Early signaling of inflammatory bursts by using estimated C-reactive Protein velocity (eCRPv): a model of acute meningitis and iso-CRPinemia

Abstract

Purpose

C-Reactive Protein (CRP) is often used to support the differentiation between acute bacterial and viral infections, including acute meningitis. However, this is not always possible due to similar CRP concentrations upon presentation in these two cohorts. We have improved the diagnostic yield of this biomarker by adjusting its concentrations to the estimated time from disease onset, transforming a single, static value into a dynamic measurement, expressed in milligrams per liter per hour.

Methods

Our study included 46 matched pairs of patients with acute bacterial and acute viral meningitis who were presented to the hospital with similar CRP concentrations (± 10%). Estimated CRP velocity (eCRPv) was calculated by dividing the first CRP measurement at the Emergency department (ED) by the estimated time of disease onset (iso-CRPinemic model).

Results

There was negligible difference (p = 0.4) between the mean ± s.d. CRP concentrations in the bacterial (110.1 ± 66.9 mg/L) versus the viral (110.0 ± 68.4 mg/L) group. However, when we adjusted these concentrations to the time from onset of the clinical picture of acute meningitis, it served to differentiate well (p < 0.001) between the groups.

Conclusion

The diagnostic yield of CRP at presentation amongst patients with acute meningitis can be improved by adjusting the value to the estimated time from disease onset.

Background

Delayed diagnosis of infections is known to negatively impact morbidity and mortality, making early detection of disease crucial to clinical practice. Acute meningitis is one such example that presents grave prognostic implications if not diagnosed and treated during the early stage of the disease. It has been repeatedly shown that C - reactive protein (CRP) values may help in differentiating between viral and bacterial infections [1–5], including in patients with acute meningitis [6]. However, no definitive CRP cut-off value has been established to reliably distinguish between bacterial and viral etiologies.

However, patients presenting with a severe inflammatory response can often produce similar CRP values with those suffering from a milder inflammatory process, complicating the differentiation between these two cohorts.

Therefore, we suggest taking into consideration the estimated time between disease onset and presentation to the Emergency Room. We hypothesize that significant inflammatory bursts are associated with a more rapid increase of the inflammatory biomarker [7].

To quantify the intensity of the inflammatory burst in relation to the time of disease onset, we have recently introduced a new concept of correcting the value of a universal inflammatory biomarker such as the C - reactive protein (CRP) [8] to time from onset of signs and symptoms [9, 10]. This is calculated by dividing the CRP values (in mg per liter) by the assumed time (in hours) elapsed from disease onset thus adding kinetic information to a single test that is taken upon presentation of the patient. It has yet to be demonstrated that patients who present with high CRP velocity [11] are indeed those who really present a more serious medical condition.

Acute meningitis is a suitable model to test the potential value of the aforementioned kinetic model. In fact, both bacteria and viruses can cause inflammation of the same organ thus permitting a comparative analysis between both groups that present similar concentrations of the biomarker upon presentation (the iso-CRPinemic model).

Methods

Inclusion criteria

Patients that were diagnosed with acute bacterial or viral meningitis. The diagnosis of bacterial meningitis was based on the presentation of typical signs and symptoms, compatible laboratory findings in their CSF, including polymorphonuclear (PMN) leukocytosis (> 100 cells/µL), decreased glucose concentration (< 40 mg/dL) and/or CSF/serum glucose ratio of < 0.4, increased protein concentration (> 0.45 g/L)] and if present, positive cultures in blood or CSF. Patients were diagnosed as having viral meningitis if they presented with typical signs and symptoms suggestive of acute viral meningitis such as fever and headache, had CSF results consistent with viral meningitis [pleocytosis, a normal or slightly elevated protein concentration in the range of 20–80 mg/dL and normal glucose concentration (> 50% of blood value)] and if available, positive serology or PCR that identified the causative viral agent.

Exclusion criteria

Excluded from the present study were patients who were taking immunosuppressive therapy and/or anti-inflammatory medications. Patients in whom the disease was related to neurosurgery or who were referred to our medical center from other hospitals were not included. Patients with an unclear record of disease onset were excluded from the study as well.

Laboratory methods

Measurement of CRP

We have used the Wide-Range C-Reactive Protein (wrCRP) [12]. The measurements were performed by ADVIA Chemistry, Siemens Healthcare Diagnostics Inc., Tarrytown, NY 10,591 − 5097 USA. The ADVIA^® Chemistry wrCRP method measures CRP in serum and plasma by a latex-enhanced immunoturbidimetric assay.

Estimated CRP velocity

Estimated CRP velocity (eCRPv) was calculated by dividing the concentration (mg per liter) of the first CRP measurement that was obtained on presentation to the Emergency Department by the estimated time of disease onset as previously described [9, 10]. This created a biomarker expressed in mg per liter per hour (mg/L/hour).

Estimated time of disease onset was taken from the medical files. This information was provided directly from the patient, a caregiver, or a family member at presentation.

We describe three estimated times (in hours) from disease onset as follows:

1. Estimated time from any relevant sign or symptom that heralded the infectious event including acute otitis media, upper respiratory tract infection, general malaise, fatigue, myalgia etc. – mode 1.

2. Estimated time from fever onset – mode 2.

3. Estimated time from meningitis onset (severe headache, decreased level of consciousness, confusion, lethargy, seizures or coma) – mode 3.

Since there were three different times of estimation we calculated three different velocities: eCRPv1 (from any symptomatology heralding the disease), eCRPv2 (from fever onset) and eCRPv3 (from meningitis onset).

The iso-CRPinemic model

We compared all the CRP values obtained in both groups of patients with bacterial or viral meningitis and could find only 46 pairs of patients that presented with CRP concentrations that did not differ by more than 10% in the large cohort (cohort A) and 32 pairs in the limited cohort (Cohort B), that contained only patients in whom a causative organism was isolated, whether from a positive culture, or a positive PCR or serology.

Microbiological methods

CSF and blood culture samples were processed at the TASMC microbiological laboratory according to the ASM guidelines [13]. Bacterial identification was performed using the VITEK-MS^® system (bioMerieux, Marcy l`Etoile France).

Nucleic acid extraction from CSF samples was carried out by the easyMAG^® system (bioMerieux, Marcy l`Etoile France). Detection of viral agents by PCR was done using the Xpert^® EV (Cephid^®) PCR (Enterovirus) or the RealStar^® alpha Herpesvirus PCR (HSV1, HSV2 and VZV) assays. Since June 2016, CSF samples were also tested by the BioFire FilmArray meningitis/encephalitis (ME) assay. WNV infection was diagnosed by the presence of IgM in the CSF using the Focus WNV IgM capture ELISA kit.

Statistical analysis

Categorical variables were summarized as frequency and percentage. Continuous variable distribution was evaluated using histogram. Since not all variables were normally distributed, they were reported as median (interquartile range (IQR)). The McNemar test was used to compare categorical variables between the matched groups, while Wilcoxon test was used to compare the continuous variable. All statistical tests were two sided and p < 0.05 was considered as statistically significant. Multivariable analysis was not applied due to the limited number of patients in each cohort. SPSS software was used for all statistical analysis (IBM statistics for windows, version 29, Armonk, New York, USA, 2022).

Study design and setting

We conducted a retrospective cohort study, analyzing all patients admitted to the Emergency Department in Tel Aviv Sourasky Medical Center, Israel, between July 2nd 2007 and 24th February 2024 who had an admission CRP test as well as a clinical and laboratory diagnosis of acute meningitis. We used the MDClone (mdclone.com) system [14] in order to extract information from patients with a diagnosis of acute bacterial or viral meningitis. The medical records were then reviewed by a resident of Internal Medicine (NSM) as well as a specialist in infectious diseases (AW). The study was reviewed and approved by the Institutional Review Board according to the declaration of Helsinki (TLV-0645-23). Informed consent by the patients was waived due to the retrospective nature of our study.

Patients

From the MDClone system we could extract 1326 patients with the diagnosis of CNS infection (962 of viral and 364 of bacterial origin), who were admitted to the department of Emergency Medicine during the time period of the study. Only 60 patients were finally included in the bacterial group and met the criteria of acute bacterial meningitis. The other 304 patients were excluded from the bacterial group due to the following reasons: 136 patients had an inconclusive diagnosis; 89 patients were following a recent neurosurgical intervention; 40 patients had a viral diagnosis and not a bacterial one; 22 patients had insufficient information in their medical files regarding the time of onset of the disease and 17 patients were on immunosuppressive therapy.

Following the definition of 60 patients with bacterial meningitis, we have matched the 60 patients with acute bacterial meningitis with patients with viral meningitis by creating pairs of patients, having a similar CRP concentration upon admission, using the IsoCRPinemic model. The process of matching was carried out by screening the group of 962 patients who had a diagnosis of viral meningitis in the MDclon system. During this process of matching, we were able to recruit as many as 46 pairs (Cohort A) and 32 pairs (Cohort B). The Demographic and clinical characteristics of both cohorts are now reported in Tables 1 and 2. The bacterial and viral causative agents are reported in Tables 3 and 4 respectively. The detailed information on the cerebrospinal fluid findings of the patients included in the study, appears in Table 1, see in the Appendix. As shown, in 3 of the bacterial patients and in 21 of the viral patients, no pathogen was isolated. In these patients, the distinction between viral and bacterial etiology was made based on the CSF findings. In order to avoid inaccuracies, we created an additional and limited cohort (Cohort B) that contained only patients in whom a causative organism was isolated, whether from a positive culture, or a positive PCR or serology. In this cohort, we were able to pair 32 patients from the bacterial and viral groups.

Table 1.

Demographic and clinical characteristics of pairs of patients with bacterial as opposed to viral acute meningitis - Cohort A

Pairs n.	Age		Gender		Charlson score		CRP values (mg/L)		Causative pathogen
	Bacterial	Viral	Bacterial	Viral	Bacterial	Viral	Bacterial	Viral	Bacterial	Viral
1	70	37	male	female	4	2	254	263	S. pneumoniae	-
2	64	36	male	male	3	0	228	229	S. pneumoniae	-
3	62	45	female	female	4	0	197	193	S. pneumoniae	-
4	78	40	male	male	3	-	192	191	S. pneumoniae	-
5	89	33	female	male	9	0	176	179	S. pneumoniae	-
6	56	35	female	female	3	0	172	177	-	-
7	55	25	female	male	2	-	159	161	S. pneumoniae	-
8	77	35	female	male	3	0	159	160	S. pneumoniae	-
9	48	44	female	male	0	0	148	152	S. pneumoniae	EV
10	63	26	female	male	2	0	147	149	-	-
11	60	49	female	male	10	0	146	148	S. pneumoniae	-
12	86	35	female	female	5	0	140	139	S. pneumoniae	-
13	69	28	male	female	6	0	138	127	S. pneumoniae	-
14	33	28	female	female	0	0	137	138	N. meningitidis	-
15	42	53	male	male	0	1	141	144	N. meningitidis	-
16	46	37	male	male	0	0	124	117	S. pneumoniae	-
17	55	34	female	male	1	0	114	112	S. epidermidis	-
18	61	19	male	male	2	1	106	110	S. pneumoniae	-
19	69	28	female	male	4	0	89	92	S. epidermidis	EV
20	55	30	female	female	1	0	88	86	S. pneumoniae	-
21	61	29	female	male	5	-	44	83	S. pneumoniae	-
22	62	34	female	male	2	0	78	66	S. pneumoniae	HSV
23	72	34	female	female	3	0	78	77	S. pneumoniae	WNV
24	51	40	female	male	1	0	62	61	S. pneumoniae	EV
25	87	38	female	male	6	0	57	55	S. pneumoniae	-
26	55	35	female	female	1	0	50	50	S. pneumoniae	EV
27	83	38	female	male	3	0	42	35	L. monocytogenes	EV
28	76	29	female	female	4	0	39	43	Haemophilus influenzae	EV
29	39	32	male	female	0	0	37	34	Haemophilus influenzae	EV
30	48	23	female	female	3	0	35	35	Streptococcus pneumoniae	EV
31	53	38	female	male	1	0	32	32	Streptococcus pneumoniae	EV
32	44	27	female	female	4	0	32	31	viridans group streptococci	EV
33	53	31	female	female	3	0	27	27	S. pneumoniae	EV
34	55	30	female	female	1	0	25	26	S. pneumoniae	-
35	67	37	female	male	3	0	25	24	S. pneumoniae	-
36	59	35	male	male	3	0	23	23	L. monocytogenes	EV
37	68	34	female	male	2	0	22	22	S. pneumoniae	EV
38	76	40	male	female	4	0	21	21	S. pneumoniae	EV
39	77	61	male	male	5	2	19	20	viridans group streptococci	EV
40	71	31	male	male	3	0	19	18	S. pneumoniae	EV
41	36	63	female	female	0	2	16	16	S. pneumoniae	EV
42	64	23	male	female	3	0	11	11	L. monocytogenes	VZV
43	26	39	female	female	0	0	9	9	-	VZV
44	51	32	female	female	3	0	9	9	S. pneumoniae	VZV
45	76	38	female	male	15	0	1.8	1.9	L. monocytogenes	VZV
46	44	37	female	female	-	2	1.5	1.5	E. faecalis	VZV

Table 2.

Demographic and clinical characteristics of pairs of patients with bacterial as opposed to viral acute meningitis - Cohort B

Pairs n.	Age		Gender		Charlson score		CRP values (mg/L)		Causative pathogen
	Bacterial	Viral	Bacterial	Viral	Bacterial	Viral	Bacterial	Viral	Bacterial	Viral
1	48	38	female	male	0	0	148	152	S. pneumoniae	EV
2	32	32	female	female	0	0	137	103	N. meningitidis	EV
3	46	34	male	male	0	-	124	92	S. pneumoniae	EV
4	55	42	female	male	1	0	114	80	S. epidermidis	EV
5	61	71	male	male	2	-	106	77	S. pneumoniae	WNV
6	69	32	female	female	-	0	89	75	S. epidermidis	EV
7	61	34	female	male	5	-	44	71	S. pneumoniae	EV
8	62	39	female	male	1	0	88	66	S. pneumoniae	EV
9	72	30	female	female	2	0	78	66	S. pneumoniae	HSV
10	51	34	female	male	3	0	78	62	S. pneumoniae	EV
11	87	34	female	male	*	0	5	54	S. pneumoniae	EV
12	55	40	female	male	6	0	57	50	S. pneumoniae	EV
13	83	38	female	male	1	0	50	44	S. pneumoniae	EV
14	76	32	female	female	-	0	42	42	L. monocytogenes	EV
15	38	35	female	female	4	0	39	35	Haemophilus influenzae	EV
16	38	38	male	male	0	0	37	35	Haemophilus influenzae	EV
17	47	32	female	female	-	0	35	32	S. pneumoniae	EV
18	53	29	female	female	1	0	32	32	S. pneumoniae	EV
19	44	23	female	female	-	0	32	27	viridans group streptococci	EV
20	52	40	female	male	3	0	27	27	S. pneumoniae	EV
21	55	30	female	female	1	0	25	23	S. pneumoniae	EV
22	66	37	female	male	3	0	25	23	S. pneumoniae	EV
23	59	35	male	male	-	0	23	22	L. monocytogenes	EV
24	68	30	female	male	2	0	22	21	S. pneumoniae	EV
25	76	40	male	female	4	0	21	18	S. pneumoniae	EV
26	77	34	male	male	5	0	19	20	viridans group streptococci	EV
27	71	32	male	male	3	0	19	16	S. pneumoniae	EV
28	35	63	female	female	0	2	16	11	S. pneumoniae	VZV
29	63	39	male	female	3	0	11	9	L. monocytogenes	VZV
30	50	32	female	female	-	0	9	2	S. pneumoniae	VZV
31	76	38	female	male	3	0	1.8	1.5	L. monocytogenes	VZV
32	43	38	female	male	0	0	1.5	152	E. faecalis	EV

Abbreviations: S. pneumoniae – Streptococcus pneumoniae, N. meningitidis – Neisseria meningitidis, S. epidermidis – Staphylococcus epidermidis, L. monocytogenes – Listeria monocytogenes, E. faecalis – Enterococcus faecalis, EV – Enterovirus, VZV – Varicella-Zoster Virus, WNV – West Nile Virus, HSV – Herpes Simplex Virus

Table 3.

Place and methods of isolation of the different Pathogens – Cohort A (46 pairs)

Bacterial pathogens
	Total	Blood stream	CSF detection	Ear drainage	Sputum
Streptococcus pneumoniae	30	23	17	3	1
Neisseria meningitidis	2	1	1	0	0
Enterococcus Faecalis	1	0	1	0	0
Enterococcus faecalis	4	2	3	0	0
Listeria monocytogenes	2	1	1	0	0
viridans group streptococci	2	1	1	0	0
Staphylococcus epidermidis	2	0	0	2	0
	43
Viral pathogens
	CSF detection			Serological IgM
Enterovirus	17			0
West Nile virus	1			0
Herpes simplex virus	0			1
Varicella-zoster virus	5			0
	24

Table 4.

Place and methods of isolation of the different Pathogens – Cohort B (32 pairs)

Bacterial pathogens
	Total	Blood stream	CSF detection	Ear drainage	Sputum
Streptococcus pneumoniae	20	15	12	2	1
Neisseria meningitidis	1	1	0	0	0
Enterococcus faecalis	1	0	1	0	0
Listeria monocytogenes	4	2	3	0	0
Haemophilus influenzae	2	1	1	0	0
viridans group streptococci	2	1	1	0	0
Staphylococcus epidermidis	2	0	0	2	0
	32
Viral pathogens
	CSF detection			Serological IgM
Enterovirus	27			0
West Nile virus	1			0
Herpes simplex virus	0			1
Varicella-zoster virus	4			0
	32

Results

The results are presented separately for Cohorts A and B, as described in the Methods section. While both cohorts were analysed using the same methodology, Cohort A includes all matched patients based on CRP values, whereas Cohort B consists solely of cases with a microbiologically confirmed diagnosis. In each cohort, bacterial and viral meningitis cases were compared with regard to demographic features, CRP concentrations at presentation, and estimated CRP velocities (eCRPv) calculated from different clinical onset points.

Cohort A

Gender and age were different within bacterial and viral groups. there were 33 vs. 18 women in the bacterial and viral groups respectively, and median age being 60 years [IQR 50.7–71.2] in the bacterial group as opposed to 35 years [IQR 29–38] in the viral group. Detailed values of CRP concentrations in these pairs are presented in Table 1. The mean ± s.d. CRP concentration upon presentation to the department of Emergency medicine of the bacterial and viral groups was 85.1 ± 68.2 mg/L, and 84.9 ± 69.1 mg/L respectively, with the differences between these two groups being non-significant (p value = 0.7), see Table 5.

Table 5.

Admission as well as estimated CRP velocities in the three modes of presentations [from general symptoms (eCRPv1), time from fever onset (eCRPv2) as well as time from meningitis symptomatology (eCRPv3)] - Cohort A

	Bacterial group	Viral group	P value
Admission CRP concentrations (mg/L)	85.07 (24.22-142.53)	84.96 (23.82-145.23)	0.709
eCRPv1 (mg/L/h), median (IQR)	2.68 (0.66–3.08)	1.11 (0.18–1.17)	< 0.001
eCRPv2 (mg/L/h), median (IQR)	8.02 (1.83–11.65)	1.30 (0.37–1.75)	0.001
eCRPv3 (mg/L/h), median (IQR)	11.86 (2.56–17.74)	1.45 (0.38–1.57)	< 0.001

Our results show that patients with bacterial meningitis presented higher velocities as opposed to those affected by viruses despite their presentation to the medical center with similar biomarker concentrations.

The estimated CRP velocity for general symptoms of disease onset (eCRPv1) was more than 2 times higher in the bacterial group compared to the viral one (median eCRPv1 2.7 and 1.1 mg/L/hour respectively, P < 0.001) as is demonstrated in Table 5; Fig. 1, with an AUC of 0.7 (0.6–0.8) as shown in Table 6; Fig. 2.

Fig. 1.

Estimated CRP velocity from general symptoms of disease (eCRPv1) – Cohort A

Table 6.

The area under the curve of estimated CRP velocities in the three modes of presentations [from general symptoms (eCRPv1), time from fever onset (eCRPv2) as well as time from meningitis symptomatology (eCRPv3)] - cohort A and cohort B

	Predictor	AUC (95%CI)	P value
Cohort A	eCRPv1	0.703 (0.596–0.809)	< 0.001
	eCRPv2	0.768 (0.622–0.913)	< 0.001
	eCRPv3	0.845 (0.759–0.931)	< 0.001
Cohort B	eCRPv1	0.762 (0.644–0.879)	< 0.001
	eCRPv2	0.711 (0.529–0.893)	0.023
	eCRPv3	0.830 (0.727–0.932)	< 0.001

Fig. 2.

eCRPv1 – ROC curve – Cohort A

The next eCRPv (eCRPv2) was calculated according to the estimated time from fever onset (temperature ≥ 38 °C) and was found to be more than 6 times higher in the bacterial group compared to the viral group (median eCRPv2 8.0 and 1.3 mg/L/hour respectively, P = 0.001), see Table 5; Fig. 3, with an AUC of 0.8 (0.6–0.9) as shown in Table 6; Fig. 4.

Fig. 3.

Estimated CRP velocity from fever onset (eCRPv2) – Cohort A

Fig. 4.

eCRPv2 – ROC curve – Cohort A

Finally, we calculated the estimated CRPv from the onset of meningitis symptoms (eCRPv3). It was found that this estimated CRP velocity was more than 8 times faster in the bacterial group than in the viral group (median eCRPv3 11.9 and 1.5 mg/L/hour respectively, P < 0.001), see Table 5; Fig. 5, with an AUC of 0.8 (0.8–0.9) as shown in Table 6; Fig. 6.

Fig. 5.

Estimated CRP velocity from meningitis symptoms onset (eCRPv3) – Cohort A

Fig. 6.

eCRPv3 – ROC curve – Cohort A

Cohort B

Gender and age were different within bacterial and viral groups. there were 23 vs. 12 women in the bacterial and viral groups respectively, and median age being 58 years [IQR 47–70] in the bacterial group as opposed to 36 years [IQR 32–38] in the viral group. Detailed values of CRP concentrations in these pairs are presented in table Table 2. The mean ± s.d. CRP concentration upon presentation to the department of Emergency medicine of the bacterial and viral groups was 48.6 ± 41.3 mg/L, and 45.6 ± 33.3 mg/L respectively, with the differences between these two groups being non-significant (p value = 0.17), see Table 7.

Table 7.

Admission as well as estimated CRP velocities in the three modes of presentations [from general symptoms (eCRPv1), time from fever onset (eCRPv2) as well as time from meningitis symptomatology (eCRPv3)] - Cohort B

	Bacterial group	Viral group	P value
Admission CRP concentrations (mg/L)	48.60 (19.5–78)	45.56 (21.27–69.75)	0.174
eCRPv1 (mg/L/h), median (IQR)	2.00 (0.51–2.56)	0.56 (0.15–0.73)	< 0.001
eCRPv2 (mg/L/h), median (IQR)	4.96 (0.825–6.12)	0.96 (0.37–1.24)	0.011
eCRPv3 (mg/L/h), median (IQR)	8.21 (2.08–13.86)	1.09 (0.26–1.21)	< 0.001

Consistent with our findings in cohort A, cohort B results also exhibited increased velocities in bacterial meningitis cases relative to viral cases, notwithstanding similar initial biomarker levels.

The estimated CRP velocity for general symptoms of disease onset (eCRPv1) was almost 3 times higher in the bacterial group compared to the viral one (median eCRPv1 2 and 0.6 mg/L/hour respectively, P < 0.001) as is demonstrated in Table 7; Fig. 7, with an AUC of 0.8 (0.6–0.9) as shown in Table 6; Fig. 8.

Fig. 7.

Estimated CRP velocity from general symptoms of disease (eCRPv1) – Cohort B

Fig. 8.

eCRPv1 – ROC curve – Cohort B

The eCRPv2 in this cohort (estimated time from fever onset) was found to be more than 5 times higher in the bacterial group compared to the viral group (median eCRPv2 5.0 and 1.0 mg/L/hour respectively, P = 0.011), see Table 7; Fig. 9, with an AUC of 0.7 (0.5–0.9) as shown in Table 6; Fig. 10.

Fig. 9.

Estimated CRP velocity from fever onset (eCRPv2) – Cohort B

Fig. 10.

eCRPv2 – ROC curve – Cohort B

Finally, it was found that the eCRPv3 (estimated CRPv from the onset of meningitis symptoms) was more than 7 times faster in the bacterial group than in the viral group (median eCRPv3 8.2 and 1.1 mg/L/hour respectively, P < 0.001), see Table 7; Fig. 11, with an AUC of 0.8 (0.7–0.9) as shown in Table 6; Fig. 12.

Fig. 11.

Estimated CRP velocity from meningitis symptoms onset (eCRPv3) – Cohort B

Fig. 12.

eCRPv3 – ROC curve – Cohort B

Discussion

The differentiation between bacterial and viral meningitis in clinical practice involves several laboratory tests and clinical parameters. For instance, serum procalcitonin, absolute Neutrophil Count, serum Ferritin, erythrocyte Sedimentation Rate, neutrophil-to-lymphocyte ratio (NLR) and CSF/serum glucose ratio are significantly elevated in bacterial meningitis compared to viral meningitis.

It is also known, that serum CRP can also help to distinguish between bacterial and viral meningitis, presents a diagnostic challenge in individuals who present with similar CRP concentrations. Therefore, we have adopted a kinetic approach in order to improve discrimination by considering also the time from disease onset. By adopting this approach, we turn a single value of this biomarker into a dynamic one that reflects not only the severity of the inflammatory response but also the rapidity of its development. Our research hypothesis is that high CRP velocities are a result of bacterial as opposed to a viral pathogen. This concept was previously demonstrated by our group in the past [15, 16].

The notion of using CRP kinetics for clinical purposes is not new and has been shown in the past by using two different CRP tests and calculating the time difference between them, the so-called CRP velocity [17–20]. Estimated CRP velocity presents an opportunity to obtain dynamic information regarding the speed of the development of the inflammatory burst. It is thought that a higher CRP value in the early stages of infection reflects a more rapid and aggressive inflammatory response. It is also assumed that this rapidity is related to the virulence of the invading agent although this requires further investigation. Accordingly, this study suggests that bacterial meningitis causes a more significant and rapid inflammatory response than viral meningitis.

The iso-CRPinemic model is the preferred approach to investigate differences between bacterial and viral infections in patients who present with similar CRP concentrations. The significant contribution of this study is the notion that by correlating the estimated time from disease onset, one can improve the differentiation between both etiologies despite the presentation of the patients with similar CRP concentrations. We believe that this approach is not necessarily limited to patients who present with similar results but might be relevant to most, if not all, patients due to the possibility that the velocity reflects the virulence of the invading agent. Our hypothesis is that a slow CRP increment is associated with a more indolent disease and vice versa although this hypothesis requires further studies in different models of virulent as well as more indolent infections.

The advantage of this novel approach is that it makes possible the early identification of patients presenting with a severe inflammatory response using information gathered very early on during a patient’s stay in the Emergency Room department. It is needless to emphasize the potential diagnostic and therapeutic advantages of obtaining these early alarm signals.

Despite CRP-based matching, the bacterial and viral groups differ substantially in age, sex, and comorbidity. We can not exclude the possibility that these imbalances could have an effect on CRP kinetics. However, our study represents a “real world” situation suggesting that significant information can be obtained from CRP kinetics despite the above mentioned differences between the two cohots.

In the present study, we included time from fever in addition to the time from symptoms onset since using symptoms only to define disease onset might overlap with features of more severe bacterial illness risking inflated eCRPv differences yet we observed that estimated times from both fever and symptoms onset were significantly different in both types of meningitis. We do believe that in daily practice, clinicians could benefit from taking into consideration both of them.Our study is subject to several limitations, primarily its retrospective nature. In fact, in such a design it is not possible to optimize the assessment of time of disease onset by obtaining additional information from the patient or his relatives. Another limitation of the retrospective nature of the study is the difficulty in simultaneously obtaining additional biomarkers, such as procalcitonin, which we know can help as well to differentiate between bacterial and viral infections [21]. The same is true for N/L ratio or CSF data. This could be evaluated only in a prospective approach. Another limitation is the fact that most cases of bacterial infections were due to Streptococcus Pneumonia and therefore it is not clear if the present findings are applicable to other bacterial etiologies as well. In addition, the core variable is based on retrospective patient recall with no standarized method and this can introduce a bias. Finally, it is clear that the exact time of onset of meningeal inflammation is not known and presently is based on a rough and semi quantitative estimation solely. On the other hand, we can see that the differences between the groups remained significant in all the three calculations namely from onset of any symptomatology, from fever as well as from the appearance of a more specific presentation of acute meningitis.

Conclusions

We conclude that the diagnostic yield of CRP at presentation amongst patients with acute meningitis can be improved by adjusting the value to the estimated time from disease onset. Further studies are needed, and especially prospective ones, in order to further improve this diagnostic approach.

Abbreviations

AUC

Area under curve

CRP

C-Reactive Protein

eCRPv

Estimated CRP velocity

IQR

Interquartile range

NLR

Neutrophil-to-lymphocyte ratio

PMN

Polymorphonuclear

wrCRP

Wide-Range C-Reactive Protein

Author contributions

N.S.M., A.W., and O.A. contributed to writing the original draft. Data collection was performed by N.S.M., M.K.S., T.W., Y.M., and N.S. The study was conceptualized and supervised by A.W., A.A., S.B., O.R., and D.Z. Methodology development was led by A.W., A.A., S.B., T.Z.B., and U.O. Data analysis and figure preparation were carried out by T.Z.B. and N.S.M. All authors reviewed the manuscript.

Funding

Partial financial support was received from the Dalia and Arie Prashkovsky grant for biomedical research.

Data availability

Data supporting the findings of this study are available within the paper. Additional data that support the findings of this study are not openly available due to reasons of patient’s privacy and are available from the Prop. Shlomo Berliner (e-mail: berliners@tlvmc.gov.il) upon reasonable request. Data are located in controlled access data storage at Tel-Aviv Medical Center.

Declarations

Ethics approval and consent to participate

The study was reviewed and approved by the Institutional Review Board according to the declaration of Helsinki (TLV-0645-23). Informed consent by the patients was waived due to the retrospective nature of our study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.