Procalcitonin is not an independent predictor of 30-day mortality, albeit predicts pneumonia severity in patients with pneumonia acquired outside the hospital

Background Procalcitonin (PCT) is a useful marker for pneumonia. However, its clinical usefulness in elderly patients has not been studied extensively. This study aimed to assess the relationship between PCT and prognosis and pneumonia severity in elderly patients with pneumonia acquired outside the hospital. Methods Data considered relevant to pneumonia severity and prognosis were retrospectively obtained from clinical charts of all patients with pneumonia who were admitted to our hospital from 2010 to 2017. The primary outcome was 30-day mortality in elderly patients (aged ≥75 years), and the relationship between PCT levels and pneumonia severity, as determined by the pneumonia severity index (PSI) was also examined. Results Data were collected from 667 patients, of which 436 were elderly patients. Multivariate and receiver operating characteristic curve analysis revealed that albumin, body mass index, and PSI class rather than PCT are important factors related to 30-day mortality in elderly patients. PCT was also not an independent prognostic factor in younger patients. PCT levels significantly differed by pneumonia severity (mild, moderate, and severe) in both younger (p < 0.001) and elderly (p < 0.0001) patients, with levels increasing as severity increased. In contrast, C-reactive protein (CRP) levels and white blood cell counts did not significantly differ by pneumonia severity in younger and elderly patients. A subgroup analysis focused on Streptococcus pneumoniae pneumonia revealed that PCT levels differed by severity in elderly patients (p = 0.03), with levels increasing as severity increased. Conclusion PCT was not an independent predictor of 30-day mortality in both of elderly and younger patients. PCT levels, but not CRP levels, significantly increased with increasing pneumonia severity in younger and elderly patients, although the degree of increase tended to be lower in elderly patients compared to younger patients for the same severity. PCT levels also significantly increased with increasing pneumonia severity in elderly patients with Streptococcus pneumoniae pneumonia.


Background
Community-acquired pneumonia (CAP) is a prominent cause of morbidity and mortality throughout the world [1]. Procalcitonin (PCT) can be used as a marker to differentiate bacterial infection from other non-bacterial infections or non-infectious inflammation [2][3][4]. Many investigators have reported the clinical usefulness of PCT for assessing the severity of pneumonia or pathogen involved, and for guiding antibiotic use [5][6][7]. Elderly patients, although representing the large majority of cases of CAP admitted to acute-care hospital wards, are often excluded from clinical studies due to the possible presence of multiple confounders. Although pneumonia occurs mainly in the elderly, the usefulness or clinical significance of PCT in these patients has not been studied extensively. While some studies have reported that PCT levels do not elevate according to the severity in elderly patients [8,9], others have argued that PCT is useful for predicting the severity of CAP [10,11].

Methods
Clinical charts of all consecutive patients admitted to our hospital from October 2010 to July 2017 with primary diagnoses of CAP and healthcare-associated pneumonia (HCAP) were retrospectively reviewed. We enrolled patients in the study with the same method as we previously reported [34], except for those with HCAP. HCAP included any patients who (1) were hospitalized in an acute care hospital for two or more days within the past 90 days, (2) resided in a nursing home or long-term care facility, or (3) continuously visited a hospital or hemodialysis clinic for intravenous antibiotic therapy, chemotherapy, or hemodialysis [35]. Data on admission considered related to pneumonia severity and prognosis were examined, including age, sex, category of pneumonia occurring outside the hospital (CAP or HCAP), PSI class, comorbidities, BMI, serum albumin, C-reactive protein (CRP), PCT and blood urea nitrogen (BUN) levels, and white blood cell (WBC) count. As for comorbidities, Charlson comorbidity index [36] was also obtained.
Serum PCT levels were determined by chemiluminescence enzyme immunoassay (SphereLight BRAHMS PCT, Wako Pure Chemical Industries, Ltd., Osaka, Japan). The PCT assay has a detection limit of 0.05 ng/mL. When serum PCT levels were under the detection limit, they were assigned a value of 0.05 ng/mL. Serum albumin, CRP and BUN levels were measured with improved bromocresol purple method, latex-turbidimetry and enzymatic method (urease-glutamate dehydrogenase), respectively, using auto-analyzer BioMajesty™ JCA-BM6050 (japan electron optics laboratory co ltd, Tokyo, Japan).
Microbiologic examinations performed were the same as our previous study [34].
The primary outcome was the relationship between PCT levels and 30-day mortality. In addition, we also investigated the relationship between PCT levels and pneumonia severity determined by PSI in younger (age ≤ 74 years) and elderly (age ≥ 75 years) patients, respectively. Pneumonia severity was determined by PSI as follows: PSI class I -III = mild, PSI class IV = moderate, PSI class V = severe. We also compared PCT levels for each pneumonia severity between younger and elderly patients.
At first, we checked normality of data distribution using Shapiro-Wilk test, which revealed that all data examined did not show normal distribution. Therefore, data were expressed as median (1st quartile, 3rd quartile). Relationship between PCT levels and 30-day mortality was examined using logistic regression and receiver operating characteristic (ROC) curve analysis. Variables with p < 0.05 in univariate analysis were further examined with multivariate analysis. Differences in PCT levels, CRP levels and WBC counts between younger and elderly patients, and among each pneumonia severity, were analyzed with the Mann-Whitney U test and Kruskal-Wallis test, respectively. When the Kruskal-Wallis test was significant, differences in PCT levels, CRP levels and WBC counts between each pneumonia severity were analyzed with the Dunn test. Comparisons of each category between younger and elderly groups were analyzed with the Mann-Whitney U and chi-square test. Statistical analysis was performed using Excel Tokei 2015 (Social Survey Research Information, Co., Ltd., Tokyo, Japan). P < 0.05 was considered statistically significant.
This human study was performed in accordance with the Declaration of Helsinki, and was approved by the Fukuoka University Medical Ethics Review Board (approval: R17-042). The review board exempted the acquisition of informed consent from patients included in the study.

Study population
We identified 682 patients with a primary diagnosis of pneumonia during the study period. Of these, 15 with hospital-acquired pneumonia were excluded. The final study population consisted of 667 patients. Clinical features of these patients are summarized in Table 1. The Klebsiella pneumoniae and Escherichia coli were more frequent in the elderly group, while Mycoplasma pneumoniae was more frequent in the younger group.

Factors associated with 30-day mortality
Though PCT was significantly associated with 30-day mortality in both of the younger and elderly groups in univariate analysis, multivariate analysis revealed that PCT was not an independent prognostic factor in both groups. Multivariate analysis revealed that BMI and albumin remained significant as factors related to 30-day mortality in both groups. In addition, BUN in the younger group and PSI in the elderly group were significant prognostic factors, respectively (Tables 2 and 3). According to ROC curve analysis, albumin and PSI, and albumin, BMI, and PSI revealed significantly wide AUC values compared to that of PCT in the younger and elderly groups, respectively (Figs 1 and 2).

Relationship between pneumonia severity and PCT, CRP, and WBC
Median (1st quartile, 3rd quartile) PCT levels in younger and elderly groups were 0.14 (0.     (Fig. 3, left  lower). WBC counts did not significantly differ by pneumonia severity in the elderly group (p = 0.74), but did differ by severity in the younger group (p = 0.04), although the counts did not increase with increasing severity.
Next, we investigated the relationship between pneumonia severity and PCT levels, CRP levels, and WBC counts in elderly patients in which pneumonia was caused by S. pneumoniae, the most common pathogen identified in the study. Of those with mild, moderate, and severe pneumonia, 21 (19.1%), 42 (18.4%) and 24 (24.5%) patients had S. pneumoniae pneumonia, respectively. PCT levels significantly differed by severity (p = 0.03) and increased with increasing severity. (Fig. 4, left upper). CRP levels (p = 0.68) and WBC counts (p = 0.31) did not significantly differ by severity (Fig. 4, right upper and left lower).

Discussion
PCT has been used to differentiate between infectious and noninfectious inflammatory respiratory diseases, or bacterial and non-bacterial respiratory infectious diseases [2][3][4]. Many investigators have reported the clinical usefulness of PCT in patients with CAP [5][6][7][17][18][19][20][21][22][23][24][25][26], HCAP [37,38], ventilator-associated pneumonia (VAP) [39][40][41][42][43], and hospital-acquired pneumonia (HAP) [44]. Although these studies included many elderly patients, the clinical usefulness of PCT in those patients was not consistent. Advanced age is associated with a decline in immune function, a phenomenon commonly referred to as immune senescence [45]. Immune senescence is generally characterized by chronic, low-grade, systemic inflammation and impaired responses to immune challenge. In elderly patients, complex immunologic rearrangements with chronic exposure to a variety of antigens may promote a subclinical inflammatory status defined as inflammaging, which can result in impaired PCT release in response to antigens [46]. It is currently unclear how the increased inflammation and declined functional reserve of parenchymal cells in elderly people would affect PCT release. In addition, the etiology of pneumonia may also be associated with different cytokine activation patterns. Thus, elderly patients with pneumonia might exhibit poor PCT elevation compared to their younger counterparts.
The geriatric population represents the fastest growing segment of the general population. Therefore, the clinical significance of PCT as diagnostic and prognostic factors in elderly patients is an important issue. While some investigators reported usefulness of PCT for diagnosing bacterial infection in the elderly [8,11,47], others revealed that PCT was inferior to CRP for predicting bacterial infection in these population [9,48,49]. As for predicting capacity for clinical course, Shi et al. reported that there were no differences in PCT between the improved and not-improved subjects in nosocomial pneumonia in the elderly [50]. To our knowledge, this is the first study to examine the relationship between PCT levels by pneumonia severity in younger and elderly patients with pneumonia, particularly patients with S. pneumoniae pneumonia, occurring outside the hospital setting. In addition, the present study included more elderly patients than those that previously assessed the clinical usefulness of PCT in the elderly population [8][9][10][11][47][48][49][50]. We assessed the relationship between PCT and PSI, because mortality in this frail elderly population is frequently not directly due to pneumonia [51], and PSI could be more specific of pneumonia severity than all cause mortality. We found that PCT levels, but not CRP levels, significantly increased with increasing severity of pneumonia in both younger and elderly patients, although the degree of increase tended to be smaller in elderly patients under conditions of same severity. Kim et al. also reported the usefulness of PCT for predicting the severity of CAP in elderly patients, although they defined elderly as > 65 years of age, the sample size was small, and they did not analyze pneumonia caused by a specific pathogen [10]. The present study revealed that PCT increased with increasing severity of pneumonia even in patients older than 75 years, and also those with S. pneumoniae pneumonia.
Elderly patients, although representing the large majority of cases of pneumonia admitted to acute-care hospital wards, are often excluded from clinical studies due to the possible presence of multiple confounders. Elderly patients included in the present study also had many comorbidities. While Charlson comorbidity index was significantly associated with 30-day mortality in elderly patients in univariate analysis, it did not remain an independent prognostic factor in these patients in multivariate analysis.
The definition of the elderly varies according to the aim and design of research studies. Although the definition of elderly in previous studies that have investigated PCT ranged from 65 to 85 years, many have adopted the definition of > 75 years. The average age of the overall population in the present study was 76 years. Based on this, we defined the elderly as patients older than 75 years.
This study has some limitations worth noting. First, the study had a retrospective design and was conducted in a single hospital. A prospective cohort study involving multiple hospitals will be needed to confirm our results. Second, the number of younger patients with severe Fig. 4 Box plots showing procalcitonin (PCT) levels, C-reactive protein (CRP) levels, and white blood cell (WBC) counts according to pneumonia severity determined by the pneumonia severity index of elderly (≥75 years old) patients with Streptcoccus pneumoniae pneumonia occurring outside the hospital setting. The box plots show 25th, 50th, and 75th percentiles, and outliers. **: p < 0.01 pneumonia was small. However, the number of such patients is naturally small, and several thousand pneumonia patients may be needed to obtain a sufficient number of younger patients with severe pneumonia. Thirdly, brain natriuretic peptide (BNP) or N-terminal-proBNP (NT-proBNP), which has been reported to be a prognostic marker of pneumonia [32,33], was not included in our study, because these markers were measured in only about half of patients included in the study. Prognostic significance of these peptides in elderly patients with pneumonia needs to be investigated. Finally, analysis of PCT levels in bacteremia patients was not performed in the present study, because blood culture was performed in only 60% of patients, and bacteria grew in only 8 patients.

Conclusions
PCT was not an independent predictor of 30-day mortality in both of elderly and younger patients with pneumonia acquired outside the hospital. PCT levels, but not CRP levels, significantly increased with increasing pneumonia severity in younger and elderly patients, although the degree of increase tended to be lower in elderly patients compared to younger patients for the same severity. PCT levels also significantly increased with increasing pneumonia severity in elderly patients with Streptococcus pneumoniae pneumonia.