A Multicenter Cohort Study to Explore Differentiating Factors between Tumor Fever and Infection among Advanced Cancer Patients

Abstract

Background

: Tumor fever and infection are common febrile etiologies among advanced cancer patients. To date, only few studies have been conducted to differentiate between tumor fever and infections.

Objective:

This study aimed to identify discriminating factors that provide rapid results and are feasible and minimally invasive for discriminating between tumor fever and infection in advanced cancer patients.

Methods

: This is a retrospective cohort study. Advanced cancer patients with clinically diagnosed tumor fever or infection, who received medical treatment from palliative care specialists in 10 nationwide Japanese hospitals, were consecutively identified during August 2012 and November 2014. The symptoms, physical findings, blood test results at baseline and during fever, imaging findings, and sociodemographic factors of these patients were retrospectively extracted.

Results

: Thirty-three patients with tumor fever and 72 patients with infection were identified. Their mean age was 68.8 years, 68 (64.8%) were men, and the median palliative performance status (PPS) was 50. Statistically significant factors predicting tumor fever by logistic regression analysis were as follows: deterioration of PPS (odds ratio, 0.078), shaking chills during fever (0.067), and change from baseline data of neutrophil/lymphocyte ratio of ≥5 (0.14).

Conclusions

: Shaking chills during fever, and changes from baseline of performance status and white blood cell differentiation can be useful to differentiate between tumor fever and infection among advanced cancer patients. Further confirmatory studies are needed.

Background

Seventy percent of cancer patients experience fever,¹ and tumor fever and infection are common febrile etiologies among cancer patients. Tumor fever is a common cause of fever of unknown origin² and occurs in 27% of cancer patients.³ Tumor fever occurs in advanced cancer patients in particular, because it is reportedly more common with metastasis.⁴ Meanwhile, the prevalence of infection among advanced cancer patients is reported to be 40%–80%.^5–7

Some studies^8–10 showed that the magnitude of responsiveness to nonsteroidal anti-inflammatory drugs (NSAIDs) can identify tumor fever. Chang proposed the diagnostic criteria for neoplastic fever¹¹: body temperature (BT) >37.8°C at least once each day; fever duration >2 weeks; lack of evidence of infection on physical examinations, laboratory examinations such as smears or cultures, and radiologic examinations; absence of allergic mechanisms; lack of response of fever to an empiric, adequate antibiotic therapy for >7 days; and prompt, complete lysis of fever, as determined by the naproxen test, with sustained normal temperature, while receiving naproxen. However, it is difficult to apply these criteria to advanced cancer patients because of the prolonged period for observing fever and antibiotic response and invasive culture tests and radiologic examinations, which can cause unnecessary distress for patients. Tumor fever therapy can improve symptoms and activities of daily living.^8,12,13 Thus, a rapid and low-invasive diagnosis of tumor fever among advanced cancer patients could quickly improve these patients' quality of life (QOL) without prescribing unnecessary antibiotics.

Objective

The purpose of this study was to identify factors that provide rapid results and are feasible and minimally invasive for discriminating between tumor fever and infection in advanced cancer patients, which could improve patients' QOL without prescribing unnecessary treatment with antibiotics.

Design

This cohort study was a retrospective analysis of consecutive cases.

Setting/Subjects

This study was conducted from August 2012 to November 2014 at 10 nationwide Japanese hospitals in which each patient's BT was measured at least once a day. We recruited patients consecutively and prospectively, who met the following inclusion criteria: age >20 years; locally advanced or metastatic cancer and treatment from palliative care specialists or admission into palliative care units; fever >38°C at least once a day, which was clinically diagnosed as tumor fever or infection; and blood test, including complete blood cell count and C-reactive protein (CRP), at <1 month before fever and during fever. The following were exclusion criteria to rule out other febrile etiologies: anti-cancer treatment such as radiation therapy or chemotherapy; medical history of febrile endocrine disease such as hyperthyroidism or adrenal deficiency; medical history of autoimmune disease; loss of consciousness or symptoms indicating increased intracranial pressure with central nervous system metastasis; clinically diagnosed drug fever; and transfusion within three days before fever onset. These exclusion criteria could rule out cancer therapy-associated fever like febrile neutropenia or tumor lysis syndrome, but the subject of this study were patients in the palliative phase.

Diagnosis of tumor fever and infection

Tumor fever was diagnosed based on the following criteria: excluding infection with symptoms, physical examinations, laboratory and imaging tests (even including no response to antibiotics), and BT <37°C throughout the day with regular NSAIDs therapy administered for ≥1 day. These criteria were based on the abovementioned Chang's criteria¹¹ and excluded smear, culture, and radiologic examinations due to feasibility and invasiveness. Specified NSAIDs were not designated because of a report revealing no difference in tumor fever response among several NSAIDs.¹⁴

Infection was diagnosed based on the following criteria: two or more of the systemic inflammatory response syndrome criteria were met (i.e., BT of >38°C or <36°C; respiratory rate [RR] >20/minute or alveolar carbon dioxide tension <32 mmHg; pulse rate [PR] >90/minute; white blood cell [WBC] count >12,000/μL or <4000/μL; or proportion of immature granulocyte >10%)¹⁵; diagnoses of one of the following three diseases from symptoms, physical examination, and laboratory or imaging findings: sepsis with positive blood culture, lower respiratory tract infection with infiltrative shadow on chest imaging or positive sputum culture, and upper urinary tract infection with positive urinary culture; and symptom improvement after antibiotic therapy. We targeted only systemic infectious diseases with clinically reliable diagnoses to exclude transient infections such as viral infection or transient inflammation associated with aspiration.

Measurements

After patient enrollment, the primary physicians retrospectively extracted the patients' data, which were determined through systematic reviews.^2,16–20 Baseline symptoms and physical findings meant those on the day before fever occurrence, and symptoms and physical findings during fever meant those on the day fever occurred. Baseline blood tests meant the most recent test performed within one month before fever occurrence, and blood tests during fever meant the first tests during the time of fever. Change from baseline data meant values obtained by subtracting continuous variables of baseline physical findings and blood tests from those during fever. Deterioration meant that categorical variables of symptoms and physical findings during fever were worse than those at baseline. Imaging findings meant findings on computed tomography (CT) obtained within one month before and after fever occurrence; all patients underwent CT within this period.

Symptoms

Pain, fatigue, and dyspnea with a Support Team Assessment Schedule—Japanese version score that is scored from 0, indicating no such symptoms, to 4, indicating severe and persistent overwhelming symptoms, along with an inability to perform any other activities.²⁰

Physical findings

Palliative performance status (PPS),²¹ which indicates performance status (PS) among advanced cancer patients and is scored from 10 to 100; Communication Capacity Scale (CCS) item 4, which indicates consciousness level and is scored from 0, indicating explicit and complex communication, to 3, indicating incoherent or not verbally responsive even when stimulated²²; shaking chills during fever; and vital signs: BT, PR, and RR. ΔPR/ΔBT was calculated by (PR during fever − PR on baseline)/(BT during fever − BT on baseline), and values >20/(minute · °C) suggest infection.²³

Blood tests

Lactate dehydrogenase (LDH) and alkaline phosphatase as indicators of the entire or liver tumor volume; and CRP, WBC, and neutrophil/lymphocyte ratio (NLR) as indicators of inflammation.

Imaging findings

Tumor necrosis findings (i.e., low-density or noncontrast area within the tumor) on CT.

Sociodemographic factors

Age, sex, primary cancer lesion, liver metastasis, which is associated with increased tumor fever occurrence,¹⁷ and admission ward (general ward or palliative care unit).

Analysis

The tumor fever and infection data were compared using Mann-Whitney's U test or Fisher's exact test, with particular attention to change from baseline data and deterioration. Then we compared data with p-values ≤0.100 in univariate analysis using logistic regression analysis. For data with significant difference in multivariate analysis, we calculated odds ratio (OR), sensitivity, specificity, and positive and negative likelihood ratio (LR), and their 95% confidence intervals (CIs) to predict tumor fever. Since NLR increased during fever compared to those at baseline in all patients, we set cutoff values of the change from baseline of NLR as 5 by sensitivity analysis. We explored useful diagnostic factors by identifying factors with LR >2 or <0.5.²⁴ Statistical analyses were performed using SPSS 19.1 software (I.B.M., Chicago), and p-values <0.05 were considered statistically significant. We did not analyze missing data.

We conducted a single-center, small-scale, retrospective study using tumor fever and infection groups of 12 patients each, and deterioration of consciousness (tumor fever, 16.7%, vs. infection, 75.0%), change from baseline HR of ≥20/minute (25.0% vs. 50.0%), change from baseline WBC count of ≥5000/μL (0% vs. 50.0%), change from baseline NLR of ≥5 (8.3% vs. 83.3%), and change from baseline CRP of ≥5 mg/dL (8.3% vs. 58.3%) were identified as significant factors.²⁵ In reference to the above study, we calculated the sample size to be 53 for both tumor fever and infection with an α score of 0.10 on both sides and a β score of 0.20. However, we could not recruit an adequate number of patients with tumor fever over the initial two years of the study or over an extended three-month period.

Ethics

This study complied with the ethical guidelines of the 1975 Declaration of Helsinki and the “Ethical guidelines for epidemiological research” by the Ministry of Health, Labor and Welfare (2008). This study was approved by the ethics boards of all participating institutions. This is a retrospective observational study, and all data were extracted using medical records from usual clinical practice, so informed consent was not obtained. This study was registered in the university hospital medical information network in Japanese (UMIN 000008896).

Results

Thirty-three patients had tumor fever and 72 had infection. Their mean age was 68.8 years (standard deviation was 12.0), median baseline PPS was 50 (range, 20–90), and the total number of male patients was 68 (64.8%). The majority of cancers were colorectal, lung, hepato-pancreato-biliary, and esophagus and stomach. There were 51 hospitalized patients in palliative care units (48.6%). Among patients with infection, 32 (44.4%) had cultures. Regarding the baseline clinical data and sociodemographic factors, only CRP before fever was significantly higher among patients with tumor fever (Table 1).

Table 1.

Univariate Analysis of Background and Baseline Data

Variable (mean, SD)	Tumor fever (N = 33)	Infection (N = 72)	p-Value	Defect (N)
Age	69.2 (15.0)	68.5 (10.8)	0.810
Sex (N, %)
Men	21 (63.6)	47 (66.2)	0.827	1
Origin of cancer (N, %)
Head and neck	2 (6.1)	0 (0.0)	0.216
Lung	6 (18.2)	14 (19.4)
Esophagus, stomach	5 (15.2)	9 (12.5)
Colorectal	5 (15.2)	17(23.6)
Hepato-pancreato-biliary	5 (15.2)	14 (19.4)
Kidney, urinary bladder, prostate	1 (3.0)	4 (5.6)
Ovary, uterus	2 (6.1)	3 (4.2)
Breast	5 (15.2)	2 (2.8)
Bone marrow, lymph node	0 (0.0)	4 (5.6)
Skin	1 (3.0)	2 (2.8)
Other	1 (3.0)	3 (4.2)
Liver metastasis (N, %)	10 (31.3)	32 (44.4)	0.206
Treatment location (N, %)
General ward	20 (60.6)	34 (47.2)	0.203
Palliative care unit	13 (39.4)	38 (52.8)
Laboratory cultures (N, %)		32 (44.4)
Symptoms
STAS-J score of pain	0.82 (0.73)	0.97 (0.84)	0.341
STAS-J score of dyspnea	0.36 (0.82)	0.36 (0.54)	0.967	2
STAS-J score of fatigue	0.94 (0.61)	0.94 (0.75)	0.976	1
Physical findings
Palliative performance status^a	47.3 (17.4)	47.2 (15.3)	0.989
Communication Capacity scale^b	0.48 (0.80)	0.39 (0.80)	0.568
Body temperature (°C)	36.70 (0.33)	36.60 (0.45)	0.288
Pulse rate (/minute)	86.6 (12.0)	83.9 (16.2)	0.350
Respiratory rate (/minute)	16.8 (3.8)	16.5 (3.1)	0.723	29
Blood tests
Lactate dehydrogenase (IU/L)	309.4 (294.7)	279.9 (171.3)	0.595	1
Alkaline phosphatase (IU/L)	583.4 (610.1)	698.6 (929.5)	0.462	5
C-reactive protein (mg/dL)	10.09 (8.78)	5.08 (4.81)	0.005	3
White blood cell (/μL)	7950.9 (3392.6)	8538.6 (5794.2)	0.519	1
Neutrophil/lymphocyte ratio^c	8.30 (7.07)	7.31 (6.58)	0.497	3

“Baseline data” means symptoms and physical findings on the day before fever occurrence and the most recent blood tests within one month before fever occurrence.

Palliative Performance Status indicates performance status among advanced cancer patients.

Communication Capacity Scale indicates consciousness level.

Neutrophil/lymphocyte ratio = neutrophil rate/lymphocyte rate.

The bold values indicate that the p-value significance is less than 0.05.

SD, standard deviation; STAS-J, Support Team Assessment Schedule—Japanese version.

Regarding clinical data during fever, change from baseline data and deterioration, and imaging findings, the significant factors in univariate analysis were deterioration of PPS, deterioration of CCS, shaking chills during fever, change from baseline LDH, change from baseline NLR, and tumor necrosis on CT (Table 2). The significant factors in multivariate analysis were deterioration of PPS (OR 0.101), shaking chills during fever (OR 0.067), and change from baseline NLR ≥5 (OR 0.122; Table 3). The factors with LR >2 or <0.5 were deterioration of PPS (positive LR 0.36 and negative LR 4.6), shaking chills (positive LR 0.095), and change from baseline NLR of ≥5 (positive LR 0.27; Table 3).

Table 2.

Univariate Analysis of Data During Fever, Change from Baseline Data, and Deterioration

Variables (mean, SD)	Tumor fever (N = 33)	Infection (N = 72)	p-Value	Defect (N)
Symptoms
Deterioration^a of STAS-J score of pain (N, %)	5 (15.2)	21 (29.2)	0.122
Deterioration^a of STAS-J score of dyspnea (N, %)	4 (12.1)	18 (26.1)	0.109	3
Deterioration^a of STAS-J score of fatigue (N, %)	17 (51.5)	48 (67.6)	0.115	1
Physical findings
Deterioration^a of PPS	10 (30.3)	61 (84.7)	<0.001
Deterioration^a of CCS (N, %)	7 (21.2)	38 (52.8)	0.002
Shaking chills during fever (N, %)	1 (3.0)	23 (31.9)	0.001
Change from baseline data^b of BT (°C)	1.8 (0.6)	2.2 (0.8)	0.069
ΔPR/ΔBT^c [/(minute · °C)]	7.0 (8.0)	10.1 (8.7)	0.097	2
Change from baseline data^b of RR (/minute)	1.4 (3.6)	2.1 (4.4)	0.483	33
Blood test
Change from baseline data^b of LDH (IU/L)	138.6 (268.7)	24.6 (105.7)	0.029	5
Change from baseline data^b of ALP (IU/L)	43.1 (252.3)	150.4 (694.6)	0.415	10
Change from baseline data^b of CRP (mg/dL)	4.18 (5.84)	6.57 (7.24)	0.119	7
Change from baseline data^b of WBC (/μL)	2809.0 (3347.7)	4323.0 (5304.2)	0.089	5
Change from baseline data^b of NLR	1.98 (5.74)	8.68 (10.76)	<0.001	12
Imaging findings
Tumor necrosis on CT^d (N, %)	10 (30.3)	3 (4.3)	0.001	3

Physical findings on the day fever developed.

The data during fever were worse than the baseline data.

Value obtained by subtracting the baseline data from data during fever.

ΔPR/ΔBT = (PR during fever − PR on baseline)/(BT during fever − BT on baseline).

Low-density area or ring-enhanced lesion within tumor at CT imaging.

The bold values indicate that the p-value significance is less than 0.05.

PPS, Palliative Performance Status; CCS, Communication Capacity Scale; BT, body temperature; PR, pulse rate; RR, respiratory rate; LDH, lactate dehydrogenase; ALP, alkaline phosphatase; CRP, C-reactive protein; WBC, white blood cell; NLR, neutrophil/lymphocyte ratio; CT, computed tomography.

Table 3.

Logistic Regression Analysis to Predict Tumor Fever and Operating Characteristic of Each Variable

Variable	Odds ratio (95% CI)	p-Value	Sensitivity (95% CI)	Specificity (95% CI)	Positive LR (95% CI)	Negative LR (95% CI)
Deterioration of PPS	0.101	<0.001	30.3	15.3	0.36	4.6
	(0.028–0.361)		(20.1–43.4)	(10.6–21.3)	(0.22–0.55)	(2.7–7.6)
Shaking chills during fever	0.067	0.018	3	68.1	0.095	1.4
	(0.09–0.518)		(0.0–8.9)	(57.3–78.8)	(0.013–0.67)	(1.2–1.7)
Change from baseline NLR ≥5	0.122	0.006	14.8	45.3	0.27	1.88
	(0.027–0.544)		(1.4–28.2)	(33.1–57.5)	(0.11–0.69)	(1.38–2.57)

Nagelkerke R² = 0.518.

Determine predictive value = 84.7%.

The bold values indicate that the p-value significance is less than 0.05 and the likelihood ratio is more than 2 or less than 0.05.

LR, likelihood ratio; CI, confidence interval.

Discussion

This is, to our knowledge, the first quantitative study exploring the factors differentiating tumor fever from infection among advanced cancer patients based on the changes in the conditions of the patients at baseline and during fever.

The most important findings were that deterioration of PPS, shaking chills during fever, and change from baseline NLR were useful to differentiate tumor fever from infection. All factors above indicated infection rather than tumor fever, and we did not identify specific findings in tumor fever.

Regarding physical examinations, shaking chills are high-degree chills and were reported to be a predictor of positive blood culture,^17,18 which was congruous with this study. On the other hand, a study reported that a high ΔPR/ΔBT value was indicative of infection,¹⁶ but that was not useful to discriminate tumor fever and infection in this study.

Deterioration of PS indicated infection rather than tumor fever. In bacterial infection, inflammatory cytokines and lipopolysaccharides can affect the brain to develop delirium.²⁶ In a review,¹⁷ tumor fever was reported to cause less worse PS or consciousness than infection, and patients with infection may appear to be more severely affected than those with tumor fever.

The blood test results indicated no significant difference between infection and tumor fever in the change from baseline CRP or WBC count, but the change from baseline NLR was significantly higher in infection. These three factors are all indicative of inflammation. Previous reports were inconclusive concerning whether CRP, change in CRP, or procalcitonin content could discriminate between tumor fever and infection.^2,14,27–29 Meanwhile, granulocyte colony-stimulating factor and granulocyte macrophage colony-stimulating factor levels in cases of tumor fever were found to be lower than those of infection,^30,31 and those cytokines were reportedly associated with increased neutrophil count.³⁰ The results of this study could reflect the difference in the quality of inflammation between WBC differentiation and CRP level or WBC count.

So far, many studies to differentiate infection and noninfection have been conducted in the setting of febrile neutropenia in hematological patients.^32–34 Advanced cancer patients are also frequently affected by tumor fever and infection, and aggressive diagnostic tests or therapies can be burden on them because of their vulnerability. Findings revealed by this study could contribute to reducing burden and improving QOL in febrile advanced cancer patients.

There were some limitations in this study. First, we used diagnostic criteria of tumor fever for research without verified validity. Although a sufficient observation period or many diagnostic tests to rule out infection might be necessary to ensure the diagnosis of tumor fever, advanced cancer patients, especially in palliative care settings, are vulnerable, tend to deteriorate in a short time period, and require rapid symptom relief. Therefore, we developed ad-hoc clinical diagnosis criteria that were as similar as possible to those of the previous study, with a shorter observation period and less invasiveness. Second, the interval of symptoms and physical findings between baseline and during fever (i.e., one day) differed from that of blood tests (i.e., within one month). This was a retrospective observational study, and we could not control the timing of blood tests. Third, we targeted only inpatients, but many advanced cancer patients who do not receive anticancer treatment are receiving home care. Future prospective studies with participation from general practitioners will reflect the real-world findings better, because most of the advanced cancer patients who do not receive anticancer treatments are not treated in the hospital. Finally, we could not recruit a sufficient sample population; a future larger study could develop classification criteria to discriminate tumor fever from infection.

Conclusions

We could discriminate tumor fever from infection among advanced cancer patients based on shaking chills, and change from baseline PS and WBC differentiation. Further confirmatory studies are needed.

Footnotes

Funding Information

No funding was received for this article.

Author Disclosure Statement

No competing financial interests exist.

References

Browder

, Huff

, Petersdorf

: The significance of fever in neoplastic disease. Ann Intern Med, 1961; 55:932–942.

Bleeker-Rovers

, Vos

, de Kleijn

, et al.: A prospective multicenter study on fever of unknown origin: The yield of a structured diagnostic protocol. Medicine, 2007; 86:26–38.

Toussaint

, Bahel-Ball

, Vekemans

, et al.: Causes of fever in cancer patients (prospective study over 477 episodes). Support Care Cancer, 2006; 14:763–769.

Chuang-Chi

, Jen-Shi

, Chen-Hsu

, et al.: Tumor fever in patients with nasopharyngeal carcinoma: Clinical experience of 67 patients. Am J Clin Oncol, 1998; 21:422–425.

Nagy-Agren

, Haley

: Management of infections in palliative care patients with advanced cancer. J Pain Symptom Manage, 2002; 24:64–70.

Reinbolt

, Shenk

, White

, Navari

: Symptomatic treatment of infections in patients with advanced cancer receiving hospice care. J Pain Symptom Manage, 2005; 30:175–182.

Lam

, Chank

, Tse

, Leung

: Retrospective analysis of antibiotic use and survival in advanced cancer patients with infections. J Pain Symptom Manage, 2005; 30:536–543.

Geisler

, Gøtzche

, Hansen

, et al.: Naproxen has greater antipyretic effect on Hodgkin's disease-related fever than on other tumours or infection. Scand J Heamatol, 1985; 35:325–328.

Chang

: How to differentiate neoplastic fever from infectious fever in patients with cancer: Usefulness of naproxen test. Heart Lung, 1987; 16:122–127.

10.

Economos

, Lucci

3rd , Richardson

, et al.: The effect of naproxen on fever in patients with advanced gynecological malignancies. Gynecol Oncol, 1995; 56:250–254.

11.

Chang

: Neoplastic fever a proposal for diagnosis. Arch Intern Med, 1989; 149:1728–1730.

12.

Alsirafy

, El Mesidy

, Abou-Elela

, et al.: Naproxen test for neoplastic fever may reduce suffering. J Palliat Med, 2011; 14:665–667.

13.

Chang

: Antipyretic effect of naproxen and corticosteroids on neoplastic fever. J Pain Symptom Manage, 1988; 3:141–144.

14.

Kallio

, Bloigu

, Surcel

, et al.: C-reactive protein and erythrocyte sedimentation rate in differential diagnosis between infections and neoplastic fever in patients with solid tumours and lymphomas. Support Care Cancer, 2001; 9:124–128.

15.

Bone

, Balk

, Cerra

, et al.: Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest, 1992; 101:1644–1655.

16.

Tsavaris

, Zinelis

, Karabelis

, et al.: A randomized trial of the effect of three non-steroid anti-inflammatory agents in ameliorating cancer-induced fever. J Intern Med, 1990; 228:451–455.

17.

Zell

, Chang

: Neoplastic fever: A neglected paraneoplastic syndrome. Support Care Cancer, 2005; 13:870–877.

18.

Tokuda

, Miyasato

, Stein

, et al.: The degree of chills for risk of bacteremia in acute febrile illness. Am J Med, 2005; 118:1417.

19.

Dasanu

, Trikudanathan

, Bandyopadhyay

, et al.: Tumor fever, paraneoplastic leukocytosis and necrotic liver metastases: A triad worth remembering. Conn Med, 2010; 74:389–391.

20.

Miyashita

, Matoba

, Sasahara

, et al.: Reliability and validity of the Japanese version of the Support Team Assessment Schedule (STAS-J). Palliat Support Care, 2004; 2:379–385.

21.

Anderson

, Drowning

, Hill

, et al.: Palliative Performance Scale (PPS); a new tool. J Palliat Care, 1996; 12:5–11.

22.

Morita

, Tsunoda

, Inoue

, et al.: Communication Capacity Scale and Agitation Distress Scale to measure the severity of delirium in terminally ill cancer patients: A validation study. Palliat Med, 2001; 15:197–206.

23.

Hamano

, Tokuda

: Changes in vital signs as predictors of bacterial infection in home care: A multi-center prospective cohort study. Postgrad Med, 2017; 129:283–287.

24.

Deeks

, Altman

: Diagnostic test 4: Likelihood ratios. BMJ, 2004; 329:168–169.

25.

Odagiri

, Morita

, Yamauchi

, et al.: Identifying factors to differentiate neoplastic fever from infection retrospectively among terminally ill cancer patients (in Japanese). Palliat Care Res, 2013; 8:273–279.

26.

Ebersoldt

, Sharshar

, Annane

: Sepsis-associated delirium. Intensive Care Med, 2007; 33:941–950.

27.

Liaw

, Huang

, Chen

, et al.: Using vital sign flow sheets can help to identify neoplastic fever and other possible causes in oncology patients: A retrospective observational study. J Pain Symptom Manage, 2010; 40:256–265.

28.

Penel

, Fournier

, Clisant

, et al.: Causes of fever and value of C-reactive protein and procalcitonin in differentiating infections from paraneoplastic fever. Support Care Cancer, 2004; 12:593–598.

29.

Yaegashi

, Izumi

, Kitagawa

, et al.: Differential diagnosis between bacterial infection and neoplastic fever in patients with advanced urological cancer: The role of procalcitonin. Int J Urol, 2014; 21:104–106.

30.

Watari

, Asano

, Shirafuji

, et al.: Serum granulocyte colony-stimulating factor levels in healthy volunteers and patients with various disorders as estimated by enzyme immunoassay. Blood, 1989; 73:117–122.

31.

Ishi

, Ohga

, Ueda

, et al.: Serum interleukin-1 and tumor necrosis factor activities in febrile children with acute leukemia. Int J Hematol, 1991; 54:79–84.

32.

Bruno

, Busca

, Vallero

, et al.: Current use and potential role of procalcitonin in the diagnostic work up and follow up of febrile neutropenia in hematological patients. Expert Rev Hematol, 2017; 10:543–550.

33.

, Wu

, Chen

, et al.: Does procalcitonin, C-reactive protein, or interleukin-6 test have a role in the diagnosis of severe infection in patients with febrile neutropenia? A systematic review and meta-analysis. Support Care Cancer, 2015; 23:2863–2872.

34.

Sakr

, Sponholz

, Tuche

, et al.: The role of procalcitonin in febrile neutropenic patients: review of the literature. Infection, 2008; 36:396–407.