Transcript Poster Ascite Diafir 2016 03 30 CCx

Mid-infrared spectroscopy has a high sensitivity and specificity for Point-Of-Care
diagnosis of spontaneous bacterial peritonitis
Anty R1,2,3, Le Corvec M4, Renou C5, Raoult A5, Nousbaum JB6, Tanne F6, Payan JC6, Triboulot C7, Mari S7, Hastier P8, Fissore-Magdelein C8, Peritore ML9,BlancAmrane V9, Marine-Barjoan E11, Vibert J12, Lemichez E2,3, Sire O4, Gual P2,3, Tariel H4, Tran A1,2,3, Ait-Oufella H13,Loréal O14, Landraud L15
1 Hepatology Unit, University Hospital, Nice; 3 University of Nice-Sophia; 3 INSERM U1065, Nice ; 4 DIAFIR, Rennes ; 5 Hospital of Hyeres; 6 University hospital Brest ; 7 Biological unit, university hospital Nice ; 8 Princesse Grace Hospital, Monaco ; 9
Antibes Hospital ; 10 Public Health department, University Hospital of Nice ; 12 Mougins ; 13 Service de réanimation médicale, AP-HP, Hôpital Saint- Antoine, INSERM UMR 991, Inserm U970, Paris Research Cardiovascular Center, Université Pierre et Marie
Curie, Paris ; 14 INSERM UMR 991, University of Rennes 1, 15 UMR 1137 INSERM, Universités Paris Diderot et Paris Nord, Sorbonne Paris Cité, Louis Mourier Hospital, Paris, FRANCE
ACTUALIZED ABSTRACT
BACKGROUND
•Spontaneous bacterial peritonitis (SBP) is a complication of cirrhotic patients
with ascites associated with a high mortality rate
•A treatment started as soon as possible is of great interest
•The availability without delay and 24/24 of cyto-bacteriology analysis of the
ascites is not present in all medical centers.
•A Point-Of-Care diagnosis is relevant for this disease
•Mid infrared (MIR) spectroscopy allows global metabolic evaluation of a
biological sample. The principle is based on the property of chemical bonds
within molecules to vibrate within the MIR region and thus to generate strong
absorption bands in this area. Thus, the use of MIR spectroscopy has been
proposed to analyse biological samples, including microbiologic studies.
•The development of chalcogenide glass optical fibers allows now to collect a
remote MIR signal in real time, based on the concept of Fiber Evanescent
Wave Spectroscopy (FEWS).
•FEWS has been recently tested for the fast diagnosis of septic arthritis (1).
METHODS (1)
Proteins
Nucleic
acids
Amino acid
Fatty acid
Lipids
Phospholipids
polysaccharides
Proteins
Calibration set of the cohort from Paris (n=90)
Results of
the model
7µl of ascitis
Diafir SPIDTM FT-IR spectrometer
(Rennes, France).
Cytology
PMN>250
PMN<250
PMN>250
13
14
PMN<250
4
59
Results of
the model
Threshold at 0.2
Sensitivity=76%
Specificity=81%
PPV=48%
NPV=94%
AUROC=0.85
Wavelength (mm)
mid-infrared spectrum
Algorithm
Ascites without
infection
Validation set of the cohort from Paris (n=45)
Cytology
SBP
PMN>250
PMN<250
PMN>250
5
2
PMN<250
4
34
Threshold at 0.2
Sensitivity=55%
Specificity=94%
PPV=71%
NPV=89%
AUROC=0.86
Step 2: Calibration on the entire cohort of Paris and external cohort
validation
PATIENTS
Calibration (cohort of Paris, n=135)
External validation (cohort of Nice, n=126)
Cytology
A cohort (of elaboration) of 135 patients from Paris (France) including 26 SBP was
used to construct a model.
An external cohort (of validation) of 126 patients from Nice (France) including 21 with
SBP was used to valide the model.
The diagnosis of SBP was done if PMN ≥ 250/mm3.
In both cohort, the ascites were collected at the time of suspicion of SBP, a cytobacteriological analysis was done, then the ascites were frozen at -80°c.
For each cohort, all samples were studied at the same time by MIR spectroscopy.
METHODS (2)
Results of PMN>250
the model PMN<250
Cytology
PMN>250
PMN<250
22
18
4
91
Results of
the model
Threshold at 0.2
Sensitivity=85%
Specificity=83%
PPV=55%
NPV=96%
AUROC=0.88
PMN>250
PMN<250
PMN>250
5
2
PMN<250
4
34
Threshold at 0.2
Sensitivity=76%
Specificity=87%
PPV=53%
NPV=95%
AUROC=0.91
*
Background and aims:
•A highly sensitive and specific Point-Of-Care (POC) diagnostic method usable at the bedside is lacking to diagnose
spontaneous bacterial peritonitis (SBP).
•The aim of this study was to test the diagnostic value of an easy-to-use, new infrared spectroscopic POC diagnostic method for
the suspicion of SBP.
Patients and Methods:
•Ascites from a calibration cohort of 135 cirrhotic patients were collected, in the university hospital of Paris France, from
February to October 2014. Patients were included in case of suspicion of SBP or for planned ascites evacuation in case of
refractory ascites. Sets of calibration and validation groups of patients were done to test the sensitivity (Se), specificity (Sp),
positive predictive value (PPV), negative predictive value (NPV) and area under the receiver operating characteristic curve
(AUROC).
•An external cohort of validation with 126 ascites from patients from Nice France was constituted.
•SBP was defined by the presence of neutrophil polymorph-nuclear cell ≥ 250/mm3 in the ascites.
•Measures were done using mid-infrared spectroscopy using LS23 chalcogenide glass fiberoptic sensors. The mid-infrared
spectra were recorded from 7µl ascitic fluid using a Diafir SPIDTM FT-IR spectrometer (Rennes, France). The concept is that a
part of the light at the fiber surface can be absorbed by element in contact with the fiber. Biological elements have typical
absorption bands in the mid infrared area. When a biological sample is in contact with the fiber, a spectrum reflecting the
sample’s molecular composition, providing a metabolic fingerprint that can be collected.
Results:
•The calibration cohort included 135 patients (male gender in 72%) with a median age of 58 [11] years. Main causes of cirrhosis
were viral hepatitis (21%), alcohol (60%), viral hepatitis + alcohol (11%), Haemochromatosis (1%) and others (18%). 26 patients
had a SBP and 109 were without.
•The external cohort of validation included 19 patients with SBP and 105 without SBP. The 2 cohort were similar regarding the
clinico-biological characteristics.
•At the threshold 0.2, Se was 85%, Sp was 83%, PPV was 55%, and NPV was 96% in the calibration cohort (AUROC=0.88). At
the threshold 0.2 , Se was 76%, Sp was 87%, PPV was 53%, and NPV was 95% (AUROC=0.91) in the validation cohort.
Conclusion:
•Mid-infrared spectroscopy is a new diagnostic POC method, which offers a high sensitivity and specificity for the diagnosis of
SBP at bedside.
Step 1: Model constructed in the cohort of Paris
Step 2: Box-plots for the prediction of SBP
Step 2: ROC curves for the prediction of SBP
Step 1, elaboration of a model : construction of a model using a set of calibration (n=90)
and a set of validation (n=45) in the cohort of Paris
Step 2, validation of the model : based on variables selected during the step 1, the model
was parametered on the entire cohort of Paris and then validated on the external cohort of
Nice (n=126)
RESULTS
Baseline characteristics
Test cohort
Validation cohort
135
126
26 (19.3)
21 (16.6)
78
90 (71.4)
58 [11]
60 [51-68.25]
Alcohol
60
70,6
Alcohol + viral hepatitis (B or C)
11
13.5
Viral hepatitis
21
4
SOFTWARE
Haemochromatosis
1
0
Operator
independant
no spectroscopic
knowledge required
Others
18
8.7
A
4
0.8
B
30
42.9
C
66
55.5
Prothrombin time (%), median [range]
51 [37-67]
57 [47.8-72.3]
Serum bilirubin (μmol/L), median [range]
57 [30-139]
43.5 [20-92.3]
Albumin (g/L), median [range]
26 [22-29]
28.15 [25-31.5]
107000 [67000-147000]
117000 [77000-181500]
97 [67-151]
70 [57-113.5]
Ascites samples (n)
SBP, n (%)
Male sex, n (%)
Age (years), mean [SD]
Repartition of patients according to
the threshold 0.2, in the calibration
and validation cohorts
Cohort of elaboration from Paris (red)
Cohort of validation from Nice (blue)
CONCLUSIONS
Causes of cirrhosis, %
Fiberoptic
evanescent wave
spectroscopy
(FEWS) SENSOR
7µl / measure
label free
single use
SPECTROSCOPE
Result in 10 mn
Compact
Intuitive use
OBJECTIVES
The aim of this study was to test the diagnostic value of an
easy-to-use, new infrared spectroscopic POC diagnostic
method for the suspicion of SBP.
Child-Pugh score, %
Platelet count (109/L), median [range]
Serum creatinine (μmol/L), median [range]
•Mid-infrared spectroscopy has a high sensitivity
and specificity for point-of-care diagnosis of
spontaneous bacterial peritonitis in 2 independent
cohorts.
•This technology offers new perspectives for
point-of-care diagnosis in various domains.
REFERENCES
Albert et al. Joint Bone Spine 2016