Roles of C-reactive protein and its composite index in predicting acute exacerbation of bronchiectasis

Abstract

Background

Acute exacerbation of bronchiectasis is a critical driver of hospitalization and mortality, yet reliable biomarkers for predicting the frequency of adverse events (AEs) remain underexplored. C-reactive protein (CRP) and its composite indices (CAR: CRP/albumin ratio; CPR: CRP/prealbumin ratio) may reflect systemic inflammation and nutritional status.

Objectives

This study aimed to evaluate the ability of the CRP level, CAR, and CPR to predict the frequency of acute exacerbation among bronchiectasis patients.

Methods

A retrospective cohort of 223 stable bronchiectasis patients (2009–2019) was stratified into three groups according to the annual acute exacerbation frequency: Groups A (0), B (1–2), and C (≥ 3). Serum CRP, albumin, prealbumin, the CAR, and the CPR were analyzed. Multivariable analyses included ANOVA, Spearman’s correlation, logistic regression adjusted for clinical confounders (dyspnea, COPD, nutritional markers), and ROC curve evaluation with DeLong’s test for AUC comparison.

Results

Significant differences were observed across exacerbation frequency groups (A: 0, B: 1–2, C: ≥ 3 events/year), with Group C exhibiting the highest serum levels of CRP (12.25 ± 8.34 mg/L), CAR (0.31 ± 0.25), and CPR (60.83 ± 52.68), and the lowest albumin (41.07 ± 5.61 g/L; all P < 0.05). Acute exacerbation frequency showed strong positive correlations with CRP (r = 0.643), CAR (r = 0.637), and CPR (r = 0.603; all P < 0.05). Following multivariable adjustment for confounders (including demographics, dyspnea, COPD, and nutritional markers), CRP (adjusted odds ratio [aOR] = 1.092, P = 0.006), CAR (aOR = 1.104, P = 0.002), and CPR (aOR = 1.025, P = 0.001) remained independent predictors of frequent exacerbations (≥ 3/year). Receiver operating characteristic (ROC) analysis for predicting frequent exacerbations identified CAR > 0.15 as the most sensitive cutoff (74.3%) and CPR > 32.1 as the most specific (70.6%). Integrating CRP, CAR, or CPR with a clinical model (age, sex, albumin, prealbumin, dyspnea, COPD) significantly enhanced predictive accuracy compared to the clinical model alone (AUC range: 0.815–0.828 vs. 0.762; all P < 0.001). However, the differences in AUC between CRP (0.815), CAR (0.828), and CPR (0.821) were minimal.

Conclusions

Elevated CRP, CAR, and CPR levels correlate with increased exacerbation frequency and independently predict high-risk patients (≥ 3 events/year). While CAR and CPR offer modest improvements over CRP alone, their additional predictive value is limited due to the inherent biological relationship between these markers. Measuring CRP alone may be sufficient for clinical risk stratification to guide early interventions. Prospective validation of these inflammatory markers is warranted. Clinical trial number: Not applicable.

Introduction

Bronchiectasis, a heterogeneous chronic inflammatory disease, arises from various etiologies, such as recurrent infections, genetic abnormalities, immunodeficiency and autoimmune disorders [1]. The prevalence has increased over the past few years due to the widespread application of high-resolution computed tomography (HRCT) in the clinic, despite improvements in tuberculosis control and childhood diseases, such as measles and whooping cough [2]. Acute exacerbation is the main cause of frequent hospitalization and high mortality for patients with bronchiectasis. It has been reported that the number of acute exacerbations and the extent of deterioration within a year are highly important for the prognosis of bronchiectasis patients [3]. For example, a retrospective study revealed that the deterioration of bronchiectasis could cause 9–35% in-hospital mortality [4]. Thus, it is necessary to explore appropriate indicators to effectively predict acute exacerbation of bronchiectasis to provide timely prevention and treatment for patients at high risk of acute exacerbation.

Chronic inflammation plays a central role in structural damage to bronchopulmonary tissue. CRP, an acute phase reactive protein synthesized by the liver, increases rapidly when people are infected and is always proportional to infection severity [5]. Studies have shown that the level of CRP is related to the severity of bronchiectasis in the acute phase [6, 7]. Albumin and prealbumin are commonly used as clinical indicators of nutritional status and not only reflect the level of internal inflammation [8] but are also strongly correlated with the severity of bronchiectasis [9, 10]. Our previous research revealed that the serum albumin (ALB) and prealbumin levels were correlated with the severity score of bronchiectasis, such as the Bronchiectasis Severity Index (BSI) and FACED (F: FEV₁, A: age, C: colonization by pseudomonas, E: extension, and D: dyspnea) [11]. Recently, the ratio of CRP to albumin or prealbumin, a new inflammatory composite index, has been increasingly applied to assess the severity of disease [12, 13]. The ratio of CRP to prealbumin (CPR) can be used as a prognostic index for many diseases, such as esophageal cancer and brain injury [14, 15], and can also be used for risk stratification of acute heart failure and acute coronary syndrome [16, 17]. In addition, the ratio of CRP to albumin (CAR) can be used as a risk indicator for the exacerbation of chronic obstructive pulmonary disease (COPD) [18]. However, the role of CRP, the CPR and the CAR in predicting acute exacerbation of bronchiectasis and its frequency remain unclear.

Here, we retrospectively analyzed the correlation between the number of acute exacerbations in bronchiectasis patients and indicators, including CRP and its composite index, CAR, and CPR aiming to investigate their potential value in predicting the exacerbation of bronchiectasis.

Methods

Study design

A total of 311 patients who were diagnosed with bronchiectasis between January 2009 and December 2019 at Zhongshan Hospital, Fudan University, were enrolled, and 223 were ultimately eligible for inclusion. The inclusion criteria were based on the 2021 “consensus of experts on the diagnosis and treatment of Chinese adult bronchiectasis [19]”. (a) Long-term (more than 8 weeks) cough, expectoration (especially purulent sputum), sputum blood, hemoptysis, repeated infection, frequent exacerbation of chronic obstructive pulmonary disease (≥ 2 times/year), severe asthma or poor asthma control, positive sputum culture of Pseudomonas aeruginosa, chronic sinusitis, immunosuppressive status and other bronchitis symptoms or high-risk factors; (b) computed tomography (CT)/high-resolution computed tomography (HRCT) findings were consistent with the manifestation of bronchiectasis. The exclusion criteria were as follows: (a) patients with active pulmonary tuberculosis; (b) patients with nontuberculous mycobacterial infection; (c) patients with pulmonary fungal infection; (d) patients with any malignancies; (e) patients with systemic conditions associated with elevated CRP or malnutrition (e.g., ischemic cardiovascular diseases, peripheral vascular diseases, hematological malignancies, gastrointestinal pathologies, or liver cirrhosis); and (f) patients with incomplete laboratory indices.

Data collection

According to the diagnostic criteria of the 2021 “consensus of experts on the diagnosis and treatment of Chinese adult bronchiectasis [19]”, acute exacerbation of bronchiectasis is defined as the presence of symptoms, such as cough, changes in sputum volume, sputum thickening, dyspnea or decreased exercise tolerance, fatigue, and discomfort, as well as hemoptysis. Three or more of these symptoms worsen for more than 48 h, and clinicians believe that the above conditions need to be treated. The demographic data, clinical symptoms, and results of laboratory tests, including CRP, albumin, and prealbumin, were recorded independently for all 223 clinically stable patients. In addition, the number of acute exacerbations experienced by each patient over the following year was documented. On the basis of the BSI scoring standard [20], participants were categorized into three groups: 0, 1–2, and ≥ 3 exacerbations (referred to as Groups A, B, and C, respectively).

All biomarker measurements (CRP, albumin, and prealbumin) were obtained during routine clinical visits when patients were in stable condition, defined as no exacerbation, antibiotic use, or hospitalization in the preceding 4 weeks. These measurements were part of standard clinical care at our institution for patients with bronchiectasis during their regular follow-up visits. The median time between biomarker measurement and the subsequent exacerbation was 3.2 months (IQR: 1.8–5.7 months).

Statistical analysis

Continuous variables are presented as the means (SDs) or medians (interquartile ranges, IQRs), and categorical data are presented as numbers (percentages). Unpaired t tests or Mann‒Whitney U tests were used to compare data between two subgroups, whereas one-way analysis of variance or the Kruskal‒Wallis test was used for comparisons of three or more groups, depending on the data distribution. Strong or moderate correlations for r ≥ 0.4 and weak correlations for r < 0.4 [21]. Multivariable logistic regression models adjusting for demographic, clinical, and nutritional confounders were used to identify independent predictors of frequent exacerbations (≥ 3 events/year). To account for potential confounding effects of treatment interventions, we performed sensitivity analyses by including additional therapeutic variables (oral antibiotics, long-term macrolides, bronchodilators, and mucoactive agents) in our multivariable models. The predictive value of each indicator was evaluated according to the area under the curve (AUC) of the receiver operating characteristic (ROC) curve of each indicator. P < 0.05 was considered statistically significant.

Results

Baseline clinical characteristics of bronchiectasis patients

We collected 311 patients'clinical data, 223 of whom were finally included in the analysis (Fig. 1). The mean age was 62.44 years, and 60.5% were female (Table 1). Patients were stratified by annual exacerbation frequency: Group A (0 exacerbations, n = 27), Group B (1–2 exacerbations, n = 87), and Group C (≥ 3 exacerbations, n = 109). Group C had a longer disease duration (16.67 ± 12.32 years) and greater radiological involvement (3.45 ± 1.65 lobes) compared to Groups A and B. While cough, expectoration, and hemoptysis were common across all groups (P > 0.05), significant differences emerged in medication use and comorbidities. Group C showed higher rates of inhaled antibiotic use (34.9% vs. 14.8% in A, P = 0.021), bronchodilators (67.0% vs. 29.6% in A, P < 0.001), mucoactive agents (42.2% vs. 18.5% in A, P = 0.009), and COPD comorbidity (41.3% vs. 18.5% in A, P = 0.007). Long-term macrolide use was highest in Group C (28.4%, P = 0.038). Dyspnea was more prevalent in Group C (74.3% vs. 51.9% in A, P = 0.013). Group C showed higher rates of inhaled antibiotic use (34.9% vs. 14.8% in A, P = 0.021), bronchodilators (67.0% vs. 29.6% in A, P < 0.001), mucoactive agents (42.2% vs. 18.5% in A, P = 0.009), and COPD comorbidity (41.3% vs. 18.5% in A, P = 0.007). Long-term macrolide use was highest in Group C (28.4%, P = 0.038). Dyspnea was more prevalent in Group C (74.3% vs. 51.9% in A, P = 0.013). No significant differences were observed in comorbidities, including osteoporosis, stroke, heart failure, coronary artery disease, or liver disease across groups (all P > 0.05).

Fig. 1.

Flow chart of the study

Table 1.

Comparison of the clinical characteristics of Groups A, B and C (n = 223)

Characteristic	Group A (n = 27)	Group B (n = 87)	Group C (n = 109)	p value
Demographics
Male, n (%)	11 (40.7)	33 (37.9)	44 (40.4)	0.941
Age, years	68.20 ± 16.25	61.52 ± 15.75	61.74 ± 13.70	0.270
Disease duration, years	11.68 ± 9.22	14.07 ± 13.36	16.67 ± 12.32	0.062
Involved lung lobes, n	2.27 ± 1.34	2.58 ± 1.27	3.45 ± 1.65	0.102
Symptoms, n (%)
Cough	24 (88.9)	80 (92.0)	103 (94.5)	0.834
Expectoration	22 (81.5)	78 (89.7)	95 (87.2)	0.687
Hemoptysis	4 (14.8)	19 (21.8)	25 (22.9)	0.463
Dyspnea	14 (51.9)	52 (59.8)	81 (74.3)	0.013
Medications, n (%)
Inhaled antibiotics	4 (14.8)	19 (21.8)	38 (34.9)	0.021
IV/oral antibiotics*	6 (22.2)	35 (40.2)	62 (56.9)	< 0.001
Bronchodilators	8 (29.6)	42 (48.3)	73 (67.0)	< 0.001
Mucoactive agents	5 (18.5)	23 (26.4)	46 (42.2)	0.009
Long-term macrolides	3 (11.1)	16 (18.4)	31 (28.4)	0.038
Comorbidities, n (%)
COPD	5 (18.5)	24 (27.6)	45 (41.3)	0.007
Asthma	4 (14.8)	14 (16.1)	19 (17.4)	0.902
GERD	3 (11.1)	11 (12.6)	24 (22.0)	0.045
Diabetes	4 (14.8)	15 (17.2)	18 (16.5)	0.958
Osteoporosis	3 (11.1)	9 (10.3)	15 (13.8)	0.778
Stroke	4 (14.8)	7 (8.0)	9 (8.3)	0.541
Heart failure	2 (7.4)	8 (9.2)	12 (11.0)	0.796
Coronary artery disease	5 (18.5)	14 (16.1)	19 (17.4)	0.952
Liver disease	0 (0.0)	2 (2.3)	1 (0.9)	0.602

Footnote. Continuous variables (age, disease duration, and number of lobes involved): one-way analysis of variance (ANOVA) was used, and the data are presented as the means ± standard deviations. Categorical variables (sex, clinical manifestations): the chi-square test (χ2 test) was used, and the data are expressed as the"number of cases (percentage)". *IV/oral antibiotics: use within 6 months prior to enrollment. p, p value

Pulmonary function of the three groups of patients

The results of pulmonary function testing revealed that patients with fewer episodes of acute exacerbation had higher percentages of FEV1/pred, FVC/pred, PEF/pred, FEF25/pred, and DLCO/pred. However, no significant differences were observed among the three groups (Table 2).

Table 2.

Pulmonary function of the three groups of patients (n = 223)

Pulmonary function	A	B	C	P
FEV1/pred	87.93 ± 11.25	79.57 ± 21.01	70.74 ± 24.79	0.281
FVC/pred	91.41 ± 19.03	82.98 ± 10.14	76.75 ± 18.70	0.104
FEV1/FVC	73.82 ± 3.90	68.41 ± 9.32	65.71 ± 11.29	0.319
FEF/pred	82.23 ± 16.93	81.77 ± 16.54	76.41 ± 30.30	0.714
FEF25/pred	85.77 ± 19.25	67.08 ± 24.65	61.46 ± 41.00	0.272
DLCO/pred	97.08 ± 8.86	86.89 ± 18.19	81.68 ± 18.73	0.415

Footnote. The data are expressed as the means ± standard deviations and were analyzed by one-way analysis of variance (ANOVA). All p values were > 0.05, and there was no significant difference. FEV1 (forced expiratory volume in 1 s); FVC (FEV1/forced vital capacity); FEF (forced expiratory flow); DLCO (transfer coefficient for carbon monoxide); p, p value

Sputum findings

Among the 223 patients, 176 had positive sputum culture results, indicating mostly gram-negative bacterial infection. The most common gram-negative bacteria were Pseudomonas aeruginosa, Escherichia coli, and Klebsiella pneumoniae. Staphylococcus aureus, Staphylococcus epidermidis and Enterococcus were the most common gram-positive bacteria. There were no significant differences in these bacteria among the three groups. The overall sputum culture positivity rates were 66.7% (18/27) in Group A, 82.8% (72/87) in Group B, and 78.9% (86/109) in Group C, with no statistically significant difference across groups (P = 0.201) (Table 3).

Table 3.

Sputum etiology among the three patients’ groups (n = 223)

Sputum pathogenic culture	A	B	C	P
Pseudomonas aeruginosa	5 (27.8%)	18 (25.0%)	24 (27.9%)	0.913
Escherichia coli	3 (16.7%)	10 (13.9%)	13 (15.1%)	0.949
Klebsiella pneumoniae	2 (11.1%)	15 (20.8%)	11 (12.8%)	0.326
Staphylococcus aureus	4 (22.2%)	13 (18.1%)	20 (23.3%)	0.720
Staphylococcus epidermidis	1 (5.6%)	8 (11.1%)	8 (9.3%)	0.766
Enterococcus	2 (11.1%)	5 (6.9%)	6 (7.0%)	0.816
Others	1 (5.6%)	3 (4.2%)	4 (4.7%)	0.966
Total number of positive cases	18 (66.7%)	72 (82.8%)	86 (78.9%)	0.201

Footnote. The data are expressed as the"number of cases (percentage of positive cases in the group)". A chi-square test (χ2 test) was used, with p values > 0.05 indicating that there was no significant difference

Comparisons of the serum CRP, CAR, CPR, albumin and prealbumin levels among the three groups

Next, the levels of serum ALB, prealbumin, CRP, CAR, and CPR among the three groups were compared. As shown in Fig. 2, all five indicators were significantly different (P < 0.05). Among the three groups, Group C presented the highest levels of CRP (12.25 ± 8.34 mg/L), CAR (0.31 ± 0.25), and CPR (60.83 ± 52.68) and the lowest levels of ALB (41.07 ± 5.61 g/L) and prealbumin (0.21 ± 0.05 g/L), whereas Group A presented the opposite trend.

Fig. 2.

Levels of serum CRP, CAR, CPR, albumin and prealbumin in the three groups. A CRP (mg/L), B CAR (ratio, no unit), C CPR (ratio, no unit), D albumin (g/L) and E prealbumin (g/L) levels were significantly different among the three groups with different durations of acute exacerbations. One-way analysis of variance was used for comparisons among the three groups. *p < 0.05, **p < 0.01, ***p < 0.001

Correlations of serum ALB, prealbumin, CRP, CAR and CPR with the number of acute exacerbations

The correlations between serum ALB, prealbumin, CRP, CAR, and CPR levels and the number of acute exacerbations were further evaluated. As shown in Fig. 3a–c, CRP, CAR, and CPR were positively correlated with the number of acute exacerbations (P < 0.05), with correlation coefficients (r values) of 0.643, 0.637, and 0.603, respectively. In contrast, the serum ALB concentration was negatively correlated with the number of acute exacerbations (Fig. 3d, r = − 0.317). However, no significant correlation between prealbumin levels and the number of acute exacerbations was observed (Fig. 3e).

Fig. 3.

Correlations of serum ALB, prealbumin, CRP, CAR and CPR with the number of acute exacerbations. The correlations between the number of acute exacerbations (p < 0.05) and (A) CRP (mg/L), (B) CAR (ratio, no unit), and (C) CPR (ratio, no unit). D Correlation between the serum ALB concentration (g/L) and the number of acute exacerbations. E Correlation between prealbumin (g/L) and the number of acute exacerbations

Logistic regression and ROC analysis of each index and number of acute exacerbations

After adjusting for demographic factors (age, sex), nutritional markers (albumin, prealbumin), and significant clinical variables (dyspnea, COPD, inhaled antibiotics), CRP (aOR = 1.092), CAR (aOR = 1.104), and CPR (aOR = 1.025) remained independent predictors of frequent exacerbations (Table 4). Notably, prealbumin emerged as a significant protective factor (aOR = 0.912, P = 0.008).

Table 4.

Unadjusted and adjusted predictors of frequent exacerbations

Variable	Unadjusted OR (95% CI)	Adjusted OR (95% CI)*	p value	Fully Adjusted OR (95% CI)†	p value
CRP (per mg/L)	1.140 (1.076–1.208)	1.092 (1.025–1.163)	0.006	1.087 (1.019–1.159)	0.009
CAR (per 0.1 unit)	1.125 (1.032–1.153)	1.104 (1.038–1.174)	0.002	1.098 (1.031–1.169)	0.004
CPR (per 10 units)	1.032 (1.018–1.046)	1.025 (1.010–1.040)	0.001	1.022 (1.007–1.038)	0.005
Age (per year)	0.992 (0.975–1.009)	0.985 (0.967–1.003)	0.102	0.987 (0.968–1.006)	0.172
Male sex	1.012 (0.587–1.745)	1.208 (0.672–2.172)	0.527	1.194 (0.658–2.167)	0.559
Albumin (g/L)	0.942 (0.899–0.987)	0.958 (0.912–1.007)	0.089	0.962 (0.915–1.012)	0.132
Prealbumin (g/L)	0.851 (0.802–0.903)	0.912 (0.853–0.975)	0.008	0.918 (0.857–0.982)	0.013
Dyspnea	2.672 (1.518–4.704)	1.894 (1.032–3.476)	0.039	1.765 (0.948–3.287)	0.073
COPD comorbidity	3.214 (1.805–5.722)	2.215 (1.192–4.113)	0.012	2.037 (1.074–3.865)	0.029
Inhaled antibiotics	3.042 (1.689–5.480)	1.876 (0.981–3.587)	0.057	1.642 (0.837–3.221)	0.149
Oral antibiotics	2.582 (1.494–4.464)	–	–	1.425 (0.764–2.658)	0.265
Long-term macrolides	2.267 (1.186–4.333)	–	–	0.682 (0.452–1.028)	0.068
Bronchodilators	3.086 (1.767–5.389)	–	–	1.518 (0.780–2.954)	0.219
Mucoactive agents	2.518 (1.422–4.458)	–	–	1.346 (0.702–2.580)	0.371

*Adjusted for age, sex, albumin, prealbumin, dyspnea, COPD comorbidity, and inhaled antibiotics. †Fully adjusted model including all variables in this table (demographic, clinical, nutritional, and therapeutic interventions). OR odds ratio; CI confidence interval

To address potential confounding by treatment interventions, we conducted additional sensitivity analyses incorporating oral antibiotics, long-term macrolides, bronchodilators, and mucoactive agents into our multivariable models (Table 4). After these adjustments, CRP (aOR = 1.087, 95% CI 1.019–1.159), CAR (aOR = 1.098, 95% CI 1.031–1.169), and CPR (aOR = 1.022, 95% CI 1.007–1.038) maintained their significance as independent predictors of frequent exacerbations (all P < 0.01). Among treatment variables, long-term macrolide use showed a protective trend (aOR = 0.682, 95% CI 0.452–1.028, P = 0.068), while other therapeutic interventions did not significantly impact the predictive capacity of inflammatory indices.

ROC analysis demonstrated that combining CRP-derived indices with clinical variables significantly improved predictive accuracy. The clinical model alone achieved an AUC of 0.762, which increased to 0.815–0.828 when adding CRP, CAR, or CPR. While CAR provided a slightly higher discriminative capacity (AUC = 0.828) compared to CRP (AUC = 0.815) and CPR (AUC = 0.821), these differences were modest and may not be clinically significant. All three biomarkers showed good sensitivity (78.9–80.3%) and specificity (78.1–79.4%) when combined with clinical variables (Table 5, Fig. 4).

Table 5.

ROC analysis for predicting frequent exacerbations (≥ 3 events/year)

Model	AUC (95% CI)	Sensitivity	Specificity	Cutoff	p value versus CRP alone
CRP alone	0.724 (0.658–0.790)	0.698	0.701	> 8.2 mg/L	–
CAR alone	0.731 (0.666–0.796)	0.743	0.692	> 0.15	0.312
CPR alone	0.718 (0.651–0.785)	0.688	0.706	> 32.1	0.407
Clinical model*	0.762 (0.701–0.823)	0.725	0.732	–	0.021
Clinical model + CRP	0.815 (0.761–0.869)	0.789	0.781	–	< 0.001
Clinical model + CAR	0.828 (0.776–0.880)	0.803	0.794	–	< 0.001
Clinical model + CPR	0.821 (0.768–0.874)	0.794	0.786	–	< 0.001

*Clinical model: age + sex + albumin + prealbumin + dyspnea + COPD

Fig. 4.

Receiver operating characteristic (ROC) curves for predicting frequent exacerbations (≥ 3 events/year). A ROC curves for individual inflammatory biomarkers: C-reactive protein (CRP), C-reactive protein-to-albumin ratio (CAR), and C-reactive protein-to-prealbumin ratio (CPR). B ROC curves for the clinical model alone and combined with inflammatory biomarkers. The clinical model includes age, sex, albumin, prealbumin, dyspnea score, and COPD severity

Discussion

In this retrospective study, we found strong positive correlations between serum CRP, CAR, and CPR levels and the frequency of acute exacerbations in bronchiectasis patients. We demonstrated that CRP alone has significant predictive value for identifying patients at risk for frequent acute exacerbations (≥ 3 per year). While the composite indices CAR and CPR were also predictive, they provided only marginal improvements over CRP alone when combined with clinical variables (AUC improvement of 0.013 and 0.006, respectively). These findings suggest that in routine clinical practice, the simpler measurement of CRP may be sufficient, potentially avoiding unnecessary additional calculations or measurements.

Patients with bronchiectasis are prone to experience recurrent airway infection and inflammation, which play pivotal roles in the acute exacerbation of bronchiectasis and promote disease progression. It is widely recognized that a considerable proportion of bronchiectasis patients exhibit systemic inflammation. Although leukocytes and neutrophils can be used as inflammatory indicators to assess the severity of infection in patients, their specificity is poor and is influenced by various factors, such as exercise and emotions. CRP is a pentameric protein structure composed of five 23 kDa subunits that can stably exist in serum [22]. Previous studies have demonstrated that CRP and its composite indicators, CAR and CPR, are superior prognostic indicators that incorporate inflammatory and nutritional factors in various types of cancers [23, 24]. Consistently, this study revealed that CRP, CAR, and CPR levels are not only positively correlated with the frequency of acute exacerbations of bronchiectasis but also help predict the occurrence of frequent exacerbations within the following year.

Studies have shown that the level of CRP is associated with the severity of asthma, COPD and cystic fibrosis [25, 26]. However, there is limited research on the association between the CRP concentration and bronchiectasis during clinical stability, particularly in relation to the frequency of acute exacerbations and disease progression within 1 year. A previous study revealed that patients with an hs-CRP level greater than 4.26 mg/L had repeated hospitalizations (≥ 2 exacerbation-related hospitalization events), which suggested that the serum hs-CRP level might be a useful biomarker that directly reflects the degree of systemic inflammation in stable non-CF bronchiectasis patients [22]. However, only a small number of patients with an idiopathic etiology of bronchiectasis were included in the present study. In this study, 223 patients with bronchiectasis were included, and these findings suggested that CRP levels may serve as valuable predictors for experiencing at least three acute exacerbations of bronchiectasis within the following year. Furthermore, there was a strong correlation between CRP levels and the number of acute exacerbations. These results are consistent with those of previous studies on chronic obstructive pulmonary disease (COPD) and cystic fibrosis (CF) [27, 28], which also revealed that higher CRP levels are associated with greater risks of disease deterioration in the future.

In addition, studies have shown that CRP and albumin levels can be used as predictive markers for mortality in patients with infectious diseases and sepsis [29, 30]. The CRP/albumin ratio (CAR) is a parameter that can provide information about both the inflammatory and nutritional status of patients with bronchiectasis. An elevated CAR may indicate the presence of a chronic systemic inflammatory response and nutritional deterioration. In recent years, the CAR has emerged as a valuable tool for assessing the activity of certain autoimmune diseases [31, 32]. In addition, research has shown that the CAR is more accurate than the CRP alone in predicting 28-day mortality in critically ill patients [33], suggesting that the CAR may have greater predictive value in determining patient outcomes than the CRP alone. This study revealed that the CAR has greater sensitivity than the CRP in predicting acute exacerbations occurring three or more times per year in patients with bronchiectasis. Furthermore, the results of the correlation analysis confirmed a positive correlation between CAR levels and the frequency of acute exacerbations, indicating the potential of the CAR as a more sensitive and reliable biomarker to assess the risk of recurrent acute exacerbations in bronchiectasis patients.

This study found that dyspnea was significantly more prevalent in patients with frequent exacerbations (Group C: 74.3% vs. Group A: 51.9%, P = 0.013). This finding is consistent with the known association between symptom burden and exacerbation risk in bronchiectasis [34]. However, it is important to note that dyspnea is also a component of exacerbation definition. The higher baseline prevalence of dyspnea in Group C might reflect more severe underlying disease, which predisposes to both chronic symptoms and frequent exacerbations. This overlap between baseline symptoms and exacerbation criteria warrants careful interpretation of exacerbation frequency data, as patients with chronic dyspnea might be more likely to meet criteria for exacerbation. Future studies using patient-reported outcome measures specifically designed to distinguish chronic symptoms from acute changes could help refine exacerbation definitions in such patients.

In addition, the prevalence of COPD comorbidity was significantly higher in frequent exacerbators (41.3% vs. 18.5% in Group A, P = 0.007). Since elevated CRP and other inflammatory markers are well-established features of COPD, this raises the question of whether the observed biomarker elevations in our cohort are driven by bronchiectasis severity or coexisting COPD. However, our multivariable analyses adjusted for COPD status, and CRP, CAR, and CPR remained independent predictors of exacerbation frequency. This suggests that the association between these biomarkers and exacerbation risk persists beyond the contribution of COPD. Nevertheless, the potential interaction between bronchiectasis and COPD in driving systemic inflammation deserves further investigation, particularly given the overlapping pathophysiological features of these conditions.

Prealbumin is synthesized in the liver and has a shorter half-life of 2–3 days, which indicates that it is present in lower amounts in the human body [35]. Prealbumin is considered more sensitive than albumin in assessing nutritional status [36]. Furthermore, prealbumin is responsive to early changes in protein balance and serves as a sensitive marker for evaluating visceral protein status [36, 37]. As a novel inflammatory indicator, the CRP-to-prealbumin ratio (CPR) has demonstrated its value in predicting the prognosis of systemic inflammatory diseases [38]. Compared with CRP and CAR, CPR may possess greater sensitivity and clinical value in the evaluation of inflammatory diseases. In this study, we discovered that the CPR exhibited a greater predictive specificity for acute exacerbations of bronchiectasis occurring more than twice within the following year than did the CRP and CAR. Similar to what was observed for CRP and CAR, a positive correlation between CPR levels and the frequency of acute exacerbations was also observed. The persistence of prealbumin's independent effect after adjustment underscores its distinct biological properties. With a short half-life (2–3 days), prealbumin responds rapidly to acute changes in inflammation and catabolism [35, 36], whereas albumin (half-life ~ 20 days) reflects chronic nutritional status. This temporal dissociation explains why prealbumin provided incremental prognostic value even when albumin was accounted for. Notably, prealbumin synthesis is directly suppressed by IL-6-driven inflammation [36], making it a sensitive integrator of inflammatory-nutritional stress in bronchiectasis.

While CRP is a nonspecific biomarker elevated in various inflammatory conditions, we minimized confounding through strict exclusion criteria (active tuberculosis, nontuberculous mycobacterial infection, fungal infections, malignancies). This ensured observed CRP elevations primarily reflected bronchiectasis-related inflammation. Importantly, our adjusted regression maintained CRP (aOR = 1.092, P = 0.006), CAR (aOR = 1.104, P = 0.002), and CPR (aOR = 1.025, P = 0.001) as independent predictors after controlling for comorbidities (COPD, GERD) and systemic factors. ROC analysis identified clinically actionable cutoffs: CRP > 8.2 mg/L (sensitivity 69.8%), CAR > 0.15 (sensitivity 74.3%), and CPR > 32.1 (specificity 70.6%). The CAR cutoff demonstrated particular utility when combined with clinical variables (AUC = 0.828), supporting its role in risk stratification. These thresholds align with Sun et al.’s CAR > 0.12 for exacerbation prediction [18] and Francis et al.’s CRP > 5 mg/L threshold in COPD [27], though bronchiectasis-specific validation is warranted. We propose CAR > 0.15 as a pragmatic threshold to flag high-risk patients for intensified monitoring or preventive therapies (e.g., macrolides), while values < 0.15 may indicate lower exacerbation probability.

Importantly, our findings remained robust even after comprehensive adjustment for potential confounding factors, including various treatment interventions. In our fully adjusted models, which accounted for oral antibiotics, long-term macrolides, bronchodilators, and mucoactive agents, all three inflammatory indices maintained their significant association with exacerbation frequency. This suggests that these biomarkers reflect underlying disease activity independent of current therapeutic management. While treatment interventions showed trends toward influencing exacerbation risk, particularly long-term macrolides, which demonstrated a potential protective effect (aOR = 0.682, P = 0.068), the predictive capacity of inflammatory markers persisted despite these interventions.

While numerous studies, including the recent analysis by Lapinel et al. in the US Bronchiectasis and NTM Research Registry [39], have established a significant association between lower FEV₁/FVC values and increased exacerbation risk, this study observed a non-significant trend of declining FEV₁/pred and FVC/pred across exacerbation frequency groups (Groups A > B > C). We acknowledge this apparent discrepancy. Although one-way ANOVA was employed for group comparisons, the lack of statistical significance despite the observed numerical decline could potentially stem from several factors related to data distribution and study design. First, the variability within groups, particularly in Group C, suggests substantial heterogeneity in pulmonary function among frequent exacerbators, which may have obscured significant differences. Second, while ANOVA assumes normality, the distribution of FEV₁ and FVC values, especially in the context of severe disease (Group C), might have deviated from normality; non-parametric tests (e.g., Kruskal–Wallis) could have been more appropriate if normality assumptions were violated, though our initial analysis did not indicate this necessity based on the data characteristics assessed. Third, our cohort size, while sufficient for primary biomarker analysis, might have limited the power to detect subtle but clinically relevant differences in pulmonary function parameters across exacerbation groups, especially given the inherent variability in spirometry. Finally, the retrospective nature limited our ability to control for precise timing of spirometry relative to the last exacerbation or ensure uniform technique across the decade-long inclusion period, potentially introducing noise. Future prospective studies with larger cohorts, rigorous spirometry protocols, and explicit checks for data distribution (employing non-parametric analyses if warranted) are needed to definitively clarify the relationship between lung function decline and exacerbation frequency in bronchiectasis populations comparable to ours.

Although our results showed that CAR achieved the highest AUC (0.828) when combined with clinical variables, compared to CRP (0.815) and CPR (0.821), it is important to acknowledge that these differences are minimal. From a clinical utility perspective, such small improvements in predictive accuracy may not justify the additional complexity of calculating these ratios in everyday practice. CRP measurement alone, which is widely available and routinely performed, provides substantial predictive value and may be sufficient for risk stratification in most clinical settings. The composite indices might offer additional insights in specific contexts, such as when nutritional status assessment is particularly relevant, but their routine use for predicting exacerbation risk appears to have limited added value over CRP alone.

It is well-established that during inflammatory states, CRP levels typically rise, while albumin and prealbumin levels decline [40]. This inverse relationship reflects the acute phase response, where hepatic synthesis shifts from albumin to acute phase proteins like CRP. This biological redundancy may partly explain why CAR and CPR offer only modest improvements over CRP alone in predicting exacerbations. The ratios may essentially be amplifying an already existing biological relationship rather than capturing truly independent physiological processes.

Despite these insights, this study has several limitations. First, the sample size in this retrospective study was limited, which may affect the generalizability of our findings. Therefore, it is necessary to conduct further research with larger populations. Second, the subjective nature of patients'descriptions of their cough, expectoration, dyspnea, and other symptom burdens may introduce information bias into the results. The use of objective quantification of symptoms, such as the mMRC score and CAT scoring system [41], could provide more accurate results. Fourth, while we stratified exacerbation frequency using BSI-based criteria (Groups A, B, and C), calculating a comprehensive BSI or FACED score for regression/ROC analysis was not feasible. BSI requires prospective data on Pseudomonas aeruginosa colonization and detailed radiological scoring, which were unavailable in our retrospective cohort. FACED scoring was limited by incomplete FEV₁% documentation and lack of standardized dyspnea assessment. This precludes direct comparison of CRP-derived indices with these established tools but highlights the need for prospective validation integrating multidimensional severity scores. In addition, due to the retrospective nature of this study, we cannot establish whether these biomarkers (CRP, albumin, and prealbumin) directly predict future exacerbations or simply reflect underlying disease activity that predisposes to exacerbations. Finally, the lack of data on certain laboratory indicators, such as the leukocyte count, hemoglobin level, and globulin level at baseline, may introduce confounding bias to the findings. Comprehensive inflammatory panels and longitudinal biomarker tracking could clarify causal relationships between CRP-derived indices and exacerbation mechanisms.

Conclusion

While CAR and CPR are significant predictors of frequent exacerbations in bronchiectasis, their incremental value over CRP alone is modest. For clinical simplicity and efficiency, measuring CRP combined with clinical assessment may be sufficient for identifying patients at risk of frequent exacerbations. Future prospective studies should evaluate whether the small additional predictive value of these composite indices translates to meaningful improvements in clinical decision-making or patient outcomes.

Author contributions

All authors contributed to the study conception and design. Material preparation were performed by Minghui Wang and Yu Yan, data collection and analysis were performed by Jun Zhang. The first draft of the manuscript was written by Jun Zhang and Sijiao Wang. All authors commented on previous versions of the manuscript. Hongni Jiang and Chunling Du obtained the funding. All authors read and approved the final manuscript.

Funding

This work was supported by Fujian Provincial Natural Science Foundation of China (No.2023J011690); Major Project of Guangzhou National Laboratory (No.GZNL2024A02003); Hospital-level Research Project of Qingpu Branch, Zhongshan Hospital, Fudan University (QYT2022-02).

Data availability

All relevant data are available within the manuscript.

Declarations

Ethics approval and consent to participate

The study was approved by Local committee of Zhongshan Hospital Affiliated to Fudan University (Approval No. B2022-288R). All methods were carried out in accordance with Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.