Haematological parameters predicting cardiac involvement in children with COVID-19 infection.

AIM: Haematological parameters obtained from the full blood count, such as neutrophil-to-lymphocyte ratio (NLR), are cost-effective tests which have been shown to be predictive of the prognosis of many diseases. We aimed to evaluate certain haematological parameters and cardiac biomarkers to test whether they could predict cardiac involvement by COVID-19 infection.
METHODS: This retrospective study included patients aged 1 month to 18 years having a positive COVID-19 PCR test but no comorbidity, who were admitted to the paediatric emergency department between 15 March 2020 and 1 February 2021.
RESULTS: There were 292 COVID-19 PCR-positive patients, 12 MIS-C patients and 70 healthy controls. A receiver operator characteristic curve analysis was performed to predict MIS-C in patients with COVID-19 infection. An NLR value of ≥5.03 could predict MIS-C with a sensitivity of 66.7% and a specificity of 91.6%; a proBNP value of ≥329.5 ng/L with a sensitivity of 91.7% and a specificity of 95.6%; a CKMB value of ≥2.95 μg/L with a sensitivity of 100% and a specificity of 77.7%; and a troponin-I value of ≥0.03 μg/L with a sensitivity of 75% and a specificity of 99.2%. A logistic regression analysis showed that an NLR value of ≥5.03 increased the risk of MIS-C 19.3 fold; a proBNP value of ≥329.5 ng/L increased the risk 238 fold; and a troponin-I value of ≥0.03 μg/L increased the risk 60 fold.
CONCLUSIONS: At the time of admission, parameters such as proBNP, troponin-I and NLR can predict the development of MIS-C in COVID-19 patients with high sensitivity and specificity.

What is already known on this topic

Full blood count is a cost‐effective test.

It has been reported that the neutrophil/lymphocyte ratio (NLR) predicts prognosis in many diseases.

MIS‐C, a post‐COVID‐19 complication, can be severe.

What this paper adds

Parameters such as NLR, proBNP, troponin‐I and CRP were shown to have high sensitivity and specificity (66.7% and 91.8%; 91.7% and 95.6%; 75% and 99.2%; 91.7% and 89.7%; respectively) in the early detection of MIS‐C.

It will be easier for clinicians to predict the development of MIS‐C in patients with COVID‐19 infection.

Although more than 1 year has passed, we still continue to experience the negative effects of COVID‐19, which was declared a pandemic on 11 March 2020. According to WHO data, 126 697 603 cases were confirmed world‐wide as of 29 March 2021, and approximately 2% of these cases have died. Although it was initially reported that children usually survive the disease with mild or even no symptoms unlike adult patients, a multisystem inflammatory syndrome (MIS‐C) associated with COVID‐19 was identified in children in late April 2020. , , It has been reported that this syndrome possesses similar characteristics with several diseases such as Kawasaki disease and hemophagocytic lymphohistiocytosis; it is reportedly a hyperinflammatory condition with fever, increased laboratory markers of inflammation, multisystemic organ involvement and cardiovascular shock.

MIS‐C is a syndrome of post‐infectious immune dysregulation. Cytokine storm plays an important role in the pathogenesis of this syndrome. Cardiac dysfunction, haematological involvement such as thrombocytopenia, coagulopathy, gastrointestinal involvement, neurological involvement and ultimately multi‐organ failure may develop. Morbidity and even death may occur.

Full blood count is a cost‐effective test that can be studied in almost any health‐care facility. In recent years, haematological parameters showing systemic inflammation, such as red cell distribution width (RDW), neutrophil to lymphocyte ratio (NLR), platelet‐to‐lymphocyte ratio (PLR), have been reported to predict prognosis in many diseases. , ,

In our study, we aimed to predict COVID‐19's cardiac involvement using these cost‐effective haematological parameters and to evaluate the correlation of MIS‐C with these cardiac biomarkers.

Methods

Patients aged between 1 month and 18 years who presented to the paediatric emergency department with suspected COVID‐19 between 15 March 2020 and 1 February 2021 were included in this single‐centre, retrospective study.

Almost everyone in Turkey can present to the emergency departments. There are no restrictions as to which patients can present to the emergency departments ‘green zone’. , , Therefore, patients with low, moderate, and severe risk were included in our study.

This study was conducted in accordance with the criteria of Declaration of Helsinki. Its ethics approval was obtained from the local ethics committee before the start of the study (2021/10‐03). Healthy children of the same age without complaints were included as the control group. Patients whose haematological parameters and cardiac biomarkers were missing, patients with comorbid conditions (patients with cardiac or neurological sequelae, asthma, immunodeficiency, haematological disorders or malignancy), and patients with a negative COVID‐19 PCR test were excluded from the study.

The demographic characteristics, and clinical and laboratory findings at the time of admission (full blood count parameters, pro‐brain natriuretic peptide (proBNP), creatine kinase, CK‐MB and troponin‐I) were obtained from the medical records of the study subjects.

MIS‐C was diagnosed according to the diagnostic criteria issued by the Center for Disease Control and Prevention (CDC). These include age < 21 years, fever ≥38°C for ≥24 h, laboratory evidence of inflammation, serious illness requiring hospitalisation, involvement of two or more systems, positive PCR, serology, or antigen tests, COVID‐19 exposure within 4 weeks before symptom onset, and the absence of any other possible cause.

Nasopharyngeal swabs for COVID‐19 RT‐PCR analysis were examined using the BIO‐RAD CFX96 Real‐Time System C1000 Touch Thermal Cycler Device, SARS‐CoV‐2 Double Gene RT‐qPCR 1000 Rxn kit and COVID‐19 RT‐PCR kit.

The Statistical Package for the Social Sciences for Windows 22 software was used for statistical analysis. Study variables were presented as number (n) – percentage (%) and mean ± standard deviation. The normality of distribution of the study variables was tested with the Kolmogorov–Smirnov test. Normally distributed numerical variables were compared by one‐way analysis of variance or Student's t‐test while non‐normally distributed ones were compared by the Kruskal–Wallis test or Mann–Whitney U test. The risk factors were evaluated with univariate and multivariate logistic regression models. The variables found to be significantly predictive of MIS‐C in the univariate analysis were included in the logistic regression analysis. The receiver operator characteristic (ROC) curve analysis was performed to find an optimal cutoff point of candidate variables for the prediction of MIS‐C. A P value of less than 0.05 was considered statistically significant.

Results

Between 15 March 2020 and 1 February 2021, a total of 2870 COVID‐19 tests were performed on patients who were admitted to the paediatric emergency service with suspected COVID‐19 infection. A total of 2784 patients underwent COVID‐19 PCR testing; 32 COVID‐19 patients underwent IgM rapid testing; 43 patients underwent COVID‐19 IgG rapid testing, and 11 – underwent both COVID‐19 IgM and IgG rapid testing. According to the inclusion criteria, 292 patients with positive COVID‐19 PCR tests, 12 MIS‐C patients and 70 healthy controls were included in this study.

The mean age of the patients was 11.04 ± 5.34 (0–18) years. The mean age of the 70 healthy controls was 11.41 ± 4.92 years; 292 patients with COVID‐19 infection who had a positive COVID‐19 PCR test but no MIS‐C had a mean age of 10.89 ± 5.53 years; and 12 patients with MIS‐C had a mean age of 8.35 ± 5.32 years. There was no statistically significant difference between the groups in terms of age and gender (P = 0.273, P = 0.479, respectively).

When the patients were evaluated according to their symptoms, the most common symptoms were fever (56.8%), cough (31.5%), weakness (18.2%), headache (15.8%) and sore throat (18.2%). The least common symptoms were loss of appetite (1.7%) and nausea (3.1%).

Abdominal pain was present in 50% of MIS‐C patients and vomiting in 35% of them; both rates were significantly higher than those of the COVID‐19 cases without MIS‐C (P < 0.001, P = 0.003, respectively) (Table 1).

Table 1

Comparison of patients according to their demographic characteristics and symptoms

		COVID‐19 infected patients
	Healthy control (n = 70)	Non‐MIS‐C (n = 292)	MIS‐C (n = 12)	P †
Age (mean ± SD)	11.41 ± 4.92	10.89 ± 5.53	8.35 ± 5.32	0.273
Gender	N (%)	N9 (%)	N (%)	P ‡
Male	34 (48.6)	143 (49)	4–33.3	0.479
Female	36 (51.4)	149 (51)	8 (66.7)
Symptoms
Fever	–	166 (56.8)	12 (100)	0.003
Cough	–	92 (31.5)	1 (8.3)	0.088
Diarrhoea	–	24 (8.2)	2 (16.7)	0.305
Loss of taste	–	20 (6.8)	0	0.348
Loss of smell	–	23 (7.9)	0	0.312
Myalgia	–	16 (5.5)	0	0.405
Loss appetite	–	5 (1.7)	1 (8.3)	0.108
Weakness	–	53 (18.2)	1 (8.3)	0.383
Vomiting	–	14 (4.8)	3 (25)	0.003
Abdominal pain	–	15 (5.1)	6 (50)	<0.001
Headache	–	46 (15.8)	2 (16.7)	0.932
Nausea	–	9 (3.1)	1 (8.3)	0.319
Runny nose	–	15 (5.1)	2 (16.7)	0.140
Throat ache	–	53 (18.2)	0	0.104

Independent Student's t‐test.

Chi‐square test.

MIS‐C, multisystem inflammatory syndrome.

When the patients were evaluated according to laboratory data; WBC, CRP, procalcitonin, IG (immature granulocyte), IG%, NLR, troponin‐I and proBNP levels were significantly higher in the MIS‐C group compared to the other groups (P = 0.005, P < 0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.001, P < 0.001, respectively). The albumin level was significantly lower in the MIS‐C group (P < 0.001). There was no statistically significant difference between the groups regarding creatine kinase and CKMB levels (P = 0.066, P = 0.051, respectively) (Table 2).

Table 2

Comparison of patients' laboratory data

	Healthy control (n = 70)	COVID‐19 infected patients (n = 304)		P †
	Mean ± SD	Non‐MIS‐C (n = 292)	MIS‐C (n = 12)
		Mean ± SD	Mean ± SD
WBC (109/L)	7.906 ± 2.058	6.947 ± 2.685	8.462 ± 2.415	0.005
CRP (mg/L)	3.139 ± 0.336	6.341 ± 10.299	154.041 ± 113.142	<0.001
Procalsitonin (μg/L)	0.034 ± 0.017	0.082 ± 0.123	9.404 ± 12.399	<0.001
Albumin (g/dL)	4.783 ± 0.226	4.721 ± 0.264	3.257 ± 0.466	<0.001
IG (106/L)	15.14 ± 9.74	16.45 ± 16.54	39.17 ± 28.11	<0.001
IG%	0.178 ± 0.112	0.221 ± 0.175	0.433 ± 0.270	<0.001
NLR	1.37 ± 0.96	2.37 ± 2.59	7.43 ± 5.61	<0.001
CK (U/L)	129.25 ± 98.89	109.19 ± 78.79	77.67 ± 58.84	0.066
CKMB (μg/L)	5.97 ± 17.47	2.76 ± 2.08	3.92 ± 4.75	0.051
Troponin‐I (μg/L)	0.00 ± 0.00	0.0013 ± 0.0107	0.0375 ± 0.0674	<0.001
ProBNP (ng/L)	50.70 ± 35.49	99.78 ± 210.71	6855.25 ± 6110.70	<0.001

Kruskal–Wallis test.

CK, creatine kinase, CRP, C‐reactive protein; IG, immature granuloycte; MIS‐C, multisystem inflammatory syndrome in children; NLR, neutrophil‐to‐lymphocyte ratio; ProBNP, pro‐brain natriuretic peptide.

The laboratory data of the patients were analysed with a ROC curve to predict MIS‐C (Table 3). The best cutoff points for MIS‐C were determined. According to these analyses, in patients with suspected COVID‐19 infection; MIS‐C can be predicted by CRP levels of ≥12.65 mg/L, with a sensitivity of 91.7% and a specificity of 89.7%; by an IG level of ≥35∙106/L with a sensitivity of 58.3% and a specificity of 88.7%; by an IG% level of ≥0.35 with a sensitivity of 73.7% and a specificity of 88.7%; by a neutrophil‐to‐lymphocyte ratio (NLR) level of ≥5.03 with a sensitivity of 66.7% and a specificity of 91.6%; by a procalcitonin level of ≥0.165 μg/L with a sensitivity of 100% and a specificity of 93.4%; by a proBNP level of ≥329.5 ng/L with a sensitivity of 91.7% and a specificity of 95.6%; by a CKMB level of ≥2.95 μg/L, with a sensitivity of 100% and a specificity of 77.7%; and by a troponin‐I level of ≥0.03 μg/L with a sensitivity of 75% sensitivity and a specificity of 99.2%.

Table 3

Determination of cutting points of laboratory data that can predict MIS‐C

	Cutoff value	AUC	Sensitivity	Specificity	Asymptotic 95% confidence interval	P †
CRP (mg/L)	≥12.65	0.907	0.917	0.897	0.752–1	<0.001
IG (≥35∙106/L)	≥35	0.749	0.583	0.918	0.586–0.912	0.003
IG%	≥0.35	0.737	0.583	0.887	0.562–0.913	0.005
NLR	≥5.03	0.823	0.667	0.916	0.672–0.974	<0.001
Procalsitonin (μg/L)	≥0.165	0.992	1	0.934	0.981–1	<0.001
ProBNP (ng/L)	≥329.5	0.980	0.917	0.956	0.988–1	<0.001
CK‐MB (μg/L)	≥2.95	0.897	1	0.777	0.800–0.995	0.007
Troponin‐I (μg/L)	≥0.03	0.870	0.750	0.992	0.613–1	0.011

ROC curve analysis.

CK, creatine kinase, CRP, C‐reactive protein; IG, immature granuloycte; MIS‐C, multisystem inflammatory syndrome in children; NLR, neutrophil‐to‐lymphocyte ratio; ProBNP, pro‐brain natriuretic peptide; ROC, receiver operator characteristic.

In addition, when a risk analysis was performed with logistic regression analysis according to the cutoff values of the laboratory data, using a ROC curve analysis for MIS‐C development, we determined that if IG was ≥35∙106/L, the risk was 15.63 times higher, if IG% was ≥0.35, the risk was 11 times higher; if the NLR was ≥5.03, the risk was 19.3 times higher, if procalcitonin was ≥0.165 μg/L, the risk was 7.5 times higher, if proBNP was ≥329.5 ng/L, the risk was 238 times higher, if troponin‐I was ≥0.03 μg/L, the risk was 60 times higher; and if CRP was ≥12.65 mg/L, the risk was 142 times higher (Table 4).

Table 4

Evaluation of risk factors for MIS‐C development by logistic regression analysis

	OR	95% Cl	P †	Risk
IG (≥35∙106/L)	15.63	4.610–53.020	<0.001	Yes
IG % (≥0.35)	11	3.298–36.607	<0.001	Yes
NLR (≥5.03)	19.3	5.522–67.417	<0.001	Yes
Procalsitonin (≥0.165 μg/L)	7.5	4537‐12 550	<0.001	Yes
ProBNP (≥329.5 ng/L)	238	28.909–1957.300	<0.001	Yes
CK‐MB (≥2.95 μg/L)	0.263	0.076–0.909	0.035	No
Troponin‐I (≥0.03 μg/L)	60	8.907–404.190	<0.001	Yes
CRP (≥12.65 mg/L)	142	17.660–1144.291	<0.001	Yes

Logistic regression analyses.

CI, confidence interval; CK, creatine kinase, CRP, C‐reactive protein; IG, immature granuloycte; MIS‐C, multisystem inflammatory syndrome in children; NLR, neutrophil‐to‐lymphocyte ratio; OR, odds ratio; ProBNP, pro‐brain natriuretic peptide.

When the patients were evaluated according to the correlation of laboratory findings, there was a weak positive correlation between proBNP and IG (r: 0.206, P < 0.001), a moderate positive correlation between proBNP and troponin‐I (r: 0.519, P < 0.001), a moderate positive correlation between proBNP and CRP (r: 0.510, P < 0.001), a moderate positive correlation between proBNP and procalcitonin (r: 0.457, P < 0.001), but no significant correlation between proBNP and CKMB (r: 0.028, P = 0.682) (Table 5).

Table 5

Correlations between patients' laboratory data

		IG	ProBNP	Troponin‐I	CKMB	CRP	Procalsitonin
IG	Pearson correlation	1	0.206†	0.067	0.012	0.398†	0.223†
	P		0.000	0.225	0.866	0.000	0.000
ProBNP	Pearson correlation	0.206†	1	0.519†	0.028	0.510†	0.457†
	P	0.000		0.000	0.682	0.000	0.000
Troponin‐I	Pearson correlation	0.067	0.519†	1	0.027	0.155†	0.321†
	P	0.225	0.000		0.693	0.005	0.000
CKMB	Pearson correlation	0.012	0.028	0.027	1	−0.027	−0.026
	P	0.866	0.682	0.693		0.692	0.724
CRP	Pearson correlation	0.398†	0.510†	0.155†	−0.027	1	0.536†
	P	0.000	0.000	0.005	0.692		0.000
Procalsitonin	Pearson correlation	0.223†	0.457†	0.321†	−0.026	0.536†	1
	P	0.000	0.000	0.000	0.724	0.000

Correlation is significant at the 0.01 level (two‐tailed).

CK, creatine kinase, CRP, C‐reactive protein; IG, immature granuloycte; MIS‐C, multisystem inflammatory syndrome in children; ProBNP, pro‐brain natriuretic peptide.

The clinical data of 12 patients followed up with the diagnosis of MIS‐C are shown in Table 6. All patients were given intravenous immunoglobulin (IVIG) and steroid. Only patient 12 was given anakinra in addition to IVIG and steroids. None of our patients died.

Table 6

The clinical data of 12 patients with MIS‐C

Patient number	Age	Sex	MIS‐C severity	Vasoactive drugs	Hospital stay (day)	First echo	Cardiac dysfunction findings
1	8 years	Female	Severe	Dobutamin 5 days	22	Ef 64	Tachycardia
						Kf 34	Prolonged capillary refilling time
						Trace MR	Lightly confused
							Abdominal pain
2	7 years	Male	Severe	Total 11 days	46	Ef 62	Clouding of consciousness
				Dobutamin 8 days		Kf 32	Tachycardia
				Adrenalin 6 days		Mild MR	Prolonged capillary refilling time
				Milrinon 7 days			Abdominal pain
3	14 years	Male	Moderate	Dobutamin 3 days	9	Ef 62	Abdominal pain
						Kf 33	Prolonged capillary refilling time
						Mild MR	Tachycardia
4	2 months	Male	Severe	Dobutamin 8 days	19	Ef 71 (while administering dobutamine)	Hepatomegaly
						Kf 38	Tachypnea
							Tachycardia
							İnability to feed
							Prolonged capillary refilling time
5	13 years	Male	Severe	Dobutamin 6 days	20	Ef 38	Abdominal pain
						Kf 19	Hypotension
						Trace AR	Tachycardia
						Mild MR	Tachypnea
							Prolonged capillary refilling time
6	1 month	Female	Moderate	Dobutamin 11 days	21	Ef 74	Prolonged capillary refilling time
						Kf 40	İnability to feed vomiting
7	6.5 years	Male	Moderate	Dobutamin 3 days	20	Ef 71	Prolonged capillary refilling time
						Kf 39	Tachycardia
						Mild MR, Trace AR
8	12 years	Female	Moderate	Dobutamin 6 days	12	Ef 62	Prolonged capillary refilling time
						Kf 33	Tachycardia
						Mild MR
9	2.4 years	Male	Moderate	Dobutamin 9 days	15	Ef 76	Prolonged capillary refilling time
						Kf 43	Tachycardia
						Pericardial effusion 3 mm at systole
10	5.4 years	Male	Mild	No	7	Ef 70, Kf 39	–
11	13 years	Male	Moderate	Dobutamin 5 days	19	Ef 71	Tachycardia
						Kf 40	Abdominal pain
							Prolonged capillary refilling time
12	8 years	Female	Severe	Dobutamin 3 days	14	EF: 44	Hypotension
				Adrenalin 2 days		KF:22	Hepatomegalyli
				Milrinon 3 days		Mild MR, Pericardial effusion 6 mm at systole, 1 mm at diastole	Tachypnea
							Tachycardia
							Abdominal pain

AR, aortic regurgitation; Ef, ejection fraction; MR, mitral regurgitation; Sf, shortening fraction.

Discussion

Our study is one of the rare studies in the literature that examines blood parameters that can shed light on cardiac involvement and prognosis in paediatric patients with COVID‐19, and it analyses the sensitivity, specificity, and the relative risk of these parameters for the development of MIS‐C. According to our results, a proBNP level of 329.5 ng/L or above and a procalcitonin value of 0.165 μg/L or above at the time of admission could predict MIS‐C with sensitivities and specificities of more than 90%.

Although the pathophysiology of MIS‐C has not yet been fully elucidated, it has been proposed that it occurs due to immune dysregulation. Patients have multisystemic symptoms such as shock, cardiac dysfunction, and abdominal pain. In a European study involving 286 MIS‐C patients, Valverde et al. reported that cardiac involvement was frequent, and the levels of BNP, ferritin, D‐dimer, troponin, CRP, and procalcitonin were increased. Although MIS‐C is characterised by multisystemic involvement, its mortality is reportedly low in children compared with adults. In a review by Hoste et al., BNP and troponin levels indicating cardiac damage as well as inflammatory markers such as CRP, ferritin and IL‐6 increased in MIS‐C cases.

Although WBC increased mostly, lymphocytopenia was widely observed. Platelet counts were found to be normal, d‐dimer and fibrinogen were increased in coagulation parameters.

Similarly, in a study reported by Minocha et al., CRP, d‐dimer and ferritin levels were increased in MIS‐C patients. Cardiac abnormalities were reported in 73% of the patients. Increased BNP was seen in 43% of these, and an increased troponin level in 21%. In a case series of 1116 patients, Feldstein et al. reported a higher neutrophil to lymphocyte ratio, CRP, and a lower platelet count in patients with MIS‐C compared with patients with COVID‐19. In our study, we found a significantly higher NLR level in the MIS‐C group, which was consistent with the results reported by Feldstein et al. In addition, in contrast to the literature, we determined a cutoff value for NLR that can predict MIS‐C (66.7% sensitivity, 91.6% specificity for an NLR level ≥ 5.03). We also showed that an NLR level ≥ 5.03 increased the risk of developing MIS‐C approximately 19 times. Hypoalbuminemia has been frequently reported in patients with MIS‐C. Consistent with the literature, we found that albumin levels were significantly lower in the MIS‐C group (P < 0.001).

In a review by Kwak et al., higher procalcitonin levels were found in MIS‐C patients compared with COVID‐19 patients. While neutrophilia and lymphopenia were observed in most patients, anaemia and thrombocytopenia were only observed in some of them. However, another study showed neutrophilia, lymphopenia, and high levels of CRP, fibrinogen and d‐dimer in children with MIS‐C. Troponin‐I was also found to be moderately high, while BNP was significantly higher in those children.

In our study, in accordance with the literature, we found significantly higher WBC, CRP, procalcitonin, NLR, troponin‐I, and proBNP levels in the MIS‐C group compared with patients with COVID 19 infection but without MIS‐C, and the healthy controls. Unlike previous studies, we determined the sensitivity and specificity of these parameters according to the cutoff values that can predict MIS‐C, and we also made a risk analysis according to these cutoff values. According to these analyses, we showed that a procalcitonin level ≥ 0.165 μg/L increased the risk of MIS‐C by 7.5 times; a proBNP level ≥ 329.5 ng/L likewise increased the same risk by 238 times; a troponin‐I level ≥ 0.03 μg/L increased the risk by 60 times; and a CRP level ≥ 12.65 mg/L increased the risk by 142 times. Moreover, in contrast to the literature, we performed correlation analyses to evaluate the relationship between proBNP, procalcitonin, CRP, CK‐MB and troponin‐I. Based on this analysis, we found a moderate positive correlation between proBNP and CRP, and procalcitonin. This is meaningful in that it shows us that as the inflammation increases, the heart is more likely to be affected.

Conclusion

Our study is a valuable one because it provided cutoff values, sensitivity‐specificity levels, and a risk analysis to predict MIS‐C using several laboratory parameters; furthermore, it also involved low‐risk patients. Parameters such as proBNP, troponin‐I, NLR and CRP were shown to have high sensitivities and specificities for the early detection of MIS‐C. These findings will make it easier for clinicians to predict the development of MIS‐C in patients with suspected COVID‐19 infection.

Limitation

The limitations of our study include its non‐prospective, single‐centre and small‐volume design.