Biomarkers in sepsis-looking for the Holy Grail or chasing a mirage!

Table 2 Biomarkers for diagnosis of sepsis

Ref.	Study characteristics			Results and inference
	Study type	Patient characteristics	Variables	AUC/95%CI	Sensitivity/specificity/PPV/NPV	Inference
Tan et al[5], 2019	Meta-Analysis; 9 studies	Pooled data. Total: 1368 patients. Sepsis: 495. Non sepsis: 873	CRP; PCT	0.73 (95%CI: 0.69-0.77), 0.85 (95% CI: 0.82-0.88)	Sensitivity 0.80 (95%CI: 0.63-0.90); spec: 0.61 (95%CI: 0.50-0.72) DOR: 6.89 (95%CI: 3.86-12.31); sensitivity 0.80 (95%CI: 0.69-0.87); specificity: 0.77 (95%CI: 0.60-0.88) DOR: 12.50 (95%CI: 3.65-42.80)	Diagnosis accuracy and specificity of PCT are higher than those of CRP
Thomas-Rüddel et al[9], 2018	Randomised control trial, Prospective, Secondary analysis	Gram negative vs Gram positive bacteremia and candidemia	PCT (Gram negative bacteremia)	0.72 (95%CI: 0.71-0.74)	Value was 10 ng/mL sensitivity 69%, specificity 35% for Gram negative bacteraemia	Streptococci, E. coli and other Enterobacteriaceae detected from BC were associated with three times higher PCT values. Urogenital or abdominal foci of infection were associated with twofold increased PCT
Lai et al[7], 2020	Meta-Analysis; 25 studies	GNBSI	CRP	0.85 (0.81–0.87)	Sens: 0.75 (0.56–0.87); Spec: 0.80 (0.68–0.88)	PCT was helpful in recognizing GNBSI, but the test results should be interpreted carefully with knowledge of patients' medical condition and should not serve as the only criterion for GNBSI
			PCT	0.87 (0.84–0.90)	Sens: 0.80 (0.60–0.91); Spec: 0.82 (0.72–0.89)
			IL6	0.83 (0.80-0.86)	Sens: 0.76 (0.58–0.88); Spec: 0.79 (0.71-0.85)
Zhao et al[29], 2014	Prospective; Observational, single centre	Total: 652; Sepsis: 452; Non sepsis SIRS: 200	PCT	0.803	Sens: 75.2%, Spec: 80.0%, PPV: 89.5%, NPV: 58.8%	Combination of PCT, IL6 and D-dimer enhances the diagnostic ability for sepsis and severe sepsis
			IL6	0.770	Sens: 81.0%, Spec: 61.0%, PPV: 82.4%, NPV: 58.7%
			D-Dimer	(0.737)	Sens: 79.9%, Spec: 59.0%, PPV: 81.5%, NPV: 56.5%
			PCT + IL6 + D-Dimer	0.866	Sens: 81.6%, Spec: 73.6%, PPV: 56.0%, NPV: 90.6%
Kondo et al[14], 2019	Meta-Analysis; 19 studies	Adult. Tot: 3012	Presepsin	0.87	Sens: 0.84 (95% 0.80-0.88); Spec: 0.73 (0.61-0.82)	Diagnostic accuracy of procalcitonin and presepsin in detecting infection was similar
			PCT	0.84	Sens: 0.80 (0.75-0.84); spec 0.75 (0.67-0.81)
Kang et al[16], 2019	Adult	Infected trauma: 89; Non infected trauma: 68; Healthy controls: 60	Presepsin	0.853 (0.784-0.922)	321.5 pg/mL; Sens: 67.2%; Spec: 91.9; PPV: 87.5; NPV: 78.2; LR+: 4.89; LR-: 0.39	Presepsin might be a superior biomarker for early differentiation of infection in trauma patients
			PCT	0.771 (0.682-0.859)	0.923 ng/mL; Sens: 61.1%; Spec: 88.2%; PPV: 79.1; NPV: 74.7; LR+: 5.21; LR-: 0.47
			Presepsin + ISS	0.939 (0.9-0.977)
Liu et al[15], 2013	Prospective, adult consecutive, emergency department	Total: 859; Control: 100; SIRS: 372; Sepsis: 372; Severe sepsis: 210; Septic shock: 98	Presepsin	0.820 (0.784-0.856)	317 pg/mL; Sens: 70.8%; Spec: 85.8%; PPV: 93.2%; NPV: 51.6%; LR+: 4.99; LR-: 0.34	Presepsin is a valuable biomarker for early diagnosis of sepsis. trauma stress elevates PCT, CRP, and WBCs even in the absence of infection
			PCT	0.724 (0.680 to 0.769)	0.25 ng/mL; Sens: 60%; Spec: 77.7%; PPV: 93.2%; NPV: 28.4%; LR+: 2.69; LR-: 0.51
Cong et al[20], 2021	Meta-Analysis	Adult 20 studies	CD 64	0.94 (0.91-0.96)	Sens: 0.88 (0.81-0.92); Spec: 0.88 (0.83-0.91); LR+: 7.2; LR-: 0.14; DOR-51 (25-101)	Neutrophil CD64 test has a high sensitivity and specificity in adult sepsis patients, and was superior to the traditional biomarkers PCT and IL6
			PCT	0.87 (0.83-0.89)	Sens: 0.82 (0.78-0.85); Spec-: 0.78 (0.74-0.82); LR+: 3.7; LR-: 0.23; DOR-16 (11-23)
			IL6	0.77 (0.73-0.80)	Sens: 0.72 (0.65-0.78); Spec: 0.70 (0.62-0.76); LR+: 2.4; LR-: 0.40; DOR-6 (4-9)
Gámez-Díaz et al[25], 2011	Prospective, cohort	Emergency, total 631 pts; based on expert consensus, Sepsis- 416	nCD-64	NA	Sens: 65.8% (95%CI: 61.1%-70.3%); Spec: 64.6% (95%CI: 57.8%-70.8%); LR+: 1.85 (95%CI: 1.52-2.26); LR-: 0.52 (95%CI: 0.44-0.62)	Patients suspected of having any infection in the ED, the accuracy of nCD64, sTREM1, and HMGB-1 was not significantly sensitive or specific for diagnosis of sepsis
			HMGB-1		Sens: 57.5% (95%CI: 52.7%-62.3%); Spec: 57.8% (95%CI: 51.1%-64.3%); LR+: 1.36 (95%CI: 1.14-1.63); LR-: 0.73 (95%CI: 0.62-0.86)
			s-TREM-1		Sens: 60% (95%CI: 55.2%-64.7%). Spec: 59.2% (95%CI: 52.5%-65.6%). LR+: 1.47 (95%CI: 1.22-1.76). LR-: 0.67 (95%CI: 0.57-0.79)
Yeh et al[19], 2019	Metaanalysis. 14 studies	Adult, pooled data: Total: 2471; Control: 1167; Sepsis: 1304	Neutrophilic CD 64	0.89 (0.87–0.92)	Sens: 0.87 (0.80-0.92); spec 0.89 (0.82-0.93)	Neutrophil CD64 levels are an excellent biomarker with moderate accuracy outperforming both CRP and PCT determinations
			PCT	0.84 (0.79–0.89)	Sens: 0.76 (0.61-0.86); spec 0.79 (0.70-0.86)
			CRP	0.84 (0.80–0.88)	Sens: 0.83 (0.78-0.86); spec 0.71 (0.56-0.85)
Dimoula et al[22], 2014	Prospective observational study	548 adult ICU patients. Sepsis: 103; Non sepsis: 445	nCD64	NR	230 MFI. sens: 89% (81%-94%); spec: 87% (83%-90%).	Combining CRP and nCD64 expression, an abnormal result for both was associated with a 92% probability of sepsis, whereas sepsis was ruled out with a probability of 99% if both were normal. In nonseptic patients, an increase in nCD64 expression ≥ 40 MFI predicted ICU-acquired infection (n = 29) with a sensitivity of 88% and specificity of 65%
Wang et al[23], 2021	Metaanalysis: 7 articles	Neonatal, paediatric and adults	IL27	0.88 (0.84-0.90)	Sens: 0.85 (95%CI: 0.72-0.93); Spec: 0.72 (95%CI: 0.42-0.90); DOR-15 (95%CI: 3-72)	IL27 is a reliable diagnostic biomarker for sepsis and should be evaluated with other clinical tests
Wong et al[24], 2013	Prospective	Adults, infective (n = 145) and non-infective (n = 125) critically ill	IL27	0.68 (0.62-0.75)		IL27 inferior to PCT in sepsis diagnosis
			PCT	0.84 (0.79-0.89)
Uusitalo-Seppälä et al[27], 2012	Prospective cohort	525 adult patients in emergency. Severe sepsis: 49; Sepsis: 302; SIRS: 58. Sirs with no bacterial infection: 53. Bacterial infection no SIRS: 63	PLA(2)GIIA	NA	OR: 1.48 (1.20-1.81, P < 0.001)	Differences in AUC between these parameters were not significant. On multivariate logistic regression analysis only PLA(2)GIIA could differentiate patients with severe sepsis from others (OR: 1.37, 95%CI: 1.05-1.78, P = 0.019
			BPI		OR: 2.66 (1.54-4.60, P = 0.001)
			CRP		OR: 1.35 (1.02-1.77, P = 0.036)
			WBC		OR: 2.81 (1.48-5.34, P = 0.002)
Aksaray et al[26], 2016	Prospective	ICU, Adult, Sepsis (52), SIRS (38)	STREM1	0.78 (0.69–0.86)	sTREM1 cut-off value ≥ 133 pg/mL. Sens: 71.1%; Spec: 67.33%; PPV: 80.43; NPV: 65.91	sTREM1, APACHES II higher in patients with positive culture than negative cultures. sTREM1, PCT and CRP levels, or WBC count performed equally to differentiate
			PCT	0.65 (95%CI: 0.53–0.76)	PCT cut-off value of 1.57 ng/mL. Sens: 67.31; Spec: 65.79%; PPV: 72.92; NPV: 70

AUC: Area under the receiver operator characteristic curve; BPI: Bactericidal/permeability-increasing protein; CRP: C-reactive protein; GNBSI: Gram negative blood stream infection; HMGB: High mobility group box 1; IL: ICU: Intensive care unit; Interleukin; NA: Data not available; NPV: Negative predictive value; NR: Data not reported; OR: Odds ratio; PCT: Procalcitonin; PPV: Positive predictive value; sens: Sensitivity, specificity.