Serum IL-17A and IL-6 in paediatric Mycoplasma pneumoniae pneumonia: implications for different endotypes

ABSTRACT

Paediatric Mycoplasma pneumoniae pneumonia (MPP) is a heterogeneous disease with a diverse spectrum of clinical phenotypes. No studies have demonstrated the relationship between underlying endotypes and clinical phenotypes as well as prognosis about this disease. Thus, we conducted a multicentre prospective longitudinal study on children hospitalized for MPP between June 2021 and March 2023, with the end of follow-up in August 2023. Blood samples were collected and processed at multiple time points. Multiplex cytokine assay was performed to characterize serum cytokine profiles and their dynamic changes after admission. Cluster analysis based on different clinical phenotypes was conducted. Among the included 196 patients, the levels of serum IL-17A and IL-6 showed remarkable variabilities. Four cytokine clusters based on the two cytokines and four clinical groups were identified. Significant elevation of IL-17A mainly correlated with diffuse bronchiolitis and lobar lesion by airway mucus hypersecretions, while that of IL-6 was largely associated with lobar lesion which later developed into lung necrosis. Besides, glucocorticoid therapy failed to inhibit IL-17A, and markedly elevated IL-17A and IL-6 levels may correlate with lower airway obliterans. Our study provides critical relationship between molecular signatures (endotypes) and clustered clinical phenotypes in paediatric patients with MPP.

KEYWORDS:

• Paediatric Mycoplasma pneumoniae pneumonia
• cytokine profile
• endotype
• phenotype
• lower airway obliterans

Introduction

Mycoplasma pneumoniae (MP) is one of the most common pathogens of community-acquired pneumonia (CAP) in school-aged children [1]. MP pneumonia (MPP) has been considered as major type of CAP among East Asian children given the increasingly reported severe/refractory cases in recent years, especially in China [2–5] where MPP is relatively more critical and prone to develop lower airway obliterans (LAO) [3], the chronic airway disease accounting for impaired lung function and restricted lung development, thus the quality of life. Of note, there was a surge of MP infections hitting Chinese children hard in 2023 that raised wide concerns [6]. Therefore, it is necessary that paediatricians attach more importance to the clinical management of MPP.

Although it is well known that excessive host immune response and direct damage of MP to respiratory epithelium play a pivotal role in the pathogenesis of MPP [7], the specific mechanisms remain unclear. Studies have demonstrated that clinical severity (clinical phenotype) of MPP varies among infected individuals [2,3,8], which could be interpreted by different degrees of immune inflammatory response elicited by MP. Tanaka et al. reported distinct radiological and pathological findings among patients with MPP based on their cell-mediated immunity levels [9,10], suggesting the underlying heterogeneity from molecular perspectives that correlates with clinical phenotypes after MP infection. This may explain the fact that some children with severe MPP (SMPP) present with lobar lesions on chest imaging while the others with diffuse bronchiolitis [3].

Cytokine profiles could be applied to evaluate disease-related molecular signatures (endotypes) as they convey potential cues of immunological and inflammatory state inside the body. In fact, extensive studies have already explored cytokine profiles in patients with MPP. However, all the studies focused on the values of cytokines for discrimination between severe/refractory and general cases, and for assistance in MPP diagnosis [11–14]. Few have noted the underlying association between the intrinsic signatures and clinical phenotypic heterogeneity. Furthermore, since the clinical dilemma of MPP-related LAO has not yet been overcome under our current therapeutic regimen including different doses of glucocorticoid (GC) and fiberoptic bronchoscopy (FOB) [3], pioneering interventions based on varied endotypes may be required.

The aim of this study was to evaluate cytokine profiles (mainly interleukins) in paediatric MPP. We performed cytokine detections both during acute stage upon admission and late stage after admission, and we conducted analyzes on the dynamic changes of specific cytokine levels for investigating their relationships with clinical phenotypes and prognosis in the infected children, so as to provide basis for future promising therapeutic strategies for MPP.

Patients and methods

Study population

We conducted a prospective longitudinal real-world study on 196 children hospitalized for acute MPP at the Respiratory Department of Beijing Children’s Hospital and the Respiratory Department of Third Affiliated Hospital of Zhengzhou University between June 2021 and March 2023. All enrolled patients had clinical symptoms suggestive of MP infection, including fever and cough. The diagnostic criteria of MPP were based on our previous study [3]: (1) serum anti-MP IgM titre ≥1:320, or/and the titre of anti-MP IgM increased by four times or more in the recovery stage than the acute stage (latex agglutination test); and (2) positive MP RNA results on RT-PCR of pharyngeal swab. MPP was confirmed if the above two were both met. The exclusion criteria were as follows: (1) disease course of <6 days or >10 days on admission; (2) lack of chest imaging during 6–10 day of disease course; (3) past history of asthma, tuberculosis, chronic malnutrition, aspiration, immunodeficiency, cystic fibrosis, primary ciliary dyskinesia, or bronchopulmonary dysplasia; (4) definite bacterial and/or adenovirus coinfection on admission; and (5) loss to follow-up (Figure 1).

Figure 1. Flowchart of the study.

The study protocol was reviewed and approved by the Medical Ethics Committee of Beijing Children’s Hospital, Capital Medical University (Approval No. [2022]-E-089-Y). Written informed consent was obtained from the patients or their legal guardian/next of kin prior to enrolment.

Patient grouping

The 196 patients were clustered and divided into three groups according to their chest imaging features [3,9,10]: those with a lobar lesion affecting at least half of a lobe (Group A); those with diffuse bronchiolitis affecting at least two lobes which were represented by centrilobular nodules, bronchiolar wall thickening, “tree-in-bud” signs and ground-glass opacities (Group B), and those with slight lesion including patchy shadows and/or cloudy opacities confined to less than half of a lobe (Group C). We further subdivided Group A into Group A1 and Group A2 based on the occurrence of lung necrosis indicated by imaging re-examination during late stage of disease course (Figure 1) [15].

Blood sampling

Blood samples were collected from all patients upon admission, and multiple blood samples were dynamically collected in about half of these patients (96/196, 49.0%). Specifically, the samples were collected twice in 51 patients (on admission; 3–4 days after admission) and thrice in 45 patients (on admission; 3–4 days after admission; 7–8 days after admission). Blood samples were collected by venepuncture into vacuum tubes (Greiner Bio-One Vacuette®) containing separation gel, then were centrifuged at 4000rpm×10 min at 4°C (Kubota S700 T Centrifuge, Fujioka, Japan) within 2 h, then stored at 4°C. Considering the possibility of degradation of cytokine concentrations resulted from varied storage duration, the upper layer of serum was extracted and transferred into sterile cryogenic tubes using micropipettes within 3 days of storage, which was prior to freezing at −80°C immediately for further use without freeze/thaw cycles.

Multiplex cytokine assay

Serum cytokine measurement was achieved using cytokine test kits (AtomLife, Nanjing, China) according to manufacturer’s protocol. Briefly, 25 μL of immunomagnetic beads mixture and 150 μL of wash buffer were added to 96-well plate, which was then shaked and washed. 25 μL of detection antibody, 25 μL of phosphate buffer solution (PBS) and 25 μL of samples were added in sequence, and the plate sealed for incubation at 800rpm× 60 min at room temperature. Then, the plate was washed twice, incubated for 5 min following addition of 50 μL of streptavidin-peroxidase (PE) to each well. The plate was washed twice again and resuspended with 150 μL of wash buffer. After that, the plate was immediately run on Luminex MAGPIX system (Luminex xMAP, Austin, Texas, USA), with the lower detection limit setting at 50 beads per analyte. Finally, the levels of cytokines were calculated by standard curves based on generated mean fluorescence intensity (MFI). The upper limits of reference values of the cytokines were shown in Figure 2. Values greater than or equal to twice the upper limits, and within the maximum concentrations of standards were considered significantly elevated (positive).

Figure 2. Box-and-whisker plot for the overall distribution of the levels of 10 serum cytokines on admission (during 6–10 day after disease onset, n = 196). Median 31.96 4.92 4.36 6.50 3.20 14.49 9.55 3.01 6.44 5.15. P25 5.70 2.10 4.00 3.09 3.00 6.49 4.00 3.00 5.00 4.00. P75 74.68 19.97 13.48 17.74 5.10 32.03 26.81 3.64 14.49 12.05. UL 19.00 7.00 8.20 11.50 8.70 9.10 8.40 12.30 16.20 8.00. Boxes in red represent the cytokines with the most remarkable variation among different patients, compared with the other eight cytokines. The lower and upper whiskers represent 5th and 95th percentiles, respectively. Each dot beyond the whiskers represents a patient. Data on the vertical axis are log-transformed. Data below are shown in pg/mL. UL: upper limit.

Fiberoptic bronchoscopy (FOB)

Safety of the procedure was evaluated for all these patients in priority including the exclusion of bleeding diathesis, haemodynamic instability and cardiac arrhythmia. After fasting for > 6–8 h, patients were given intramuscular atropine (0.01–0.03 mg/kg, total dose ≤ 0.5 mg) 30 min before FOB for inhibition of airway secretions. 1% lidocaine was applied to upper airway for topical anaesthesia, and intravenous midazolam (0.2 mg/kg, total dose ≤ 5 mg) was given for sedation. In supine position, the fiberoptic bronchoscope (Olympus Optical Co. Ltd, Tokyo, Japan) was inserted nasally and wedged in the subsegmental bronchus of the most affected lobe according to chest imaging. For those whose extensive secreted mucus was difficult to be removed by regular lavage, disposable cytology brushes (Olympus Optical Co. Ltd, Tokyo, Japan) were applied. Vital signs including heart rate, respiratory rate and saturation of pulse oxygen were monitored continuously during the procedure, and appropriate oxygen supply was to be given in case of hypoxia.

Diagnostic criteria of clinical outcomes

1. Curative criteria (if all the followings were met): (1) there was no bronchitis obliterans under FOB during hospitalization, (2) chest imaging showed good recovery after 6 months of disease onset, (3) no airflow obstruction was detected by PFT after 6 months of disease onset.

2. Bronchitis Obliterans (if one of the followings was met) [16]: (1) bronchial lumen occlusion was found under the bronchoscopy during hospitalization, (2) atelectasis accompanied by bronchiectasis or not on chest X-ray or chest CT reexamination after 6 months of disease onset, with no obvious improvement compared with that from acute stage.

3. Bronchiolitis Obliterans (if all the followings were met) [17]: (1) clinical symptoms such as coughing, dyspnoea, wheezing, hypoxaemia persisted for at least 6 weeks after MP infection, (2) mosaic pattern and/or air trapping on chest X-ray or chest CT reexamination, (3) airflow obstruction detected by PFT, (4) exclusion of acute asthma exacerbation during chest imaging reexamination.

Statistical analyzes

Continuous variables with non-normal distribution were recorded as the median (IQR, 25th–75th percentile) and compared by Mann–Whitney U test (between two groups) or Kruskal–Wallis test (among three groups). Continuous variables with normal distribution were recorded as the mean ± standard deviation (SD) and compared by Student’s t test. Categorical variables were recorded as numbers (%) and compared by Fisher’s exact probability test or chi-square test, as appropriate. Two-sided P < 0.05 was considered to be statistically significant. GraphPad Prism 9 (GraphPad Software Inc., USA) was used for the statistical analysis and preparation of figures.

Results

Remarkable variabilities in serum IL-17A and IL-6 in patients with MPP

Profiles of 10 serum cytokines on admission are shown in Figure 2. The levels of IL-17A (31.96 [5.70, 74.68] pg/mL) and IL-6 (4.92 [2.10, 19.97] pg/mL) indicated remarkable variabilities in these patients (boxes in red), whereas the levels of the other eight cytokines showed relatively concentrated distributions. About half of the patients (89/196, 45.4%) were characterized by significantly elevated IL-17A levels (77.73 [57.13, 131.10] pg/mL), while nearly one-third of them (57/196, 32.4%) revealed significantly elevated IL-6 levels (40.02 [27.27, 89.18] pg/mL). Further, IL-17A alone (without IL-6) was significantly elevated in 71 patients (71/196, 36.2%), who were labelled as “Type high-IL17A” (Cluster I); IL-6 alone (without IL-17A) was significantly elevated in 36 patients (36/196, 18.4%) labelled as “Type high-IL6” (Cluster II). Both IL-17A and IL-6 were significantly elevated in 21 patients (21/196, 10.7%) labelled as “Type high-IL17A + IL6” (Cluster III). Finally, 68 patients (68/196, 34.7%) showed neither significant elevation in IL-17A nor IL-6 on admission, who were labelled as “Type-N” (Cluster IV).

Since the first blood samplings from all patients were consistently performed during 6–10 day after disease onset, i.e. the acute stage of MPP, the influence of disease course on the cytokine levels was largely avoided. Pronounced variations of the two cytokines among these patients suggested underlying heterogeneous molecular signatures (endotypes).

Serum IL-17A and IL-6 in patients with MPP harbouring varied clinical phenotypes

The clinical features of clustered patients in Groups A–C were shown in Table 1. Although there were no significant differences in gender composition and age of onset among the three groups, patients in Group A (n = 92) presented with much more severe fever than those in Group B (n = 39) and Group C (n = 65). The proportion of patients with wheezing or dyspnoea was higher in Group A than in the other groups. Circulating inflammatory markers (during 6–10 day of disease course) including C-reaction protein (CRP), neutrophils, lactate dehydrogenase (LDH) and D-dimer were significantly elevated, whereas serum albumin was sharply decreased in Group A compared with the other groups (all P < 0.0001).

Table 1. Clustering features in three groups of patients with MPP.

	Group A (n = 92)	Group B (n = 39)	Group C (n = 65)	P*
IL-17A, pg/mL	49.92 (8.40, 77.68)	98.88 (54.90, 137.40)	7.20 (5.00, 18.75)	< 0.05 ^I
IL-6, pg/mL	19.50 (3.50, 50.35)	4.00 (2.30, 8.60)	2.63 (2.00, 4.92)	< 0.0001^II
Male gender, no. (%)	49 (53.3)	26 (66.7)	34 (52.3)	> 0.05
Age of onset, yr	7.8 (6.1, 9.3)	7.1 (4.1, 9.0)	7.6 (5.5, 9.5)	> 0.05
Atopy, no. (%) ^a	39 (41.1)	30 (76.9)	29 (44.6)	< 0.01 ^III
Hyperpyrexia duration, days ^b	5.5 (3.0, 7.0)	2.0 (1.0, 5.0)	2.0 (0, 3.0)	< 0.0001 ^II
Wheezing/dyspnoea, no. (%) ^c	25 (27.2)	17 (32.7)	4 (6.2)	< 0.001 ^IV
Chest imaging features
Lobar lesion ≥1/2 lobe	Y	N	N	/
Diffuse bronchiolitis ≥ 2 lobes	N	Y	N	/
Slight lesion ^d	N	N	Y	/
CRP, mg/L	19.50 (9.08, 67.75)	8.00 (2.00, 10.00)	8.00(4.00,12.64)	< 0.0001 ^II
Neutrophils, %	72.20(63.10, 81.50)	62.00 (52.60, 70.00)	61.30(51.75,71.90)	< 0.0001 ^II
LDH, U/L	419.0 (287.8, 629.3)	289.0 (252.0, 360.0)	266.0(235.0,301.0)	< 0.0001 ^II
ALB, g/L	35.75 (33.40, 40.00)	40.10 (38.50, 42.60)	40.40(38.55,42.80)	< 0.0001 ^II
D-dimer, mg/L ^e	0.880 (0.276, 2.520)	0.201 (0.175, 0.313)	0.261(0.166,0.490)	< 0.0001 ^II

Data shown as median (IQR) unless otherwise indicated. All laboratory indexes reflect examinations on admission (during 6–10 day after disease onset).

*: Kruskal–Wallis test for continuous variables with non-normal distribution; Fisher’s exact test or chi-square test for categorical variables. Abbreviations: CRP, C-reactive protein; LDH, lactate dehydrogenase; ALB: albumin; Y: yes. N: no.

I: Pairwise comparisons between three groups.

II: Significant differences between Group A and the other two groups.

III: Significant differences between Group B and the other two groups.

IV: Significant differences between Group C and the other two groups.

a: Refers to past history and/or parental history of allergic diseases.

b: Refers to consecutive days of hyperpyrexia (≥ 39.0°C) before admission.

c: Before admission.

d: Refers to patchy shadows and/or cloudy opacities confined to < 1/2 lobe.

e: Missing data for 12% of observations in Group A and < 10% of that in both Group B and Group C.

We overlapped the three groups with the four cytokine-based clusters mentioned above. Most patients in Group B matched with Cluster I (35/39, 89.7%) but not Cluster II. All patients in Group C matched with Cluster IV (65/65, 100%). Interestingly, two populations that matched with Cluster I (36/92, 39.1%) or Cluster II (36/92, 39.1%) coexisted in Group A, indicating that there was inflammatory heterogeneity among these patients with SMPP.

Accordingly, we sought to uncover correlations between clinical phenotypes and these intriguing endotypic characteristics within Group A. As is revealed in Table 2, patients in Group A2 (n = 27) were characterized by prominent hyperinflammatory state which was reflected in more critical symptoms including prolonged high fever and severe hypoxaemia, dramatically elevated inflammatory markers, moderate-to-massive pleural effusions as well as the subsequent lung necrosis, thus demonstrating markedly phenotypic heterogeneity compared with those in Group A1 (n = 65) (all P < 0.01). Additionally, we observed prolonged time to defervescence as well as to lung lesion absorption after admission in Group A2 (both P < 0.0001). Despite relatively lower inflammatory response, all patients in Group A1 presented with extensive airway mucus hypersecretions which led to the formation of mucus plugs or plastic bronchitis resulting in bronchial lumen obstruction, while mucosal necrosis was rarer compared with those in Group A2 (P < 0.0001). However, there was no significant difference in the proportion of patients with airway mucus hyperproduction between the two subgroups (P > 0.05).

Table 2. Clustering features in two subgroups of SMPP with lobar lesion.

	Group A1 (n = 65)	Group A2 (n = 27)	P*
Male gender, no. (%)	36 (55.4)	14 (51.9)	> 0.05
Age of onset, yr	7.8 (6.1, 9.2)	7.7 (5.9, 9.3)	> 0.05
Atopy, no. (%)	25 (38.5)	9 (33.3)	> 0.05
Hyperpyrexia duration, days	5.0 (3.0, 7.0)	7.0 (6.0, 8.0)	0.0002
Time to defervescence, days^&	1.5 (1.0, 3.0)	3.0 (2.0, 5.0)	< 0.0001
Wheezing/dyspnoea, no. (%)	12 (18.5)	13 (48.1)	0.0088
PTE, no. (%)	1 (1.5)	4 (14.8)	0.0248
Pleural effusions, no. (%) ^+	0	21 (77.8)	< 0.0001
CRP, mg/L	10.69 (8.00, 29.27)	87.00 (48.00, 138.00)	< 0.0001
Neutrophil, %	67.40 (60.40, 78.35)	79.90 (73.40, 83.40)	0.0001
LDH, U/L	320.0 (267.1, 561.0)	627.0 (476.2, 727.2)	< 0.0001
ALB, g/L	37.90 (34.85, 40.95)	32.80 (31.00, 35.60)	< 0.0001
D-dimer, mg/L	0.470 (0.217, 1.447)	2.659 (1.967, 4.150)	< 0.0001
First FOB after admission
Mucus plugs, no. (%) ^#	57 (87.7)	23 (85.2)	> 0.05
Plastic bronchitis, no. (%)	8 (12.3)	4 (14.8)	> 0.05
Mucosal necrosis, no. (%)	4 (6.2)	16 (59.3)	< 0.0001
Lung necrosis, no. (%)	0	27 (100)	/
Lesion absorption, days ^++	5.5 (4.0, 7.5)	7.0 (5.0, 9.0)	< 0.0001

Data shown as median (IQR) unless otherwise indicated. All laboratory indexes reflect examinations on admission (during 6–10 day after disease onset).

*: Mann–Whitney U test for continuous variables with non-normal distribution; Student’s t test for continuous variables with normal distribution; Fisher’s exact test or chi-square test for categorical variables. Abbreviations: PTE, pulmonary thromboembolism; FOB, fiberoptic bronchoscopy.

&: After admission.

+: Refers to moderate-to-massive pleural effusions that required thoracentesis.

#: Refers to secreted mucus taking up over half or the complete lumen of the affected bronchial subsegments.

++: Refers to time required for the lung lesion to be reduced to 50% of its original size.

As expected, we next confirmed strong variabilities of IL-17A and IL-6 between the two subgroups. Specifically, IL-17A in Group A1 was significantly higher whereas IL-6 in this subgroup was notably lower than those in Group A2 (both P < 0.0001, Table 3, Figure 3a, b). Further, we noticed that over half of patients in Group A1 (36/65, 55.4%) matched with Cluster I, 40% of whom matched with Cluster II and III, and only a few (3/65, 4.6%) matched with Cluster IV. Surprisingly, an overwhelming majority of patients in Group A2 (24/27, 88.9%) matched with Cluster II but not Cluster I (Table 3). Besides, there was no pronounced difference in elevated IL-17A between Group A1 (n = 50) and Group B (n = 39) (74.05 [57.18, 119.30] vs. 98.88 [54.90, 137.40], P > 0.05, Figure 3c), while elevated IL-6 in Group A2 (n = 27) was much higher than that in Group A1 (n = 26) (89.33 [49.70, 144.70] vs. 29.05 [19.07, 38.57], P < 0.0001, Figure 3d). Taken together, our data revealed that distinct circulating endotypic signatures may correlate with the underlying subphenotypes in patients with a lobar lesion in SMPP.

Figure 3. Levels of serum IL-17A and IL-6 on admission (during 6–10 day after disease onset). (a,b) Levels of serum IL-17A and IL-6 among four groups. Group A1, A2, B and C: n = 65, 27, 39, and 65, respectively. (c) Levels of significantly elevated serum IL-17A between Group A1 (n = 50) and Group B (n = 39). (d) Levels of significantly elevated serum IL-6 between Group A1 (n = 26) and Group A2 (n = 27). Data on the vertical axis are log-transformed. Mann–Whitney U test for continuous variables with non-normal distribution. ****P < 0.0001, ns: no significance.

Table 3. Proportions of cytokine-based clusters in four groups of patients.

	Group A1 (n = 65)	Group A2 (n = 27)	Group B (n = 39)	Group C (n = 65)
IL-17A, pg/mL	62.57 (40.09, 85.10)	5.00 (3.00, 16.80)	98.88 (54.90, 137.40)	7.20 (5.00, 18.75)
IL-6, pg/mL	7.10 (2.05, 23.44)	89.33 (49.70, 144.70)	4.00 (2.30, 8.60)	2.63 (2.00, 4.92)
Cluster I	36 (55.4)	0	35 (89.7)	0
Cluster II	12 (18.5)	24 (88.9)	0	0
Cluster III	14 (21.5)	3 (11.1)	4 (10.3)	0
Cluster IV	3 (4.6)	0	0	65 (100)

Data shown as median (IQR) for levels of cytokines. Cluster I: Type high-IL17A; Cluster II: Type high-IL6; Cluster III: Type high-IL17A + IL6; Cluster IV: Type-N.

Serum IL-17A and IL-6 in varied phenotypes of MPP after treatment

For all patients receiving dynamic blood sampling (n = 96), conventional anti-MP therapy (intravenous azithromycin) was administered after admission, and different dosages of GC (0–5 mg/kg/day, intravenous methylprednisolone) were given depending on clinical severity. As expected, the levels of the two cytokines in Group C were not significantly elevated throughout (n = 18, data not shown). Nonetheless, IL-17A failed to decrease significantly in Group A1 (n = 33) 3–4 day after admission (Table 4, Figure 4a), wherein 14 patients received 5 mg/kg/day GC before second blood sampling. In parallel, mucus hypersecretion was still observed under the second FOB although their body temperature had returned to normal. On the other hand, patients in Group A2 (n = 15) all received 5 mg/kg/day GC after admission. Interestingly, their IL-6 rapidly dropped to normal range (Table 4, Figure 4c), and their IL-17A were not markedly elevated throughout (data not shown). However, we also found the inconsistency between defervescence and continuous airway obstruction. Patients in Group B (n = 19) received 2–4 mg/kg/day GC beforehand, but their second IL-17A were still markedly high (Table 4, Figure 4b). Collectively, these findings indicate a positive correlation between persistently elevated IL-17A and airway mucus hyperproduction in Group A1, and the ineffectiveness of GC on inhibiting circulating IL-17A in patients with a lobar lesion and diffuse bronchiolitis in SMPP.

Figure 4. Variations in serum IL-17A and IL-6 after admission and clinical outcomes among four groups. (a,b) Dynamic changing curves of significantly elevated serum IL-17A in Group A1 (n = 33) and Group B (n = 19). (c) Dynamic changing curves of significantly elevated serum IL-6 in Group A2 (n = 15). (d) Clinical outcomes among four groups. (a–c) Data on the vertical axis are log-transformed. Dashed lines indicate IL-17A 38 pg/mL or IL-6 14 pg/mL, both representing twice the upper limit of their reference ranges.

Table 4. Dynamic changes of serum IL-17A and IL-6 in varied clinical phenotypes after glucocorticoid treatment.

	Group A1 matched with Cluster I (n = 33)		Group A2 matched with Cluster II (n = 15)		Group B matched with Cluster I (n = 19)
	IL-17A (pg/mL)		IL-6 (pg/mL)		IL-17A (pg/mL)
Timepoint	T1	T2	T1	T2	T1	T2
Level	74.60 (53.21, 111.90)	61.73 (46.30, 89.40)	87.90 (49.70, 119.90)	4.30 (2.00, 7.10)	121.80 (49.52, 194.90)	104.70 (46.84,188.00)
P*	> 0.05		< 0.0001		> 0.05

Data shown as median (IQR) for levels of cytokines. T1: on admission; T2: 3–4 days after admission.

*Mann–Whitney U test for continuous variables with non-normal distribution.

Clinical outcomes of patients with MPP harbouring varied clinical phenotypes

Clinical outcomes of enrolled patients are summarized in Figure 4d. Most patients with lobar lesion developed LAO (particularly bronchitis obliterans), including 56 patients in Group A1 (56/65, 86.2%) and all patients in Group A2 (27/27, 100%). Over 80% in Group B (33/39, 84.6%) developed LAO (particularly bronchiolitis obliterans, BO). All patients in Group C completely recovered without any sequelae. For the patients whose serum samples were collected twice in Group A1 and B, continuously elevated IL-17A was observed in 24 patients, most of whom developed LAO (21/24, 87.5%). Moreover, among those receiving blood samplings for three times in the two groups, this was confirmed in 18 patients, and 16 of them (16/18, 88.9%) developed LAO. In contrast, there seems to be no direct correlation between serum IL-6 and LAO given the prompt decrease in IL-6 after admission.

Discussion

Our study identified four cytokine clusters and four clinical groups based on the previous findings [9,10]. Among various inflammatory cytokines reported thus far, we focused on those relatively accessible in routine clinical practice. We demonstrated the relationship of circulating IL-17A and IL-6 with clinical phenotypes and prognosis in children with MPP. Briefly, significantly elevated IL-17A correlated with lobar lesions and diffuse bronchiolitis, while that of IL-6 was more often found in patients with lobar lesions progressing to lung necrosis. Continuously elevated IL-17A may correlate with LAO.

A particularly interesting finding is the strongly phenotypic disparities among patients with lobar lesions (Group A), which was reflected in the much more severe immune-mediated inflammatory response in Group A2 than in Group A1. Since the recruitment and activation of local alveolar macrophages upon lower respiratory tract infection by MP are crucial for subsequent alveolar inflammatory exudation [7,11], we speculate that the lobar lesion in Group A1 was mainly atelectasis due to airway obstruction by mucus hypersecretions, while that in Group A2 comprised both alveolar exudate-associated consolidation and atelectasis. This is supported by previous histopathological findings [10].

The reason why elevated IL-17A in Group A1 did not differ significantly from that in Group B, we suppose, is that patients harbouring these two phenotypes presented with airway mucus hyperproduction without exception. The main difference lies in the locations where mucus is secreted, i.e. in the segmental/subsegmental bronchi (proximal airways, Group A1) and in the bronchiolar lumen/alveolar ducts (distal airways, Group B). The fact that IL-17A promotes airway mucus production has been recognized in other respiratory diseases, which is reflected in the dose-dependent upregulation of mucin MUC5AC/MUC5B due to the enhanced expression of IL-17A [18–20]. We assume that patients in Group A1 may also suffer from bronchiolar mucus obstruction given the extensive expression of MUC5B in the entire airway [21]. Moreover, peripheral IL-17A accounts for a limited portion of that secreted at local sites of inflammation [22]. Therefore, its significant elevation might be of great value for predicting MP-induced Th17 cell infiltration and mucus hypersecretions in the airway.

Furthermore, continuously elevated IL-17A was largely associated with LAO in Group A1, which may result from long-lasting mucus hypersecretion that accounts for epithelial injury and airway remodelling [23,24]. Accordingly, circulating IL-17A may serve not only as a predictive factor for disease severity in MPP, but also as a prognostic factor for patients harbouring these two phenotypes. The primary clinical dilemma mirrored by our results is that GC therapy was ineffective on inhibition of the expression of IL-17A despite the increase in dosage, which drives us to consider possible targeted therapy against this cytokine. In fact, we have already started targeting mTOR, a master regulatory protein that can interact with IL-17A in vivo, for indirect blockade of IL-17A signalling in specific patients with MPP [3] based on our previous study [25]. Additionally, Th17-dominant immune profile and enhanced IL-17 – related pathway have also been observed in COVID-19 patients, and the viewpoint of targeting IL-17A for the alleviation of acute inflammation and sequelae has been proposed [26,27]. Given that IL-17A activates multiple downstream signalling pathways resulting in chronic lung disease [19], we put forward the feasibility of targeting IL-17A as early intervention for patients with SMPP harbouring these two phenotypes. Nevertheless, since IL-17A also plays an important role in protective immunity and repair of injured epithelium [28], side effects caused by the blockade must not be ignored. Future well-designed prospective studies are warranted for defining patient cohorts that would benefit from this therapeutic strategy.

Significant elevation of circulating IL-6 was mainly observed in patients with subsequent lung necrosis. Although lung biopsy was not conducted in these patients, lung necrosis is usually caused by a vascular event triggered by MP infection leading to hypercoagulation and thrombotic occlusion of intrapulmonary vessels [15]. These patients also suffered from severe mucosal necrosis and contained markedly high levels of D-dimer, which is consistent with the statement proposed by Bester et al. [29] that IL-6 positively correlates with hypercoagulation. Besides, elevation of IL-6 has been deemed as a prognostic factor and treatment target in acute necrotizing encephalopathy [30]. These findings further support the potential role of IL-6 in tissue necrosis. Because these patients have excessive immune-mediated inflammatory response, clinicians should pay attention to timely and initial adequate dosage of GC therapy after exclusion of coinfection [3]. However, the inconsistency between the rapid decrease in IL-6 after admission and the high incidence rate of LAO is confusing, which might be interpreted by post-effects of signalling pathways activated by IL-6 [31,32] and requires further investigations. Intriguingly, markedly high IL-6 was also found in approximately one-fifth of patients in Group A1. Considering that IL-6 signalling equally plays a role in MUC5B-induced airway fibrosis [33,34] and combined with our data, we hypothesize that there may be a cut-off value for serum IL-6 applied for predicting subsequent lung necrosis in the acute stage of SMPP with a lobar lesion.

Our study has several limitations. First, as the patients were enrolled over several years, there could be unintended bias in terms of the quantity of infected MP strains individually, and of course, the potential differences among the infected MP strains themselves. Secondly, we did not perform matching analyzes on the cytokine profiles of peripheral blood and within airway microenvironments in the same individuals. Thirdly, dynamic blood samplings were not performed on all enrolled patients due to routine clinical circumstances.

In general, we propose a classification draft for paediatric MPP in China based on the comprehensive cluster analysis of clinical symptoms, chest imaging, inflammatory markers and bronchoscopic findings (Figure 5). Although our understanding of this disease has improved over the last decade, the insights have not transformed into effective treatments for eliminating MPP associated LAO. Our findings emphasize the urgency of initiating individualized precision management for Chinese children with MPP in line with this proposal, which would be of great importance for early treatment and prevention of LAO that could lead to obstructive lung disease [35,36], and consequently for alleviation of burden on the patients and their families. The heuristic correlations between varied endotypes reflected by the two cytokines and clinical phenotypes may shed light on early targetable pathophysiologic changes as well as the possible biomarker-driven treatments for paediatric MPP. Future large-scale studies are required to exhibit further comprehensive conclusions.

Figure 5. Schematic diagram of the study content. Symptoms refer to the severity of fever and hypoxaemia before admission. Inflammatory markers refer to routine laboratory indexes including CRP, NEU, LDH and D-dimer (examined during 6–10 day after disease onset). Bronchoscopy refers to the findings under first FOB after admission. LAO: lower airway obliterans. Type N: no significant elevation in IL-17A and IL-6.

Acknowledgments

The authors would like to thank all patients and their family members who agreed to participate in this study, and all inpatient staff who assisted in the recruitment of the patients. In addition, HW is grateful to Chen Shen, PhD, and laboratorians at Beijing Children’s Hospital for providing research facilities, instruments and kind guidance. We also thank the reviewers for their suggestions to improve the manuscript.

Disclosure statement

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

Additional information

Funding

This work was supported by the Clinical Medicine Development of Special Funding Support, Beijing Hospitals Authority [ZYLX202118] and the Respiratory Research Project of National Clinical Research Center for Respiratory Diseases [HX2X-202103].