Diagnostic capacity of miRNAs in neonatal sepsis: a systematic review and meta-analysis

Abstract

Background

Neonatal sepsis is the third leading cause of mortality during the neonatal period, with manifestations atypical and obscure. But the gold standard-blood culture test, requiring 3–5 days, makes it difficult to unveil the final pathogen and leads to the increasing ratio of false-negative results. The empirical method is consulting traditional biomarkers, such as procalcitonin (PCT), C-reactive protein (CRP), and white blood cell count. However, they are not specific for neonate in diagnostic capacity, especially for infants within three days after delivery, so more novel biomarkers are urgently needed to assist diagnosing neonatal sepsis. microRNAs (miRNAs) have been widely studied in recent years for their diagnostic and prognostic values in different diseases and we conducted a meta-analysis of miRNAs on the topic that whether they are potentially novel biomarkers in early detection of neonatal sepsis.

Objectives

The purpose of the study was to assess whether circulating miRNAs could be used as potential biomarkers for neonatal sepsis, including early and late-onset neonatal sepsis, then calculate their overall accuracy (OA) via meta-analysis.

Methods

PubMed, Cochrane Library, Embase, Web of Science, Scopus, and Ovid databases were retrieved; data cutoff for this analysis was 15 January 2023. Methodological quality assessment of included studies was performed through the Quality in Prognostic Studies tool. Corresponding 95% confidence interval (95%CI) was calculated to present miRNAs’ diagnostic value including the pooled sensitivity (Sen), specificity (Spe), positive or negative likelihood ratios (PLR or NLR), diagnostic odds ratio (DOR), and area under the curve (AUC). Differences in OA between the septic group and non-septic group were compared using Chi-square test.

Results

After identification, 16 records out of 11 selected articles were eligible for systematic review of miRNAs and four records for PCT; the case group for miRNAs included 945 neonatal sepsis cases; contrast group included 190 respiratory tract infections or pneumonia cases, 60 systemic inflammatory response syndrome (SIRS) cases and 559 healthy neonates. The pooled Sen, Spe, and DOR of miRNAs were 0.87 (95%CI 0.81–0.91), 0.79 (95%CI 0.71–0.85), and 24 (95%CI 12–50), respectively. The pooled Sen, Spe, and DOR of PCT were 0.92 (95%CI 0.83–0.96), 0.64 (95%CI 0.56–0.70), and 20 (95%CI, 7–56), respectively. The OA value of miRNAs was 80.38% and that of PCT was 77.36%, which were not statistically significant difference (p = .13) after the Chi-square test. In addition, no significant publication bias was indicated (p = .92).

Conclusions

Circulating miRNA levels could be applied as diagnostic biomarkers in neonatal sepsis.

Keywords:

• Neonatal sepsis
• miRNA
• diagnosis
• biomarker
• meta-analysis

Introduction

Neonatal sepsis, a complex and heterogeneous condition produced as response to invasive blood infection mostly caused by bacteria, classified as early or late-onset, is one of the leading causes of mortality and morbidity during the neonatal period. Early onset sepsis (EOS) is defined as the presence of bacteremia or meningitis in the first three days of life. Late-onset sepsis (LOS) is defined as a systemic infection in neonates older than three days (4–28 days). The pathogens mainly include group B streptococcus (GBS), Escherichia coli, coagulase-negative Staphylococcus, Haemophilus influenzae, and Listeria monocytogenes [1]. A systematic analysis pointed out that neonatal sepsis continues to be the third leading cause of morbidity and mortality in infants, following preterm and pregnancy events [2]. Despite antibiotic administration before delivery, there are 0.5 million infants who die from sepsis [3], for there are always no obvious manifestations in early stages of sepsis. If evident symptoms were observed, multiple organ damage may have occurred, such as kidney injury and heart failure, which could even occur simultaneously. Atypical symptoms, like apnea, poor reflexes, and lethargy indicate nervous system impairment, while abdominal distension and bloody diarrhea indicate necrotizing enterocolitis or intestinal perforation, and respiratory distress and apnea always indicate pneumonia [1].

According to clinical features of neonatal sepsis above, early diagnosis and treatment are crucial for improving prognosis and reducing mortality. Hemoculture is the “gold standard” in detecting pathogens for diagnosis, however, that requires 2–5 mL blood volume from neonates and 3–5 days to maximize sensitivity (Sen) [4]. Other biomarkers, such as white blood cell count, C-reactive protein (CRP), procalcitonin (PCT), and interleukins (ILs), have been reported limited in diagnostic specificity (Spe), especially for CRP and ILs [5].

microRNAs (miRNAs) are a class of non-coding RNA molecules with an extension of 18–23 nucleotides that bind target mRNA and induce mRNA degradation or repression of translation, which is essential for development and survival [6]. miRNAs have been reported to be involved in the regulation of exacerbated inflammatory responses, impairment of endothelial function and activation of the coagulation cascade [6]. Dysregulation of several miRNAs, such as miRNA-15b, miRNA-378a, miR-211-5p, and miR-1184, was found to be differentially expressed in the serum of septic neonates [7–9].

Among the neonate populations reported by recent studies, the expression level of circulating miRNAs is inclined to show differences in demographic variation, samples, disease stages, and methodologies. Accordingly, a systematic meta-analysis of the current study could promote comprehension of the diagnostic capacity of miRNAs in neonatal sepsis.

Methods

Selection criteria

According to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [10] and Meta-analysis of Observational Studies in Epidemiology (MOOSE) reporting guidelines, we performed a meta-analysis of miRNAs in the diagnostic value of neonatal sepsis. PubMed, Cochrane Library, Embase, Web of Science, Scopus, and Ovid were screened before 15 January 2023. In lexical retrieval process, words like “Neonatal Sepsis”, “Neonatal Sepses”, “Sepses”, “Neonatal Late-Onset Sepsis”, “microRNA”, “microRNAs”, and “miRNA” were included. The inclusion criteria for the literature were as follows: (1) peripheral miRNAs were confirmed to be differentially expressed in EOS or LOS; (2) the data were publicly available, such as true positive (Tp), true negative (Tn), false positive (Fp), and false negative (Fn). The exclusion conditions were as follows: (1) reviews, letters, correspondence, expert-opinions, and editorial; (2) bioinformatic analysis, animal and cellular studies; (3) duplicated articles; (4) lacking sufficient data to obtain accurate diagnosis value; (5) the collected samples were not from blood.

Data collection

The titles and abstracts of the retrieved literature from the above databases were first screened and the full-text literature that met all the above screening conditions were selected for further analysis. Data extraction was performed including author, year of publication, region, miRNAs, specimen, regulation mode, sample size, and diagnostic power including Sen, Spe, and area under the curve (AUC).

Methodologic quality

For the methodological quality evaluation of systematic review, Quality Diagnostic Accuracy Studies-2 (QUADAS-2) was applied.

Statistical analysis

After the threshold effect evaluation, the diagnostic index, which included Sen, Spe, negative likelihood ratio (NLR), positive likelihood ratio (PLR), summary receiver operating characteristic (SROC) curve, AUC, and diagnostic odds ratio (DOR) were quantified with 95% confidence interval (95%CI). A bivariate random-effects model was used for pooled analysis of diagnostic indices; meta-regression models were used to estimate heterogeneity and publication bias was assessed by Deek’s funnel plots. All statistics were analyzed with STATA version 17 (StataCorp, College Station, TX). Using SPSS Statistics 25 (SPSS Inc., Chicago, IL), the Chi-square test was used to compare the overall accuracy (OA) value of miRNAs and PCT.

Results

Search results and methodological qualities analysis

This study found 249 original articles by retrieving six medical databases, 238 duplicates were eliminated, then bioinformatic analysis, animal or cellular studies, irrelevant articles were removed, eventually 11 full-text studies [7–9,11–18] were screened in total, which contained 16 records of miRNAs, four records of PCT. The flow of studies identified to be included in the review is displayed in Figure 1 and methodological qualities assessment is presented in Figure 2.

Figure 1. Flow diagram of the study selection.

Figure 2. A summary of methodological qualities of included articles.

Characteristics of included studies

There were 16 records of miRNAs, the control groups include two records of systemic inflammatory response syndrome (SIRS), four records of respiratory infection or pneumonia and 10 records of healthy contrast group and the septic neonates were included for suspected or proven sepsis. The details of above records are displayed in Table 1. Among the 16 records, 10 records were published in Egypt, five records were published in China, and one record in America. In addition, 15 records were obtained from peripheral blood and one from umbilical cord blood. There were four records of PCT, including three records of respiratory infection and one record of SIRS.

Table 1. Characteristics of the included studies and participants.

First author	Year	Region	Biomarker	Mode	Specimen	Case	Number 1	Control	Number 2	Sen	Spe	AUC
Marwa A. Ali	2022	Egypt	miR181b-5p	Down	Serum	Sepsis	60	Healthy	60	0.915	1.000	0.833
Marwa A. Ali	2022	Egypt	miR21-5p	Down	Serum	Sepsis	60	Healthy	60	0.924	1.000	0.917
Eman Fouda	2021	Egypt	miRNA-15b	Up	Serum	Sepsis	25	Healthy	25	0.760	0.880	0.965
Eman Fouda	2021	Egypt	miRNA-378a	Down	Serum	Sepsis	25	Healthy	25	0.600	0.880	0.952
Linda M. Ernst	2021	America	miR-211-5p	Up	Serum	Early onset neonatal sepsis	8	Healthy	21	0.625	0.900	0.787
Dan Wang	2021	China	miR-1184	Down	Serum	Sepsis	72	Respiratory infection or pneumonia	56	0.694	0.928	0.824
Dan Wang	2021	China	PCT	Up	Serum	Sepsis	72	Respiratory infection or pneumonia	56	0.694	0.928	0.824
Sally M. El-Hefnawy	2021	Egypt	miRNA-16a	Up	Serum	Suspected or proven sepsis	25	Healthy	25	0.880	0.980	0.968
Sally M. El-Hefnawy	2021	Egypt	miRNA-451	Up	Serum	Suspected or proven sepsis	25	Healthy	25	0.640	0.610	0.674
Xinyu Lin	2021	China	miR-141	Down	Serum	Sepsis	98	Respiratory infection or pneumonia	50	0.765	0.840	0.870
Xinyu Lin	2021	China	PCT	Up	Serum	Sepsis	98	Respiratory infection or pneumonia	50	0.684	0.780	0.760
Rabab F. Salim	2020	Egypt	miR-101	Up	Serum	Sepsis	50	SIRS	30	0.840	0.840	0.908
Rabab F. Salim	2020	Egypt	miR-187	Up	Serum	Sepsis	50	SIRS	30	0.720	0.760	0.789
Rabab F. Salim	2020	Egypt	PCT	Up	Serum	Sepsis	50	SIRS	30	0.720	76.700	0.856
Guozhi Liu	2020	China	miR-181a	Down	Serum	Sepsis	102	Respiratory infection or pneumonia	50	0.833	0.840	0.893
Guozhi Liu	2020	China	PCT	Up	Serum	Sepsis	102	Respiratory infection or pneumonia	50	0.755	0.980	0.874
Yue-Yan Mao	2021	China	miR-455-5p	Up	Serum	Sepsis	90	Healthy	88	0.802	0.852	0.895
Omayma O. Abdelaleem	2022	Egypt	miR-34a-5p	Down	Serum	Sepsis	90	Healthy	90	0.667	0.833	0.792
Omayma O. Abdelaleem	2022	Egypt	miR-199a-3p	Down	Serum	Sepsis	90	Healthy	90	0.666	0.644	0.701
Meng Li	2022	China	miR-129-5p	Down	Serum	Sepsis	75	Respiratory infection or pneumonia	84	0.827	0.798	0.884

Abbreviations: SIRS, systemic inflammatory response syndrome; Sen, sensitivity; Spe, specificity; AUC, area under the curve.

Threshold effect

The Spearman correlation coefficients and p values for miRNAs were obtained by the rank correlation test, the coefficient value was −0.412 (p = .113), indicating that the threshold effect did not result in heterogeneity among the included studies.

Diagnostic accuracy of miRNAs and PCT

Owing to significant heterogeneity, we selected a random-effects model to perform statistical analysis and generate data among the 16 records (I2 = 80.04% for Sen, I2 = 76.93% for Spe, and I2 = 74.7% for DOR), as shown in Figures 3 and 4. The yellow spots, which fell outside the 95%CI, were the major sources of heterogeneity in Figure 5 and both were from the same article [11]. The pooled Sen and Spe of the miRNAs were 0.87 (95%CI 0.81–0.91) and 0.79 (95%CI 0.71–0.85), respectively. The PLR and NLR were 4.1 (95%CI 2.9–5.7) and 0.17 (95%CI 0.11–0.26), respectively. The DOR of pooled studies was 24 (95%CI 12–50), the AUC for SROC was 0.9 (95%CI 0.87–0.92) as shown in Table 2 and Figure 6. Among those records, four records were used to evaluate diagnostic accuracy of PCT, the pooled Sen and Spe were 0.92 (95%CI 0.83–0.96) and 0.64 (95%CI 0.56–0.70), respectively. The PLR and NLR were 2.5 (95%CI 2.0–3.2) and 0.12 (95%CI 0.05–0.29), respectively. The DOR of pooled studies was 20 (95%CI 7–56) and the AUC for SROC was 0.74 (95%CI 0.70–0.78) as shown in Table 2. Finally, the OA value of miRNAs was 77.36% and that of PCT was 80.38%, indicating that their overall diagnostic accuracy scores of them were moderate and not statistically significant difference (p = .13).

Figure 3. Diagnostic odds ratio of included studies.

Figure 4. Forest plots of pooled sensitivity and specificity of miRNAs.

Figure 5. Heterogeneity analysis. (a) Goodness-of-fit; (b) bivariate normality; (c) influence analysis; (d) outlier detection.

Figure 6. Summary receiver operating characteristic curve value of circulating miRNAs.

Table 2. Diagnostic value of circulating miRNAs and PCT.

Index	Studies	Patients	Sen (95%CI)	Spe (95%CI)	PLR (95%CI)	NLR (95%CI)	DOR (95%CI)	AUC (95%CI)
miRNA	16	1754	0.87 [0.81, 0.91]	0.79 [0.71, 0.85]	4.1 [2.9, 5.7]	0.17 [0.11, 0.26]	24 [12, 50]	0.9 [0.87, 0.92]
PCT	4	508	0.92 [0.83, 0.96]	0.64 [0.56, 0.70]	2.5 [2.0, 3.2]	0.12 [0.05, 0.29]	20 [7, 56]	0.74 [0.7, 0.78]

Abbreviations: miRNAs, microRNAs; PCT, procalcitonin; CRP, C-reactive protein; 95%CI, 95% confidence intervals; Sen, sensitivity; Spe, specificity; PLR, positive likelihood ratios; NLR, negative-likelihood ratios; DOR, diagnostic odds ratio; AUC, area under the curve.

Meta-regression and subgroup analysis

Multiple univariate bivariate meta-regression models were used to explore the heterogeneity of included studies. The effects of each covariate on Sen and Spe were estimated as shown in Figure 7 and Table 3. The nation and regulation mode were expected to contribute significant differences in Sen and Spe (p < .05), the physical state did not affect Spe but caused difference in Sen, additionally, there were no significant differences for sample size in Sen and Spe.

Figure 7. Meta-regression and subgroup analysis.

Table 3. Summary diagnostic power based on subgroup analysis.

Parameter	Category	Studies	Sensitivity	p1	Specificity	p2
China	Yes	5	0.88 [0.81–0.95]	.04	0.75 [0.63–0.87]	.02
	No	11	0.86 [0.79–0.92]		0.81 [0.73–0.88]
Upregulated	Yes	7	0.85 [0.76–0.93]	.01	0.77 [0.67–0.88]	.04
	No	9	0.88 [0.82–0.94]		0.79 [0.71–0.88]
Healthy	Yes	11	0.86 [0.80–0.93]	.02	0.81 [0.74–0.88]	.39
	No	5	0.87 [0.79–0.95]		0.72 [0.59–0.85]
Size >100	Yes	9	0.88 [0.82–0.94]	.08	0.80 [0.73–0.88]	.17
	No	7	0.85 [0.75–0.94]		0.76 [0.64–0.87]

Publication bias

Publication bias assessment was evaluated to assess the publication bias of the included studies by funnel graph, Deek’s test result in this study indicated that there was no potential publication bias in variables, as shown in Figure 8 (p = .92 > .05).

Figure 8. Deek’s funnel plot.

Discussion

Neonatal sepsis is a systemic condition in response to invasive infection of pathogens, such as bacterial, viral, or fungal. The irreversible hemodynamic change and multiple organ impairments could result in substantial morbidity and mortality, so early detection and timely treatment are urgent to improve prognosis in the early or late periods. However, it is difficult to distinguish SIRS from sepsis [19], or local infection with systemic infection by traditional indexes, such as white blood cell count, CRP, PCT, and IL-6, which can vague disease stage and mistake therapeutic strategy. In recent years, increasing number of studies have focused on the disease prediction and prognosis value of miRNAs and analyzed the potential mechanisms by cellular tests, animal tests, bioinformatic database mining, etc. Similarly, more studies are focusing on neonatal sepsis, which is one of the major causes of mortality in neonatal period. In order to evaluate the diagnostic value of miRNAs, it is necessary to summarize recent results by systematic review and meta-analysis in neonatal sepsis.

Based on this systematic meta-analysis, the following conclusions can be obtained that the pooled Spe and DOR of circulating miRNAs were higher than those of PCT in diagnosing sepsis, but Sen was lower. The OA values of miRNAs or PCT were obviously greater than that of CRP, but there were not statistically significant difference between miRNAs and PCT. Furthermore, subgroup analysis indicated that different nations, regulation modes and health state can lead to differences in diagnosis values (p < .05).

Although there are several screening tools for inflammatory markers, such as white cell count, IL-6, CRP, and PCT, which have been applied for early diagnosis of sepsis, disease progression monitoring, and treatment efficacy determining, but none of them have been confirmed specific. For example, recent reports have pointed out that CRP had limited Spe in diagnosing sepsis, especially under conditions like neonatal asphyxia, respiratory distress, and stress response [20]. The level of PCT is a useful index in diagnosing sepsis, monitoring response to antibiotic treatment and progression phases, while within 48 h after delivery, PCT and white blood cell count can increase physiologically in neonates [21], so their Sen and Spe are comparatively lower and lack accurate cutoff values for the diagnosis of early onset neonatal sepsis [22].

Current studies have analyzed the diagnostic value of circulating miRNAs in peripheral blood. In all 16 miRNAs, miR-181a, the most studied one, which had been researched in septic groups with acute kidney and lung injury [23,24], both Liu and Chen pointed out that it was differentially downregulated in the serum of septic neonates compared with non-septic ones [15,25]. Similarly, a meta-analysis in adult group showed that overall Sen of miRNAs was 0.77, overall Spe was 0.91, AUC value was 0.87, and the DOR value was 33, indicating the potential diagnostic value of sepsis in adults [26].

Our results showed that the AUC value of miRNAs was 0.9 (95%CI 0.87, 0.92) higher than that of PTC which was 0.74 (95%CI 0.70, 0.78). However, there was no statistically significant difference in OA values between them. Although miRNAs are promising diagnostic biomarkers for sepsis, certain limitations were observed in this meta-analysis. First, heterogeneity was observed. Second, we could not control the statistical methods among the included studies, which may cause differences in the outcomes.

Conclusions

Circulating miRNAs may be potential biomarkers for distinguishing sepsis from respiratory infections or pneumonia, SIRS and healthy controls.

Author contributions

Yihong Zhao: conceptualization, methodology, software, investigation, formal analysis, and writing – original draft. Ruqin Zhu: data curation. Xiaoyan Hu: conceptualization, supervision, and writing – review and editing.

Ethics statement

An ethics statement is not applicable because this study was based exclusively on published literature.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data availability is not applicable to this article as no new data were created or analyzed in this study.

Additional information

Funding

There is no funding source for this meta-analysis.