Abstract
There are very few data on serum procalcitonin (PCT) levels in pulmonary tuberculosis (PTB) patients who are negative for HIV.
We assessed serum PCT in consecutive patients diagnosed with pulmonary tuberculosis or community-acquired pneumonia (CAP) on admission to discriminate between PTB and CAP, and examined the value of prognostic factors in PTB.
102 PTB patients, 62 CAP patients, and 34 healthy volunteers were enrolled. Serum PCT in PTB patients was significantly lower than in CAP patients (mean±sd 0.21±0.49 versus 4.10±8.68 ng·mL−1; p<0.0001). By receiver-operating characteristic curve analysis, serum PCT was an appropriate discrimination marker for PTB and CAP (area under the curve 0.866). PTB patients with ≥0.5 ng·mL−1 (normal cut-off) had significantly shorter survival than those with <0.5 ng·mL−1 (p<0.0001).
Serum PCT is not habitually elevated in HIV-negative PTB patients and is a useful biomarker for discriminating between PTB and CAP; however, when serum PCT is outside the normal range, it is a poor prognostic marker.
Procalcitonin (PCT), the precursor molecule of calcitonin, is known as a systemic inflammatory protein. Several studies have reported that serum PCT is a useful biomarker for diagnosis and for estimating the severity of community-acquired pneumonia (CAP) 1, 2. In particular, high serum PCT with CAP is associated with a high mortality rate 3–7. In contrast to CAP, there are very few data on PCT levels in pulmonary tuberculosis (PTB). According to the limited information available from small scale studies, which included 27 PTB patients 8, 30 PTB patients 9, and 34 HIV-positive PTB patients 10, serum PCT was not elevated; however, the clinical significance of serum PCT levels in PTB patients has not been well documented. In this study, we examined serum PCT levels in HIV-negative PTB patients to discriminate between PTB and CAP. Moreover, we investigated whether serum PCT levels in PTB patients are related to disease prognosis.
METHODS
Patients
From June 2008 to September 2009, consecutive patients admitted to our hospital (Tenryu Hospital, Hamamatsu, Japan) with PTB or CAP were included in this study. PTB was defined by sputum smear-positive and culture-positive Mycobacterium tuberculosis in the presence of new radiographic pulmonary infiltration. According to the European consensus on the surveillance of tuberculosis 11, PTB patients with tuberculous involvement of other organ systems were defined as having disseminated tuberculosis. Patients with other infections, such as urinary tract infection, meningitis and infectious endocarditis, were excluded. All PTB patients were initially treated with a standard four-drug regimen of isoniazid, rifampin, pyrazinamide, and ethambutol or streptomycin.
Pneumonia was diagnosed by the presence of new radiographic pulmonary infiltration and the following clinical findings: 1) axillary temperature >37.5°C; and 2) a cough, purulent sputum, pleuritic chest pain or shortness of breath. CAP was defined if pneumonia had occurred at home without antibiotic use in the previous 14 days.
As the severity index, the Pneumonia Patients Outcome Research Team score 12 was used in all PTB and CAP patients. Healthy volunteers free from respiratory disease were included as normal controls.
This study was prospective and was approved by the ethics committee of our hospital, and informed consent was obtained according to the hospital's guidelines.
Methods
Venous blood samples were drawn from PTB and CAP patients on admission. Serum PCT and C-reactive protein (CRP) were measured within 24 h of admission. Serum PCT was determined by an immunoluminometric assay (SphereLight B.R.A.H.M.S PCT; Wako Diagnostics, Tokyo, Japan). The normal range of PCT is <0.5 ng·mL−1 and the lower limit of detection is 0.1 ng·mL−1 13, 14. Serum CRP was determined by a turbidimetric assay (Nanopia CRP; Sekisui Medical, Tokyo, Japan). The lower limit of detection was 0.1 mg·dL−1. White blood cell (WBC) and differential cell counts were determined by flow cytometry (XT-2000i; SYSMEX Corporation, Kobe, Japan). If serum PCT and CRP were lower than their detection limit, we defined them as 0.05 ng·mL−1 and 0.05 mg·dL−1, respectively.
Statistical analysis
Data are expressed as mean±sd. Differences between groups were tested using the nonparametric Mann–Whitney U-test for continuous variables and Fisher's exact test for categorical variables, respectively. PTB and CAP were discriminated using receiver-operating characteristic (ROC) analysis. Mortality within 60 days of admission was analysed by Kaplan–Meier survival curves and by the log-rank test, according to cut-off values of 10.00 mg·dL−1 serum CRP 15, 16 and 0.5 ng·mL−1 serum PCT 13, 14, respectively. A probability value of <0.05 was regarded as significant.
RESULTS
Patient characteristics and serum PCT levels
102 PTB patients, 62 CAP patients, and 34 healthy volunteers were enrolled. All were seronegative for HIV. The causative micro-organism was identified in 33 CAP patients (53.2%), with the most frequently isolated micro-organism being Streptococcus pneumoniae (n = 13), followed by Haemophilus influenzae (n = 7), Pseudomonas aeruginosa (n = 5), Klebsiella pneumoniae (n = 4), Escherichia coli (n = 2), and Moraxella catarrhalis (n = 2). The characteristics of the enrolled patients are shown in table 1. There were no significant differences in age (71.5±18.5 versus 75.2±17.0 yrs; p = 0.2196) and sex (63/39 males/females versus 38/24; p>0.9999) between PTB and CAP patients. Serum PCT in PTB patients was significantly lower than in CAP patients (0.21±0.49 versus 4.10±8.68 ng·mL−1; p<0.0001) (fig. 1a). PTB patients had significantly lower serum CRP and WBC count than CAP patients (6.18±5.75 versus 15.67±9.59 mg·dL−1; p<0.0001 and 7.87±3.12 versus 13.01±8.23×109 cells·L−1; p<0.0001) (fig. 1b and c). Healthy volunteers (11/23 males/females aged 64.7±15.3 yrs) had 0.05±0.0 ng·mL−1 serum PCT, 0.14±0.27 mg·dL−1 serum CRP, and 5.95±1.33×109 cells·L−1 WBC count.
a) Serum procalcitonin, b) serum C-reactive protein and c) white blood cell count compared among pulmonary tuberculosis (PTB), community-acquired pneumonia (CAP) and healthy controls. #: p<0.0001; ¶: p = 0.0034; +: p = 0.0009.
12 of 102 PTB patients had disseminated tuberculosis (11.8%). As shown in table 2, disseminated tuberculosis patients had higher serum PCT (0.75±0.79 versus 0.14±0.39 ng·mL−1; p<0.0001) and serum CRP (11.19±7.28 versus 5.52±5.26 mg·dL−1; p = 0.0059) than non-disseminated tuberculosis patients. The PORT score (131.1±31.6 versus 84.5±32.2; p = 0.0001) and mortality within 60 days of admission (58.3% versus 5.6%; p<0.0001) were significantly higher in disseminated than non-disseminated tuberculosis patients.
Discrimination between PTB and CAP
On ROC curve analysis (fig. 2), the area under the ROC curve was 0.866 for serum PCT, 0.815 for serum CRP and 0.749 for the WBC count. Using different cut-off values, sensitivity, specificity and positive (PPV) and negative predictive values (NPV) are shown in table 3. With a cut-off value of 0.25 ng·mL−1, serum PCT had a sensitivity of 86.3%, specificity 74.2%, PPV 84.6% and NPV 76.7%; with a cut-off value of 10.00 mg·dL−1, serum CRP had a sensitivity of 76.5%, specificity 72.6%, PPV 82.1%, and NPV 65.2%. Serum PCT was a better discriminative marker for PTB and CAP than the WBC count or serum CRP, but there was no significant difference among them.
Receiver-operating characteristics curve for discrimination between pulmonary tuberculosis and community-acquired pneumonia using procalcitonin (–––––), C-reactive protein (· · · · ·) and white blood cell count (– – – –).
Prognosis
Mortality within 60 days of admission was 11.8% (12 out of 102) in PTB patients. Two died of severe tuberculous intestinal involvement and the others of respiratory failure. All 12 patients had drug-sensitive tuberculosis. With a cut-off value of 10.00 mg·dL−1 serum CRP, there was no significance between groups (p = 0.1074) (fig. 3a). Using a cut-off value of 0.5 ng·mL−1 serum PCT, patients with ≥0.5 ng·mL−1 had significantly shorter survival than those with <0.5 ng·mL−1 (p<0.0001) (fig. 3b). The characteristics of patients with PCT ≥0.5 or <0.5 ng·mL−1 are shown in table 4. A significant proportion of patients with PCT ≥0.5 ng·mL−1 had disseminated TB compared with PCT <0.5 ng·mL−1 (55.6% versus 7.5%).
Kaplan–Meier curves for 60-day mortality with pulmonary tuberculosis patients grouped according to a) 10 mg·dL−1 C-reactive protein (•: <10 mg·dL−1; ▪: ≥10 mg·dL−1) and b) 0.5 ng·mL−1 procalcitonin (•: <0.5 ng·mL−1; ▪: ≥0.5 ng·mL−1).
DISCUSSION
The current study demonstrates that serum PCT in HIV-negative PTB patients is basically low and is a useful biomarker for discriminating PTB and CAP; however, when serum PCT is above the normal cut-off point (0.5 ng·mL−1), it is a poor prognostic marker in PTB patients.
It has been reported that serum PCT is not elevated in PTB compared with in CAP 8–10, 17. Although the reasons why the serum PCT response in PTB is poor remain to be clarified, there are several possible explanations. First, secreted cytokine patterns are different between tuberculous infection and common bacterial infection; for example, interferon (IFN)-γ is a more critical cytokine for growth inhibition of mycobacteria than common bacteria 18–20. According to in vitro observation, IFN-γ attenuates the secretion of PCT from human adipose tissue 21. Secondly, as previously reported, serum PCT concentration increases slightly in intracellular infection, including that caused by Mycoplasma, viruses and Pneumocystis jiroveci 22, 23. Since PCT synthesis and release are determined by the inflammatory cytokine cascade during systemic infection, the intensity depends on the number of organisms entering the systemic circulation. The number of organisms in PTB is probably lower than in typical bacterial pneumonia.
Conversely, Naderi et al. 24 reported that serum PCT is not a reliable marker to discriminate between PTB and non-PTB disease due to its low sensitivity and specificity; however, since the study included mild cases, such as bronchitis, and used semi-quantitative serum PCT measurement, we consider that it differed from our study.
Several studies have shown a factor related to prognosis in tuberculous patients. Regarding nutritional status, it is reported that malnutrition is a risk factor for fatality in patients with miliary tuberculosis 25, and low serum albumin levels are strongly and independently associated with death in tuberculous patients 26. In addition to these findings, our study demonstrated that serum PCT levels over the normal cut-off point (0.5 ng·mL−1) predict a poor prognosis. To our knowledge, this is the first report on serum biomarkers determining prognosis in PTB patients.
Our study has several limitations. First, since outpatients were excluded, the enrolled patients might have had relatively severe conditions. As a result, serum PCT was elevated compared with previous studies including outpatients 1, 5. Secondly, PTB patients could be associated with bacterial co-infection. In severe cases, we performed non-tuberculous culture of blood, urine or sputum, but pathogenic bacteria were not found. Additionally, PCT levels in PTB patients who died were significantly lower than in CAP (0.90±0.92 versus 12.26±18.54 ng·mL−1; p = 0.0073). Taken together, we do not consider that bacterial co-infection was a factor in severe PTB patients; however, we cannot completely reject the possibility of bacteria co-infection with PTB.
In conclusion, serum PCT is a better biomarker than CRP or WBC count for the differentiation between PTB and CAP in HIV-negative patients. However, it is important to realise that elevated PCT values were observed in a limited number of patients with disseminated tuberculosis, which was associated with higher mortality.
Acknowledgments
We thank H. Ohta and A. Shimaya for microbiological analysis, H. Fukui, T. Sakurai, M. Uchida and M. Ogai for pleural fluid analysis, and Y. Fujisaka and N. Nakamura for data collection (Tenryu Hospital, National Hospital Organization, Hamamatsu, Japan).
Footnotes
Statement of Interest
None declared.
- Received January 23, 2010.
- Accepted May 24, 2010.
- ©ERS 2011
REFERENCES
