Clinical Features and Biomarkers for Early Prediction of Refractory Mycoplasma Pneumoniae Pneumonia in Children

Objective The study aimed to analyze the clinical characteristics of children with RMPP and to explore the biomarkers for the early prediction of RMPP, thus providing references for the clinical diagnosis and treatment of RMPP in children. Methods Baseline clinical characteristics, clinical symptoms, physical examination, chest imaging, and laboratory indicators between children with RMPP and general refractory mycoplasma pneumoniae pneumonia (GMPP) were compared. Multiple logistic regression analysis was used to determine independent risk factors for RMPP. ROC curves were adopted to analyze the predictive values of biomarkers. Results The RMPP group had more severe clinical symptoms and manifestations on imaging (including pleural effusion, pulmonary consolidation, and pulmonary atelectasis), a higher incidence of extrapulmonary complications, and a longer duration of hospital stays. Results of multiple logistic regression analysis showed that serum D-dimer (OR = 8.169, P < 0.001), C-reactive protein (CRP) (OR = 1.146, P < 0.001), and lactate dehydrogenase (LDH) (OR = 1.025, P < 0.001) levels were independent risk factors for RMPP. The area under the receiver operating characteristic curve (AUROC) in RMPP prediction was 0.841, 0.870, and 0.893 for serum levels of D-dimer, CRP, and LDH, respectively (P < 0.001), with a cutoff value of 1.47 ng/ml, 39.34 mg/L, and 379 IU/L, respectively. Conclusions Serum D-dimer, CRP, and LDH levels were related to the severity of mycoplasma pneumoniae pneumonia in children and had potential as biomarkers for the early prediction of RMPP, suggesting great applicative values for the early diagnosis and timely intervention of children with RMPP in clinical practice.


Introduction
Mycoplasma pneumoniae pneumonia (MPP) is a common disease of the respiratory system in children caused by mycoplasma pneumoniae (MP), with an incidence of approximately 10%-40% [1,2].MP is one of the most common causes of community-acquired pneumonia in children and adolescents.In recent years, the risk of MP infection has increased in the Chinese population.
Epidemiological studies have shown that MPP has an epidemic outbreak every 3-4 years in European countries [3,4].More than 85% of MP strains among pediatric patients in China have been reported as macrolide-resistant MP, which can potentially cause severe and even extrapulmonary diseases.Te global increase in macrolideresistant MP is of concern due to limited therapeutic options [1].Based on this fact, MPP is divided into refractory mycoplasma pneumoniae pneumonia (RMPP) and general mycoplasma pneumoniae pneumonia (GMPP) [5].Te prevalence of RMPP in children was 14.30%, with 6.83%, 20.86%, and 40.84% in those aged <4 years, 4-7 years, and ≥7 years, respectively [6].
Te clinical presentation and manifestations of RMPP vary widely in diferent individuals and can afect all organs of the body.Children with RMPP may be complicated by pleural efusion, pulmonary atelectasis, gas accumulation in the mediastinum, pneumothorax, and necrotizing pneumonia.Some children with RMPP may develop respiratory distress, followed by rapid deterioration of their pulmonary functions, even requiring mechanical ventilation or extracorporeal lung support using extracorporeal membrane oxygenation (ECMO) [7], resulting in death.After conventional treatments, a recurrence of pulmonary lesions or a prolonged course of disease may still be observed, contributing to structural and/or functional lung abnormalities manifested by bronchiectasis [8].In fact, these conditions are often associated with recurrent pulmonary infections in children and even exert a signifcant impact on pulmonary function in adults.To prevent progression and reduce relevant complications, early recognition and diagnosis are crucial for the appropriate treatment of RMPP in patients who are prone to clinical and radiological exacerbations during macrolide therapy.With the increase in the incidence and mortality of RMPP, the prevention of high-risk patients with RMPP has become a major concern in clinical practice [9,10].
Currently, the diagnosis of RMPP has been challenging, with few signifcant features detected in laboratory or radiological evaluations, suggesting that there is no specifc tool available for the diagnosis of RMPP [11].Studies have shown that C-reactive protein (CRP), lactate dehydrogenase (LDH), erythrocyte sedimentation rate (ESR), percentage of neutrophils (NEPs), and the percentage of lymphocytes, together with the presence of dense solid pulmonary shadows, were signifcant predictors of RMPP [12][13][14][15].Based on the above, this study was aimed to investigate clinical characteristics and explore biomarkers for early prediction of RMPP in children, providing references for the establishment of an efcient protocol for the early diagnosis of RMPP.Te study was approved by the Ethics Committee of Fujian Children's Hospital, Fujian Medical University.Patients were appropriately informed about treatment decisions.Informed consent was obtained from all patients.

Grouping of Patients.
All patients were divided into 2 groups, including the GMPP group and the RMPP group.Te diagnosis of RMPP was based on the presence of persistent fever (≥37.5 °C) accompanied by clinical and radiological deterioration after azithromycin treatment for ≥7 days [9,15].Patients who met the diagnosis of RMPP were allocated into the GMPP group; others were classifed into the GMPP group.

Data Collection.
In the study, patient information was collected, including baseline clinical characteristics, laboratory results, and radiological fndings.Baseline clinical characteristics, including age, sex, month and season of onset, hospital stay, clinical symptoms and signs, and extrapulmonary manifestations, were collected from both groups of children.Within 24 h of admission, all children were tested for respiratory pathogens and 2-3 ml of fasting venous blood was drawn for relevant laboratory tests, including MP-specifc antibody titer tests, WBC count, percentage of NEP, CRP, LDH, procalcitonin (PCT), D-dimer, and tumor necrosis factor alpha (TNF-α).Imaging examinations mainly included chest radiographs and/or CT throughout the course of the disease, from which the extent of involvement, type of lesions, and intrapulmonary complications such as pulmonary atelectasis, pleural efusion, and pulmonary necrosis were collected.

Comparison of Clinical Characteristics between the RMPP and GMPP Groups.
In this study, 8 of 476 patients withdrew from the study.A total of 468 children with MPP were included fnally, of which 156 (33.33%) were in the RMPP group and 312 were in the GMPP group.
As shown in Table 1, there were 55.77% males and 51.28% females in the RMPP group, with a mean age of 6.23 ± 2.89 years.RMPP occurs primarily in summer (36.54%) and in autumn (30.13%).there were no signifcant diferences between the two groups in terms of sex, age, and season of onset (P > 0.05).
In terms of clinical symptoms, all patients in the RMPP group had fever, and the proportion of shortness of breath was signifcantly higher in the RMPP group than in the GMPP group (23.08% vs. 6.41%,P < 0.001); in contrast, the proportions of runny nose (36.90% vs. 8.50%, P � < 0.001) and gastrointestinal symptoms (26.92% vs.18.59%,P � 0.047) in the GMPP group were signifcantly higher than those in the RMPP group.However, the percentage of cough was as high as 94.87% in the RMPP group and 98.08% in the GMPP group, but their diference was not statistically signifcant (P > 0.05).
In terms of physical examination, the proportion of three concave signs (manifested as the suprasternal fossa, supraclavicular fossa, and concave intercostal space) was signifcantly higher in the RMPP group than in the GMPP group (10.90% vs. 2.56%, P < 0.001).Te proportion of moist rales was 52.56% in the RMPP group and 46.79% in the GMPP group, but their diference was not statistically signifcant (P > 0.05).
Extrapulmonary complications refer to the symptoms and signs of damage to other systems other than the respiratory system, including the digestive system, cardiovascular system, blood system, nervous system, urinary system, skin sores, and joint pain.
Te mean duration of hospital days in the RMPP group was also signifcantly longer than that of the GMPP group (11.09 ± 4.25 vs. 8.27 ± 3.12, P < 0.001).

Comparison of Laboratory Indicators between the RMPP and GMPP Groups.
As shown in Table 2, comparisons of laboratory indicators between the RMPP and GMPP groups showed that serum levels of CRP, LDH, PCT, and D-dimer were signifcantly higher in the RMPP group than in the GMPP group.However, the percentage of NEPs was signifcantly higher in the GMPP group than in the RMPP group.Diferences in WBC count and TNF-α between the two groups were not statistically signifcant (P > 0.05).

Predictive Values of Serum D-Dimer, CRP, and LDH
Levels for RMPP Using ROC Curves.To explore the predictive value of serum D-dimer, CRP, and LDH levels for RMPP, the ROC curves were plotted, and the AUC was calculated.Te results showed good predictive values of serum D-dimer, CRP, and LDH levels for RMPP, with AUROC of 0.841, 0.870, and 0.893, respectively (Figure 2).Te optimal cutof point was determined according to the Youden index.Specifcally, the optimal cutof point of 1.47 ng/ml for the serum D-dimer level revealed a sensitivity of 64.74% and a specifcity of 98.72% for the detection of RMPP, the optimal cutof point of 39.34 mg/L for the serum CRP level revealed a sensitivity of 60.89% and a specifcity of 94.55% for the detection of RMPP, and the optimal cutof point of 379 IU/L for the serum LDH level revealed a sensitivity of 66.67% and a specifcity of 93.91% for the detection of RMPP (Table 4).

Discussion
RMPP is usually characterized by a long course of disease, a poor therapeutic efcacy, and numerous complications that can even endanger the lives of children.Te present study found that serum levels of D-dimer, CRP, and LDH were independent risk factors for RMPP, which laid a basis for early identifcation of RMPP and may be of great help in the diagnosis and prognosis of these children.
As reported, the most common clinical symptoms of RMPP were composed of cough (no sputum at the beginning and small to moderate bloodless sputum later), fever, Emergency Medicine International chills, sore throat, headache, hoarseness, myalgia, and general malaise [16] and would worsen after 7 days of macrolide therapy, accompanied by persistent fever, pulmonary exacerbation in radiological fndings, and extrapulmonary complications [17].All of this was consistent with the clinical symptoms of children with RMPP observed in this study.Additionally, we also found that clinical symptoms were more severe in the RMPP group compared to the GMPP group, with fever observed in all children and a percentage of shortness of breath of up to 23.08% in the RMPP group.Furthermore, compared to those with GMPP, the images also presented with more severe manifestations in children with RMPP, with proportions of pulmonary consolidation, pulmonary atelectasis, and pleural efusion of 79.49%, 5.77%, and 30.77%, respectively.
In this study, 126 children (37.18%) with RMPP had extrapulmonary complications, which was signifcantly more than those in the GMPP group.Furthermore, the mean duration of hospital stays was 11.09 ± 4.25 in the RMPP group, which was also higher than the GMPP group with statistical signifcance.Te abovementioned fndings were in agreement with those reported in previous studies.Gong et al. [15] identifed that persistent fever (>10 days), pleural efusion, extrapulmonary complications, pulmonary consolidation detected in chest radiography, and CRP > 40 mg/L could be used for early evaluation of RMPP by using a fxedefects model or a random-efects model.In addition, Choi et al. [18] showed that respiratory distress, oxygen saturation <90%/cyanosis, oxygen support during hospitalization, lobar pneumonia on admission, and extrapulmonary Te pathogenic mechanisms of RMPP are complex and include mainly direct pulmonary cell injury and immune response-induced injury.Currently, RMPP is considered to be related to airway mucus hypersecretion, hypercoagulable state, bacterial or viral infection, and excessive immune response due to the communityacquired respiratory distress syndrome (CARDS) toxin [19,20].Based on relevant basic research results and literature reports, this study selected 7 highly correlated biomarkers for the study of RMMP prediction indicators.CRP is one of the important indicators of infammatory response [21], which has been widely used for assessing disease severity and treating infammatory conditions.In addition, serum D-dimer levels have been recognized as a specifc marker of the fbrinolytic system and an indicator of monitoring infammations and severe infections [22].Te elevated level of D-dimer is possibly attributed to the injury of vascular endothelial cells caused by the excessive infammatory response, which may be related to the mechanism of pulmonary injury in RMPP.Additionally, increased LDH activity has been found to be associated with pulmonary infammation and hypoxia.High serum LDH levels showed the potential to predict an inadequate response to glucocorticoid treatment [23].
In this study, the results of multiple logistic regression analysis showed that serum D-dimer, CRP, and LDH levels were independent risk factors for RMPP; AUROCs for serum D-dimer, CRP, and LDH levels in the prediction of RMPP were all more than 0.8 (0.841, 0.870, and 0.893, respectively), suggesting their good predictive values in RMPP.Although the sensitivity of these cutofs is lower than 65%, the specifcity is higher than 93%, which can efectively exclude non-RMPP patients, thereby reducing the misdiagnosis rate.Similar to these fndings, Zhang et al. also observed that the levels of CRP, LDH, and interleukin-6 (IL-6) were signifcantly higher in patients with RMPP than those with GMPP, indicating that they may be important predictors of RMPP in children and could facilitate the early identifcation of RMPP [21].Meanwhile, the optimal cutof points for serum D-dimer, CRP, and LDH levels for detecting RMPP were found to be 1.47 ng/ml, 39.34 mg/L,  Emergency Medicine International and 379 IU/L, respectively, which were comparable to the values of a case-control study (2.10 ng/ml, 343.08 mg/L, and 375 IU/L for serum levels of D-dimer, CRP, and LDH, respectively) [24].Based on the abovementioned results, it could be inferred that serum levels of D-dimer, CRP, and LDH may have a certain value for clinical application.
Tere were several limitations in this study.First, as a single-center study, this study may have selection bias compared to multicenter studies.Second, a paired design was not performed and the patients in the RMPP group were not matched with those in the GMPP group for certain parameters, which may afect the statistical efcacy of the results.Finally, no joint prediction using multiple indicators was performed.Terefore, in the future, a multicenterpaired study should be conducted to further study the joint prediction of RMPP based on multiple indicators.

Conclusions
In conclusion, serum levels of D-dimer, CRP, and LDH are independent risk factors for RMPP and have high specifc predictive values though low sensitivity for the early identifcation of RMPP.Early detection of RMPP within 24 hours of hospital admission may guide therapy revision for patients to reduce mortality.

Figure 2 :
Figure 2: ROC curves for D-dimer, CRP, and LDH for the prediction of RMPP.
Statistical analysis was performed using SPSS 23.0.Continuous data with normal distribution were described with mean ± standard deviation (SD), and Student's t-test was used for comparison between the groups.Continuous data with a skewed distribution were presented with median and interquartile range (IQR), and the Mann-Whitney U test was used for comparison between the groups.Categorical data were expressed as frequency and percentage (%), and the chi-square test was used for comparison between the groups.Laboratory indicators of statistical signifcance in the comparisons were included as risk factors in the prediction of RMPP by using stepwise backward logistic regression.Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated to evaluate the predictive value of laboratory indicators for RMPP.A two-sided P < 0.05 was considered as statistically signifcant.

Table 1 :
Comparison of clinical characteristics between the RMPP and GMPP groups (n � 468).

Table 2 :
Comparison of laboratory indicators between the RMPP and GMPP groups (n � 468).

Table 3 :
Logistic regression analysis for risk factors of RMPP.

Table 4 :
Predictive values of serum D-dimer, CRP, and LDH levels for RMPP.