate or cence was observed within 1–4 (median 2) days. respiratory rate above 2 Â s.d.). In addition, microbiologi- Similar peak temperatures (37.6–40.61C/median 39.51C) cal proof of Gram-negative bacteraemia was required for were measured during sepsis. inclusion. Blood cultures grew Escherichia coli (n ¼ 3), In group A (T-cell antibody-treated patients), PCT levels Klebsiella pneumoniae (n ¼ 2), Pseudomonas aeruginosa, (n ¼ 228) were evaluated from the start of conditioning Stenotrophomonas maltophilia, Enterobacter sakazakii and therapy until day 11 after the start of T-cell antibody Proteus mirabilis (n ¼ 1, each). The WBC on the day of treatment. fever onset was 100–1500/ml (median 150/ml). All PCT baseline levels prior to T-cell antibody infusion In all patients with Gram-negative sepsis blood for the were below 1.0 ng/ml (o0.1–0.78, median 0.18 ng/ml). In first PCT and CRP determination was drawn simulta- 12 episodes, the baseline PCT was in the normal range neously with blood samples for microbiological cultures (o0.5 ng/ml). prior to the start of empirical therapy (merope- After T-cell antibody infusion the median increase in nem n ¼ 8, imipenem/cilastatin n ¼ 2, additional netilmicin serum PCT exceeded the baseline level 39 times. In all but n ¼ 5). Follow-up blood samples were drawn every morning two cases blood samples collected 13–22 h (median 18 h) on the following 2–9 days, until clinical improvement. after starting the infusion showed PCT concentrations In all patients, body temperature was measured at least five times daily. 25

Laboratory methods 20

CRP was measured turbidimetrically using the Hitachi 917 15 analytic device and reagents of Roche Diagnostics (Vienna, Austria), PCT was determined by means of an immunolu- 10 minometric assay (BRAHMS Diagnostika GmbH, Berlin, PCT (ng/ml) Germany). 5

Statistical analysis 0 −2 −1 0 1 2 3 4 5 6 7 Data are presented as medians and ranges or as counts Days and proportions. Considering other potential stimuli than Figure 1 Daily median PCT serum levels (ng/ml) during T-cell antibody T-cell antibodies for the release of PCT and CRP, we also treatment (circles) or during sepsis (squares). Day 0: Start of T-cell examined the influence of cytotoxic chemotherapy and antibody treatment/onset of sepsis. Error bars display the interquartile radiation treatment. The increase of PCT and CRP serum ranges.

Bone Marrow Transplantation PCT and CRP under anti-T-cell antibody treatment HJ Dornbusch et al 943 Table 2 Comparison of body temperature, CRP and PCT between groups A and B

Temperature CRP PCT

Normal range o37.01C o5 mg/la o0.5 ng/mla

Group A Group B Group A Group B Group A Group B

Values on day 1 after fever onset 37.61C 37.71C 40.0 mg/l 129.0 mg/l 7.2 ng/ml 5.1 ng/ml (36.9–40.2) (36.7–40.6) (14.0–131.0) (65.0–329.0) (0.2–38.9) (0.57–22.02)

Episodes with peak values above normal values 15/15 (100%) 9/9 (100%) 15/15 (100%) (9/9) (100%) 14/15 (93%) (9/9) (100%)

Increase versus baseline 1.0b — 3.8c — 39.3c — (À0.4–4.2) (0.07–23.3) (À0.5–199)

Differences between group A and B/day 1 after fever onset P ¼ 1.0 (NS) P ¼ 0.003 P ¼ 0.799 (NS)

NS ¼ nonsignificant. aAccording to the respective product information. bCentigrade. c(Value after T-cell antibody infusionÀbaseline prior to infusion)/baseline prior to infusion.

a PCT 250 1.00 200

150 0.75 100 CRP (mg/ml) 50 0.50 0 Sensitivity −2 −1 0 1 2 3 4 5 6 7 Days AUC = 0.552 0.25 Figure 3 Daily median CRP serum levels (mg/l) during T-cell antibody treatment (circles) or during sepsis (squares). Day 0: Start of T-cell antibody treatment/onset of sepsis. Error bars display the interquartile ranges. 0.00 0.00 0.25 0.50 0.75 1.00 1-specificity Table 3 Median increase of PCT and CRP serum levels related to each component of conditioning treatment b CRP Conditioning treatment Median increase vs baselinea 1.00 PCT CRP

Radiation treatment À0.11 0.13 Busulphan 0.11 0 0.75 Cyclophosphamide 0 0 Thio-tepa À0.32 0 Etoposide À0.6 0.5 Fludarabine À0.67 0 T-cell antibodyb 39.33 3.8 0.50 a(Level of the day after first administrationÀbaseline level before

Sensitivity administration)/baseline level. bDifferences between thymoglobuline, ATG and OKT III statistically not AUC = 0.861 significant. 0.25 AUC = 0.861

Figure 2 ROC curves for discrimination between sepsis and reaction due 0.00 to T-cell antibody treatment by means of PCT (a) and CRP (b) on the first 0.00 0.25 0.50 0.75 1.00 day after fever onset.

Bone Marrow Transplantation PCT and CRP under anti-T-cell antibody treatment HJ Dornbusch et al 944 above the upper normal value with a maximum of 38.9 and a No significant change of PCT or CRP levels was median of 7.15 ng/ml. Thereafter, PCT levels decreased and observed with any component of conditioning treatment returned to normal between 2 and 11 (median 5.5) days after except for T-cell antibodies (Table 3). the start of T-cell antibody treatment. One patient developed Similarly, administration of blood products did not a first PCT increase (1.2 ng/ml) later than 24 h after T-cell influence the inflammatory markers. antibody infusion with no more concomitant conditioning treatment. The PCT remained within the normal range with T-cell antibody treatment in only one patient. In group B (septic patients), PCT levels (n ¼ 41) were Discussion evaluated from onset of fever for 3–10 days depending on the clinical course. On the first day after fever onset Although PCT has been recognized as a rather specific PCT levels were above the upper normal value in all marker of sepsis,5,6 there are also recent reports of high septic patients (Table 2) and reached a maximum on day 1 PCT levels after T-cell antibody infusion in adult renal (n ¼ 6), 2 (n ¼ 2) or 3 (n ¼ 1). Thus, a wide overlap was transplant patients.9–11 This evidence is confirmed and observed in the levels of T-cell antibody-treated patients. broadened by our data derived from paediatric patients Neither PCT peak levels nor PCT concentrations on the 3 with haematological or oncological diseases. days after onset of fever differed significantly between T-cell antibody treatment was followed by an increase in septic patients and patients treated with T-cell antibodies PCT levels in all patients. In 14 of 15 episodes (including (Figure 1). the patient with conventional T-cell antibody treatment for The ROC curve for PCT levels on the day after fever SAA), PCT concentrations exceeded the upper normal onset (Figure 2a) showed an area under the curve of 0.552 value. For all 15 patients, the median PCT elevation on day (P ¼ 0.77), suggesting that PCT has no diagnostic value in 1 after T-cell antibody infusion was nearly 40 times the predicting Gram-negative sepsis in patients receiving T-cell baseline. antibodies. In all, 14 patients received conditioning treatment in In group A, CRP levels (n ¼ 296) were evaluated from the addition to T-cell antibodies, but no component of start of conditioning therapy until day 14 after the start of conditioning therapy except T-cell antibodies was followed T-cell antibody treatment. by a significant increase in PCT levels, neither did blood CRP levels were elevated after T-cell antibody infusion in products have any influence on serum levels of the all patients, with peak values of up to 215.8 mg/l inflammatory parameters. (14.0–215.8, median 40 mg/l). The median increase was On day 1 of T-cell antibody-related fever, when four times the baseline. Peak levels were reached within 1 laboratory parameters are most relevant for therapeutic day in 10 patients, and CRP returned to normal at the decisions, PCT even showed a tendency towards higher earliest on day 4, with a median of day 7 after T-cell serum levels than in patients with Gram-negative sepsis, antibody infusion. In five patients, the CRP remained and during the entire course of the studied episodes the two elevated for more than 20 days, independent of the groups were statistically indistinguishable by means of PCT duration of T-cell antibody treatment. levels. According to the ROC curve, PCT is not predictive In group B, CRP levels were evaluated until at least 5 of sepsis while T-cell antibody treatment is in pro- days after onset of fever (n ¼ 100). CRP concentrations gress. Thus, according to this study and to data derived peaked on the second (n ¼ 6) or third (n ¼ 3) day of fever from adult kidney transplant patients, PCT does not offer and showed highest levels between 74 and 394 mg/l (median any diagnostic benefit in patients treated with T-cell 129 mg/l). CRP levels on the 3 days after fever onset as well antibodies. as the peak levels showed significant differences when Interestingly, CRP, which is commonly considered less compared with group A, with P-values of 0.003 and 0.018. specific than PCT, seems to have a higher diagnostic value However, there was a wide overlap between both groups in distinguishing T-cell antibody-related fever from sepsis. (Figure 3). Yet, although significant differences are observed on the The ROC curve for differentiating T-cell antibody first 4 days after onset of fever and although the com- reaction from Gram-negative sepsis by CRP serum levels parison between ROC curves seems to favour CRP, on day 1 after fever onset (Figure 2b) shows an area under specificity of this parameter is insufficient, since there is the curve of 0.861 (P ¼ 0.005). also a marked influence of T-cell antibody treatment on Using an optimized cutoff point of 62 mg/l, the CRP concentration, which causes a median increase of four sensitivity of CRP in indicating Gram-negative sepsis in times the baseline and a wide overlap with the levels this sample is 100%, while specificity is 60%. measured in septic patients. Exclusion of the six patients with SAA (a diagnosis In summary, febrile episodes during treatment with present in group A only) from statistical analysis had no T-cell antibodies go along with markedly elevated levels significant influence on the results shown above. of PCT and CRP. There is a tendency toward higher serum There was no significant difference between T-cell levels of CRP during sepsis. However, this does not permit antibody preparations (Thymoglobuline Merieuxs, ATG- sufficient diagnostic value in clinical practice for the Freseniuss, OKT 3) concerning their influence on PCT and individual case. Thus, during T-cell antibody treatment, CRP levels; however, validity of the statistical evaluation is none of these parameters seems to be helpful in distinguish- questionable due to the low number of patients in each ing fever due to infection from that caused by nonspecific subgroup. cytokine release.

Bone Marrow Transplantation PCT and CRP under anti-T-cell antibody treatment HJ Dornbusch et al 945 Acknowledgements interleukin-8, soluble interleukin-2 receptor and soluble tumour necrosis factor receptor II. Br J Haematol 2000; 111: We are indebted to Andrea Berghold and Petra Ofner from the 1093–1102. Institute for Medical Informatics, Statistics and Documentation 7 Nijsten MW, Olinga P, The TH et al. Procalcitonin behaves as for their professional assistance in statistical analysis. a fast responding acute phase protein in vivo and in vitro. Crit Care Med 2000; 28: 458–461. 8 Lacour AG, Gervaix A, Zamora SA et al. Procalcitonin, References IL-6, IL-8, IL-1 receptor antagonist and C-reactive protein as identificators of serious bacterial infections in children 1 Dearden C, Foukaneli T, Lee P et al. The incidence and with fever without localising signs. Eur J Pediatr 2001; 160: significance of fevers during treatment with antithymocyte 95–100. globulin for aplastic anaemia. Br J Haematol 1998; 103: 9 Kuse ER, Jaeger K. Procalcitonin increase after anti-CD3 846–848. monoclonal antibody therapy does not indicate infectious 2 Peltola H, Jaakkola M. C-reactive protein in early detection of disease. Transplant Int 2001; 14: 55 (letter). bacteremic versus viral infections in immunocompetent and 10 Sabat R, Hoflich C, Docke WD et al. Massive elevation of compromised children. J Pediatr 1988; 113: 641–646. procalcitonin plasma levels in the absence of infection in 3 Legouffe E, Rodriguez C, Picot MC et al. C-reactive protein kidney transplant patients treated with pan-T-cell antibodies. serum level is a valuable and simple prognostic marker in Intens Care Med 2001; 27: 987–991. non Hodgkin’s lymphoma. Leukemia Lymphoma 1998; 31: 11 Eberhard OK, Langefeld I, Kuse ER et al. Procalcitonin 351–357. in the early phase after renal transplantation – will it add to 4 Dornbusch HJ, Kerbl R, Lackner H et al. Procalcitonin and C- diagnostic accuracy? Clin Transplant 1998; 12: 206–211. reactive protein levels at the time of engraftment after stem cell 12 American College of Chest Physicians/Society of Critical Care transplantation. Acta Chir Austr 2000; 32: 37 (abstract P38). Medicine Consensus Conference Committee. Definitions for 5 Assicot M, Gendrel D, Carsin H et al. High serum sepsis and organ failure and guidelines for the use of procalcitonin concentrations in patients with sepsis and innovative therapies in sepsis. Crit Care Med 1992; 20: infection. Lancet 1993; 341: 515–518. 864–874. 6 Fleischhack G, Kambeck I, Cipic D et al. Procalcitonin 13 Hayden WR. Sepsis terminology in pediatrics. J Pediatr 1994; in paediatric cancer patients: its diagnostic relevance is 124: 657–658. superior to that of C-reactive protein, interleukin-6,

Bone Marrow Transplantation