Procalcitonin predicts death risk in COPD flare-ups without pneumonia

Patients hospitalized for a flare-up of chronic obstructive pulmonary disease (COPD) but with no sign of pneumonia may still carry a hidden warning signal in their blood. A large prospective study from China reports that procalcitonin, a biomarker usually associated with bacterial infection, independently predicts which of these patients are most likely to die during their hospital stay. The finding, drawn from nearly 7,000 patients treated at ten tertiary care hospitals, suggests that an elevated procalcitonin level marks a high-risk group even when chest imaging shows no pneumonia.

The study drew its participants from the MAGNET AECOPD Registry, a multicenter cohort established to examine how acute exacerbations of COPD are managed and what adverse outcomes patients experience. Between September 2017 and July 2021, the registry enrolled hospitalized adults aged 40 or older who had received a physician diagnosis of acute exacerbation of COPD according to GOLD criteria and who had a serum procalcitonin measurement taken during the index admission. From an initial pool of 14,007 registry patients, the investigators excluded those without procalcitonin data, patients younger than 40, individuals with malignant tumors, and anyone diagnosed with pneumonia or sepsis. Pneumonia diagnoses required compatible symptoms with radiographic confirmation, and sepsis was defined by the Sepsis-3 consensus criteria. After all exclusions, 6,791 patients remained for analysis.

To identify what distinguished patients who died in hospital from those who survived, the researchers ran both univariate and multivariate Cox regression analyses across baseline characteristics, comorbidities and laboratory values. In the multivariable model, several factors held up as independent risk factors for in-hospital mortality: increasing age, the presence of interstitial lung disease, cor pulmonale, lower diastolic blood pressure, lower hemoglobin levels, and procalcitonin. Each unit increase in procalcitonin carried a hazard ratio of 1.016, with a 95 percent confidence interval of 1.003 to 1.029 and a p-value of 0.014. Other markers that looked promising in univariate analysis, including C-reactive protein, heart failure and chronic kidney disease, lost their statistical significance once the model adjusted for the full set of variables.

Procalcitonin is the precursor of calcitonin, normally produced by the C cells of the thyroid gland and barely detectable in the bloodstream under healthy conditions. Outside the thyroid, mononuclear cells and the liver can also synthesize the molecule when stimulated by bacterial endotoxin and proinflammatory cytokines. Unlike C-reactive protein, which rises in response to a broad range of inflammatory stimuli whether infectious or not, procalcitonin is preferentially induced by systemic bacterial invasion. This specificity has made it a widely used tool for guiding antibiotic decisions in respiratory medicine, and previous work had shown that levels above 0.25 nanograms per milliliter correlate with poor prognosis in exacerbation patients. What remained unclear was whether procalcitonin still carried prognostic weight in exacerbations where no pneumonia could be found.

To convert the biomarker into a practical risk threshold, the team applied receiver operating characteristic analysis and the Youden index. Continuous procalcitonin levels predicted in-hospital mortality with an area under the curve of 0.840, with a 95 percent confidence interval of 0.802 to 0.875. The optimal cut-off came out at 0.11 nanograms per milliliter, a threshold that achieved a sensitivity of 0.844 and a specificity of 0.717 for detecting in-hospital mortality. Patients were then split into a high group above this value, comprising 1,880 individuals, and a low group at or below it, comprising 4,911.

The clinical contrast between the two groups was stark. Overall, 109 patients died in hospital, an all-cause mortality rate of 1.6 percent, but the deaths were heavily concentrated in the high-procalcitonin group: 4.7 percent versus 0.4 percent in the low group. Intensive care admission followed the same pattern, at 15.6 percent versus 3.9 percent, as did the use of invasive mechanical ventilation, at 9.1 percent versus 1.5 percent. Median length of stay stretched to 10 days in the high group compared with 8 days in the low group, and all these differences reached statistical significance. Kaplan-Meier survival curves confirmed that patients with elevated procalcitonin had markedly lower survival, and the separation persisted in subgroup analyses of patients who did and did not receive antibiotics during hospitalization.

The high-procalcitonin group also looked different before any outcome was measured. These patients were older, had lower body mass index, and carried a heavier load of comorbidities, including hypertension, coronary heart disease, heart failure, chronic kidney disease, arrhythmia, cor pulmonale, diabetes and cerebrovascular disease. Their laboratory profiles showed higher white cell counts and neutrophil percentages, lower hemoglobin and albumin, and elevated creatinine, blood urea nitrogen, D-dimer, fibrinogen, C-reactive protein and erythrocyte sedimentation rate. They more often presented with fever, purulent sputum, wheezing and chest tightness, while cough, sputum production and dyspnea were actually more common in the low group. Notably, respiratory failure rates did not differ significantly between the groups, and antibiotic use was far more frequent in the high group, at 82.3 percent versus 60.3 percent, suggesting clinicians empirically sensed an infection risk in these patients even without a pneumonia diagnosis.

One of the study’s most intriguing results came from the correlation analysis. Although the high-procalcitonin group had markedly higher median C-reactive protein levels, the individual-level correlation between the two biomarkers was weak, with a coefficient of just 0.14. Procalcitonin also showed no significant association with white cell count, neutrophil percentage, D-dimer, fibrinogen, erythrocyte sedimentation rate, uric acid, albumin, hemoglobin, platelet count, blood urea nitrogen or creatinine. Patients with higher procalcitonin did have higher age-adjusted Charlson comorbidity index scores, 4.6 versus 4.1, but the correlation between the biomarker and the comorbidity score was very weak at 0.032. The authors interpret this pattern as evidence that procalcitonin captures a distinct biological signal rather than simply mirroring general inflammation or comorbidity burden, potentially reflecting subclinical bacterial activation, tissue ischemia or organ stress.

The investigators propose several mechanisms that could explain procalcitonin elevation in the absence of confirmed pneumonia. Patients in the high group had lower albumin, hemoglobin and body mass index, along with a higher proportion of diabetes and slightly greater corticosteroid use, factors that leave them immunocompromised and susceptible to occult infections in the urinary tract, bloodstream or gastrointestinal tract. Alternatively, procalcitonin is known to rise after major surgery, extensive burns and postoperative lung injury, where sterile proinflammatory cascades can drive its synthesis independently of infection. In this cohort, the higher prevalence of cardiovascular and cerebrovascular comorbidities in the high group raises the possibility that elevated procalcitonin also signals systemic stress on the heart and brain, or tissue injury from ischemia, hypoxia or necrosis. The weak link between procalcitonin and kidney function measures suggests impaired renal clearance is unlikely to be the main driver of elevation.

The authors caution that their study has limitations inherent to its non-interventional design. Treatment protocols were not standardized across the ten centers, so variation in ventilatory management between hospitals cannot be excluded. Although 88.3 percent of patients underwent pulmonary CT imaging, some may have had occult radiographic changes, atypical bacterial pathogens or viral pneumonia that escaped detection. Antibiotics given before admission could have suppressed procalcitonin levels in some patients, and the biomarker may have been ordered preferentially in sicker patients or those already suspected of infection, introducing potential selection bias. Despite these caveats, the study stands as the largest investigation to date of procalcitonin as a prognostic marker in exacerbations of COPD without pneumonia, and it points to a practical implication: patients arriving with an exacerbation and a procalcitonin level above 0.11 nanograms per milliliter, but no radiographic pneumonia, warrant close monitoring and a systematic search for hidden infection or tissue injury.

Subject of Research: Procalcitonin as a prognostic biomarker for in-hospital mortality in COPD exacerbations without pneumonia

Article Title: Procalcitonin as a prognostic factor in patients with acute exacerbation of COPD without pneumonia: a prospective multicenter cohort study

Article References: Qi, M., Zhou, C., Luo, Y., Wei, H., Ge, H., Liu, H., Zhang, J., Li, X., Pan, P., Xie, X., Yi, M., Cheng, L., Zhang, J., Peng, L., Pu, J., Chen, X., Zhou, H., Tang, Y., & Yi, Q. (2026). Procalcitonin as a prognostic factor in patients with acute exacerbation of COPD without pneumonia: a prospective multicenter cohort study. Frontiers in Medicine, 13, Article 1905875. https://doi.org/10.3389/fmed.2026.1905875

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1905875

Keywords: procalcitonin, COPD, acute exacerbation, in-hospital mortality, pneumonia, biomarker, multicenter cohort study, sepsis, inflammation, ICU admission, antibiotics, risk prediction