Retrospective analysis of cancer patients admitted to a tertiary centre with suspected neutropenic sepsis: Are C-reactive protein and neutrophil count useful prognostic biomarkers?

Abstract

Historically, neutropenic sepsis has been associated with high mortality rates. However, there has been limited research into cancer patients admitted with suspected sepsis who are found to be non-neutropenic. C-reactive protein has been shown to be raised in cancer patients for reasons other than infection and there have been limited studies to look as its utility as a prognostic biomarker in suspected sepsis in this population. This study looked at 749 patients admitted to a tertiary cancer centre between January 2015 and February 2016 with suspected sepsis. The neutrophil count and C-reactive protein level was taken in all these patients on admission and at 72 h and compared to the primary outcome of 30-day all-cause mortality rates and hospital length of stay. There were 49 patients who died within 30 days (6.5%). Patients who died were found to have both higher neutrophil counts and C-reactive protein level on admission and at 72 h compared to survivors. Prolonged grade 4 neutropenia was shown to have higher mortality rates. There was only weak correlation between either neutrophil counts or C-reactive protein level and length of hospital stay. This study suggests that higher C-reactive protein level and neutrophil counts and prolonged grade 4 neutropenia are associated with higher mortality rates in cancer patients admitted with suspected sepsis and have utility as prognostic biomarkers in this population.

Keywords

Antibiotics cancer C-reactive protein neutropenic sepsis neutrophil

Background

Sepsis is more prevalent in patients with cancer.^1,2 Neutropenic sepsis, in particular, has high mortality rates and prolonged inpatient hospital stays.^3,4 The diagnosis of infection in this cohort is difficult as its early signs and symptoms are mimicked by non-infective causes, including the cancer itself and responses to systemic anti-cancer treatment (SACT).⁵

The white cell count is not specific in septic patients with cancer as it is influenced by SACT, the use of steroids and granulocyte colony-stimulating factor (G-CSF), and the disease itself.⁶ Patients can develop neutropenic fever without an infective source, most commonly due to the production of pro-inflammatory pyrogenic cytokines such as IL-2, IL-6 and tumour necrotic factor (TNF).⁷ Biomarkers of infection, such as C-reactive protein (CRP) and procalcitonin, are potentially useful in its diagnosis and monitoring the response to antimicrobial therapy.^8–10 Surviving Sepsis Campaign 2016 International Guidelines suggest biomarkers may be used to support shortening the duration of antimicrobial therapy in septic patients.¹¹

In cancer populations, the utility of CRP has not been widely studied and mainly focused on the neutropenic and intensive care populations.⁶ Both neutropenic and non-neutropenic patients have been shown to mount a CRP response, the course of which was not influenced by the neutrophil count.¹² A study of 314 patients showed significantly lower day 3 to 5 CRP values in survivors versus non-survivors.¹³

There are limited studies looking at length of stay as a clinical outcome to assess neutropenia or CRP. A 2-year study of 307 febrile neutropenic patients showed each day of neutropenia added 2% to the patient’s total hospital stay.¹⁴ A single hospital study of 96 COPD patients failed to find a statistical link between CRP on admission and length of hospital stay.¹⁵

It is currently not clear whether CRP levels or neutrophil counts can be used as prognostic biomarker, assessed by both overall 30-day mortality rates and overall length of hospital stay in an unselected cancer population with suspected sepsis, this study aims to address these questions.

Methods

A retrospective analysis was performed at a specialist oncology hospital in the North West of England from 1 January 2015 to 31 January 2016. The hospital has a medical admissions unit (MAU),¹⁶ which receives patients via a number of access points including directly from the hospital patient hotline, outpatient clinics and as tertiary referrals from other hospitals.

Basic demographic data were collected for each patient who presented with suspected sepsis (based on SIRS criteria in operation at the time),¹⁷ alongside site of primary cancer, neutrophil count and CRP level on admission and at 72 h (or the result on discharge if the patient was discharged within 72 h). The primary outcome was overall 30-day mortality. Secondary analysis was performed on the length of hospital stay.

The data were analysed using GraphPad Prism 7. The p value for statistical significance was 0.05 and was tested using student’s t test. Correlation coefficients were used to compare the neutrophil count and CRP level at admission and at 72 h to create gradients of the effect on length of hospital stay.

Results

During the study period 749 patients presented to the MAU with suspected sepsis. The median age of patient was 60 (range 16 to 87). The most common malignancies presenting with sepsis were haematological (n = 219; 29.1%), colorectal (n = 107; 14.3%) and breast (n = 93; 12.4%). The demographic data, including site of primary tumour, and the neutrophil and CRP values at admission and 72 h are shown in Table 1. 49 (6.5%) patients died within 30 days of admission.

Table 1.

Patient demographics and neutrophil and CRP levels.

	Number	Mean (range)	Percentage
Age	749	54.6 (27–84)
Cancer type
Gynaecological	47		6.3
Breast	93		12.4
Upper gastrointestinal	45		6
Hepatopancreatobiliary	40		5.3
Lower gastrointestinal	107		14.3
Urological	58		7.7
Haematological	218		29.1
Melanoma	18		2.4
Sarcoma/germ cell	46		6.1
Lung	50		6.7
Central nervous system	12		1.6
Unknown primary	11		1.5
Neuroendocrine	4		0.5
30-day mortality	49		6.5
CRP on admission (ng/dl)		98.1 (0–499)
CRP at 72 h (ng/dl)		94.8 (0–499)
Neutrophil count on admission (10⁹/L)		5.5 (0–67.5)
Neutrophil count at 72 h (10⁹/L)		4.6 (0–65.4)
Length of hospital stay (days)		7.8 (1–78)

CRP: C-reactive protein.

The mean neutrophil count on admission of the 49 patients who died was 9.6 (95%CI = ±3.13), compared to 5.23 (95%CI = ±0.4) for the 700 patients who survived (p ≤ 0.0001). At 72 h patients that died within 30 days had significantly higher neutrophil counts than survivors, 9.4 (95%CI = ±3.0) vs 4.21 (95%CI = ±0.34) (p ≤ 0.0001) (see Table 2 and 3).

Table 2.

Thirty-day mortality rates with neutrophil count split into grades of neutropenia on admission and at 72 h/discharge.

Time	Neutrophil count (grade)	Total pts	Pts with 30-day mortality	% Mortality	Range LOS (days)	Mean LOS (days)
On admission	0–0.5 (4)	112	5	4.5	1–49	6.6
	0.51–1 (3)	43	3	7	1–39	8
	1.01–1.5 (2)	32	1	3.1	1–18	4.9
	1.51–2 (1)	43	1	2.3	1–27	5.4
	Non-neutro.	519	39	7.5	1–78	8.4
At 72 h/DC	0–0.5 (4)	66	7	10.6	1–49	8.9
	0.51–1 (3)	60	1	1.7	1–73	6.7
	1.01–1.5 (2)	86	1	1.2	1–26	5.3
	1.51–2 (1)	60	2	3.3	1–48	6.7
	Non-neutro.	477	38	8	1–78	8.3

DC: discharge; LOS: length of stay.

The mean CRP level on admission of the 49 patients who died was 135.8 (95%CI = ±22.2), compared to 95.8 (95%CI = ±5.9) for the 700 patients who survived (p = 0.0004). At 72 h patients that died within 30 days had significantly higher CRP levels than survivors, 143.2 (95%CI = ±25.1) vs 87.1 (95%CI = ±5.6) (p ≤ 0.0001).

Table 3.

Split of CRP on admission and at 72 h/discharge compared to 30-day mortality rate and length of stay.

Time	CRP	No. of pts	Mortality	% Mortality	Range LOS (days)	Mean LOS (days)
On admission	0–25	147	1	0.7%	1–64	5.1
	26–75	202	10	5.0%	1–76	7.1
	76–125	133	10	7.5%	1–71	7.7
	126+	254	26	10.2%	1–78	10
At 72 h/DC	0–25	159	1	0.6%	1–64	5.2
	26–75	219	10	4.6%	1–55	6.5
	76–125	157	9	5.7%	1–76	8.5
	126+	211	29	13.7%	1–78	10.4

CRP: C-reactive protein.

The overall 30-day mortality rates increase in patients with higher CRP levels on admission and at 72 h (see Table 3). The mortality rate of patients who had a reduction in CRP between admission level and 72 h level was 5.5% compared to 10.3% in those patients with an increasing CRP level over the 72 h. Two patients died with a CRP of less than 25, one from progressive small cell lung cancer and one from pneumocystis jiroveci pneumonia.

At initial presentation, neutrophil count and CRP levels showed poor correlation. At 72 h, patients with neutrophilia (>8 × 10⁹/L) and grade 4 neutropenia (<0.5 × 10⁹/L) had significantly raised CRP levels (see Figure 1).

Figure 1.

Box plots assessing CRP level and neutrophil count on admission and at 72 h. CRP: C-reactive protein.

Both on admission and at 72 h, CRP levels and neutrophil counts had very weak correlation with length of hospital stay (length of stay compared to: CRP OA, r = 0.16; CRP at 72 h, r = 0.19; neutrophils OA, r = 0.05; neutrophils at 72 h, r = 0.13). The correlation worsens when looking at the change in both CRP and neutrophil over 72 h (r = 0.05 and r = 0.09, respectively) (see Figures 2 and 3).

Figure 2.

Scatter graphs of neutrophil count on admission and at 72 h against length of hospital stay in days.

Figure 3.

Scatter graphs of CRP on admission and at 72 h against length of hospital stay in days. CRP: C-reactive protein.

Discussion

This study suggests that the CRP level and neutrophil counts could be useful biomarkers in predicting outcomes in cancer patients presenting with signs and symptoms suggestive of sepsis. Both CRP level and neutrophil counts were significantly higher in patients with 30-day mortality but neither was predictive of a longer inpatient stay.

The identification of infection in patients with cancer is a key clinical challenge. Numerous biomarkers, such as CRP, procalcitonin, IL-6 and IL-8, have been studied with regards to diagnosis of infection, identification of infective agents and sepsis outcome prediction, particularly in the neutropenic population.^{6,10,11,18–20} Procalcitonin concentrations in septic patients with cancer are lower in those with leucopenia.²¹ Furthermore, infection has been confirmed in neutropenic patients with procalcitonin levels <0.5 ng/ml.²² In our study, a patient with a CRP <25 ng/dl died of an infective illness.

This study almost demonstrates an increasingly noted trend of lower mortality rates in patients with neutropenic sepsis compared to non-neutropenic sepsis in cancer patients. This may reflect that high mortality rates in neutropenic patients reflect a different era of both sepsis management and chemotherapy prescribing. This study adds to the evidence that with appropriate patient selection for myelosuppressive chemotherapy regimens and high quality care of patients with suspected neutropenic sepsis, mortality rates should be less than 5%. Patients at the highest risk of mortality include those with ongoing grade 4 neutropenia at 72 h, persistent neutrophilia or those with a rising CRP at 72 h.

This study is limited by being a retrospective single centre analysis. We also only studied neutrophil and CRP levels for the first 72 h. It is also difficult to determine definitively if the patients died from sepsis or progressive cancer but this reflects the diagnostic challenge in this population. Due to use of retrospective data, this study hasn’t analysed disease severity or accounted for other confounders such as co-morbidities and disease severity. As of 2016, SIRS criteria were replaced by the quick sequential organ failure assessment (qSOFA) after it was shown to be more prognostic for in hospital mortality,²³ however, using our selected population allows comparison with previous studies. This study is strengthened by being the largest single centre analysis of cancer patients with suspected sepsis in the UK and the fact that 96.4% of patients presenting with suspected sepsis to the OAU receive their first dose of intravenous antibiotics within 1 h through a previously described nurse-led protocol.²⁴ Delays in administration of antibiotics in septic patients with cancer have been shown to affect both mortality and length of stay.^25,26

Conclusion

Increasing CRP, neutrophilia or 72 h of sustained neutropenia appear to be useful independent prognostic biomarkers of 30-day mortality in cancer patients with suspected sepsis. Additional prospective research of these biomarkers analysing for confounders such as disease severity, comorbidities and qSOFA scores would add further clarity to their utility in cancer patients.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

National Chemotherapy Advisory Group, http://webarchive.nationalarchives.gov.uk/20130104173757/http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/DH_104500 (2009, accessed 21 November 2017).

Cooksley

Rice

. Emergency oncology: Development, current position and future direction in the US and UK. J Support Care Cancer 2017; 25: 3–7.

Klastersky

Paesmans

. The Multinational Association for Supportive Care in Cancer (MASCC) risk score: 10 years of use for identifying low risk neutropenic cancer patients. Support Care Cancer 2013; 21: 1487–1495.

Cooksley

Holland

Klastersky

. Ambulatory outpatient management of patients with low risk febrile neutropaenia. Acute Med 2015; 14: 178–181.

Loizidou

Aoun

Klastersky

. Fever of unknown origin in cancer patients. Crit Rev Oncol Hematol 2016; 101: 125–130.

Povoa

Souza-Dantas

Soares

et al.

C-reactive protein in critically ill cancer patients with sepsis: influence of neutropenia. Crit Care 2011; 15: R129.

Bisoendial

. C-Reactive protein elicits white blood cell activation in humans. Am J Med 2009; 122: 582.e1–e9.

Reinhart

Bauer

Reidemann

et al.

New approaches to sepsis: Molecular diagnostics and biomarkers. Clin Microbiol Rev 2012; 25: 609–634.

Shomali

Hachem

Chaftari

et al.

Can procalcitonin distinguish infectious fever from tumor-related fever in non-neutropenic cancer patients?

Cancer 2012; 118: 5823–5829.

10.

Allin

Nordestgaard

. Elevate C-reactive protein in the diagnosis, prognosis and cause of cancer. Crit Rev Clin Lab Sci 2011; 48: 155–170.

11.

Rhodes

Evans

Alhazzani

. Surviving sepsis campaign: International guidelines for management of sepsis and septic shock: 2016. Crit Care Med 2017; 45: 486–552.

12.

Póvoa

Ces Souza-Dantas

Soares

et al.

C-reactive protein in critically ill cancer patients with sepsis: influence of neutropenia. Crit Care 2011; 15: R129.

13.

Devran

Karakurt

Adıgüzel

et al.

C-reactive protein as a predictor of mortality in patients affected with severe sepsis in intensive care unit. Multidiscip Respir Med 2012; 7: 47.

14.

Rosa

Goldani

. Factors associated with hospital length of stay among cancer patients with febrile neutropenia. PLoS One 2014; 9: e108969.

15.

Lake

Patel

Hurst

. Co-morbidities, CRP and length of hospital stay at exacerbation of COPD. Am J Respir Crit Care Med 2010; 181: A1502.

16.

Rooney

Moloney

Bennett

et al.

Impact of an acute medical admission unit on hospital mortality: a 5 year prospective study. QJM: Int J Med 2008; 101: 457–465.

17.

Levy

Fink

Marshall

et al.

International sepsis definitions conference. 2001 SCCM/ESICM/ACCP/ATS/SIS International sepsis definitions conference. Intensive Care Med 2003; 29: 530–538.

18.

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.

19.

Miedema

de Bont

Elferink

et al.

The diagnostic value of CRP, IL-8, PCT and sTREM-1 in the detection of bacterial infections in pediatric oncology patients with febrile neutropenia. Support Care Cancer 2011; 19: 1593–1600.

20.

Lima

Nobre

de Castro Romanelli

et al.

Procalcitonin-guided protocol is not useful to manage antibiotic therapy in febrile neutropenia: a randomized controlled trial. Ann Hematol 2016; 95: 1169–1175.

21.

Schuttrumpf

Binder

Hagemann

et al.

Utility of procalcitonin concentration in the evaluation of patients with malignant diseases and elevated C-reactive protein plasma concentrations. Clin Infect Dis 2006; 43: 468–473.

22.

Sakr

Sponholz

Tuche

et al.

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

23.

Singer

Deutschman

Seymour

et al.

The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA 2016; 315: 801–810.

24.

Mattison

Bilney

Haji-Michael

et al.

A nurse-led protocol improves the time to first dose intravenous antibiotics in septic patients post chemotherapy. J Support Care Cancer 2016; 24: 5001–5005.

25.

Rosa

Goldani

. Cohort study of the impact of time to antibiotic administration on mortality in patients with febrile neutropenia. Antimicrob Agents Chemother 2014; 58: 3799–3803.

26.

Perron

Emara

Ahmed

. Time to antibiotics and outcomes in cancer patients with febrile neutropenia. BMC Health Serv Res 2014; 14: 162.