INT J TUBERC LUNG DIS 18(4):470–477 Q 2014 The Union

Value of procalcitonin in differentiating pulmonary tuberculosis from other pulmonary infections: a meta-analysis S-L. Huang,* H-C. Lee,† C-W. Yu,‡ H-C. Chen,‡ C-C. Wang,* J-Y. Wu,‡ C-C. Lee§¶# *Department of Emergency Medicine, Chang Gung Memorial Hospital, and Chang Gung University College of Medicine, Taoyuan, †Department of Anaesthesiology, Chang Gung Memorial Hospital, Taoyuan, and Graduate Institute of Clinical Medical Sciences, Chang Gung University, Taoyuan, ‡Department of Emergency Medicine, Chang Gung Memorial Hospital, Keelung, Taiwan and Chang Gung University College of Medicine, Taoyuan, § Department of Emergency Medicine, National Taiwan University Hospital Yunlin Branch, Douliou, ¶Department of Emergency Medicine, National Taiwan University Hospital, Taipei, Taiwan; #Department of Epidemiology, Harvard School of Public Health, Boston, USA SUMMARY OBJECTIVES:

To systematically and quantitatively summarise the current evidence on the utility of the procalcitonin test (PCT) in discriminating pulmonary tuberculosis (TB) from other pulmonary infections. M E T H O D S : We searched MEDLINE, EMBASE and the Cochrane database up to August 2013 for studies that reported the performance of PCT alone or compared with other biomarkers in diagnosing pulmonary TB. We summarised PCT using forest plots, hierarchical summary receiver operating characteristic curves and bivariate random effects models. R E S U LT S : We found nine qualifying studies covering 951 episodes of suspected TB along with 426 confirmed TB cases. The bivariate pooled sensitivity and specificity of PCT to distinguish TB from non-TB were respectively

42% (95%CI 30–56) and 87% (95%CI 63–96). The bivariate pooled sensitivity and specificity for PCT in distinguishing TB from bacterial pneumonia were respectively 78% (95%CI 67–86) and 85% (95%CI 78–90). Low heterogeneity was noted in studies comparing TB with bacterial pneumonia patients. C O N C L U S I O N : The results suggest consistently acceptable sensitivity and specificity of the PCT test in distinguishing TB from bacterial pneumonia. However, given the imperfect sensitivity and specificity of the test, medical decisions should be based on both the PCT test results as well as on clinical findings. K E Y W O R D S : pneumonia; procalcitonin; tuberculosis; meta-analysis

EVERY YEAR, an estimated 9 million new cases of tuberculosis (TB) occur, resulting in 1.7 million deaths worldwide.1 TB is the single most important cause of death due to an infectious agent in low- and middle-income countries. In industrialised countries, TB remains an important public health problem due to the high prevalence of TB in human immunodeficiency virus (HIV) patients.2 Microscopic detection of TB bacilli in sputum with Ziehl-Neelsen staining remains the main diagnostic tool in most clinical settings. The estimated detection rate for new sputum smear-positive cases of TB is as low as 62%. Clinical symptoms and chest radiographic findings for TB are often undifferentiated from those of other bacterial infections, and the gold standard test, microbiological culture, has the disadvantage of a high turnaround

time. There is therefore an urgent need for a biomarker test for the diagnosis of TB. Procalcitonin (PCT), a 116-amino-acid N-terminal end precursor protein of calcitonin, has become an infectious disease marker of increasing relevance in recent years.3 During bacterial infection, the production of PCT is induced by tumour necrosis factoralpha (TNF-a) and interleukin 2.4–6 Unlike traditional inflammatory biomarkers, such as C-reactive protein, PCT is inhibited by interferon-gamma (INFc), and does not increase in response to viral infection.3 In patients with clinical TB, both TNF-a and IFN-c play key roles in the cellular host response. PCT seemed to increase only moderately in patients with active TB. The usefulness of PCT in the clinical differential diagnosis of TB from other causes of pulmonary infection has been investigated

Correspondence to: Chien-Chang Lee, Department of Emergency Medicine, National Taiwan University Hospital Yunlin Branch, Douliou, 640, Taiwan. Tel: (þ88) 623 123 456; ext 5926. Fax: (þ88) 6 2322 3150. e-mail: [email protected] Jiunn-Yih Wu, Department of Emergency Medicine, Chang Gung Memorial Hospital, Keelung, Taiwan and Chang Gung University College of Medicine, Taoyuan, 3F No. 3, Ln 325, Sec 2, Shipai Rd, Beitou Dist, Taipei City 11267, Taiwan. Tel: (þ886) 3328 1200; ext 2505. Fax: (þ886) 3328 7715. e-mail: [email protected] Article submitted 2 July 2013. Final version accepted 11 December 2013.

PCT for diagnosis of PTB


in a small number of studies, with inconsistent results.7–15 We performed a systematic review and metaanalysis to determine the sensitivity and specificity of the serum PCT test with regard to distinguishing pulmonary TB from other pulmonary diseases.

METHODS Our systematic review and meta-analysis was conducted in accordance with the published guideline for Systematic Reviews of Diagnostic Test Accuracy.16,17 Search strategy and selection criteria We searched three electronic databases (MEDLINE, EMBASE and the Cochrane database) using broad search terms, ‘procalcitonin’, ‘tuberculosis’ and ‘TB’, for studies published up to August 2013. We searched the bibliographies of primary studies and review articles and contacted authors for additional data. MEDLINE was searched using MeSH terms and free text. The same strategy with Emtree tools was used to search the EMBASE database. We did not set any time or language restrictions for these searches. Initial eligibility was determined independently by two reviewers. Discrepancies between the reviewers were resolved by consensus between the third and fourth co-authors. Eligible studies The titles and abstracts of the studies were screened in the first round, and potentially relevant articles were retrieved for full-text review in the second round. Eligible studies were primary studies that assessed the diagnostic accuracy of PCT alone or compared with other tests, with regard to discrimination between pulmonary TB and bacterial pneumonia or other lower respiratory tract infections. For inclusion, studies were required to have a study population of adult patients. We excluded case reports, case series, review articles, editorials and clinical guidelines. Our review was focused on adult populations; studies among children were excluded. Two authors independently assessed all titles and abstracts to determine whether the inclusion criteria were satisfied. Full-text articles were retrieved if any of the reviewers considered the abstract suitable. The study inclusion and exclusion process is summarised in Figure 1. Information about the publication (title, authors, journal), study population, comparison group (bacterial pneumonia), study design (crosssectional or cohort) and prespecified covariates, such as the cut-off value, was collected, as were descriptions of reference and index tests and data for 232 contingency tables. Quality assessment The methodological quality of the selected studies

Figure 1

Flow chart of study identification and inclusion.

was evaluated independently by two reviewers with a validated tool for the assessment of the quality of diagnostic accuracy studies (Quality Assessment of Diagnostic Accuracy Studies [QUADAS]).18 Discrepancies were resolved by consensus between the third and fourth co-authors. Data generation and analysis We calculated the pooled sensitivity, specificity and likelihood ratio, as well as the diagnostic odds ratio for the included studies. As simple pooling by weighted summary of sensitivity and specificity is only valid if all primary studies report test performance using the same threshold,16 we performed a random-effects bivariate regression analysis to calculate summary estimates of sensitivity and specificity and their 95% confidence intervals (CIs).19 Summary likelihood ratios (LRs) were calculated from pooled sensitivity and specificity data. Hierarchical summary receiver-operating characteristic curves (HSROCs) were also generated with the parameters of the HSROC model, taking into account the threshold heterogeneity of PCT tests in the included studies.20 Confidence regions for the summary points and the prediction region in which 95% of future studies are predicted to lie were plotted. To deal with values of zero in 232 contingency tables, we performed continuity correction by adding ½ to each cell, thereby reducing the small study bias. Diagnostic accuracy studies are expected to show considerable heterogeneity. To formally quantify the extent of between-study variation (i.e., heterogeneity), we calculated the inconsistency index, I2, which de-


The International Journal of Tuberculosis and Lung Disease

scribes the variation of effect estimate that is attributable to heterogeneity across studies.21 We prespecified several additional analyses to examine the potential effects of different methodological quality factors, adjust for covariates and assess the robustness of our results. We also used Galbraith plots to explore sources of heterogeneity.22 Statistical analyses were conducted using Stata 11.0 (Stata Corp, College Station, TX, USA). All statistical tests were two-tailed, and statistical significance was defined as P , 0.05.

RESULTS A total of 2390 studies (excluding duplicates) were identified using the search strategy outlined above (Figure 1). After the first round of title screening and abstracts, 2366 ineligible studies were excluded; 24 potentially relevant studies were retrieved for full text evaluation, of which a further 15 were excluded for varying reasons, leaving 9 that met the inclusion criteria. The nine eligible studies included 951 episodes of suspected TB, with 426 (44.8%) confirmed TB episodes. Table 1 presents a summary of the characteristics of the included studies and patients. The number of TB patients in each study ranged from 18 to 102, and the mean/median ages ranged from 23.1 to 69.3 years. Most studies used both Ziehl-Neelsen stain positivity of sputum or bronchial lavage fluid and growth of TB culture of sputum or pleural fluid as

the reference test. The two exceptions were a study by Polzin et al., which used smear together with chest radiography, and a study by Nyamande and Lalloo, which used smear or culture as the reference standard.10,11 Three studies also included histological examination or staining of pleural or lung tissue specimens, and one study included pleural fluid adenosine deaminase concentration .65 international units (IU)/dl as an additional criterion. The included studies can be classified into three broad categories: 4 studies comparing pulmonary TB with bacterial pneumonia, 4 comparing pulmonary TB with nonTB patients with various other types of pulmonary infection, and 1 comparing TB pleurisy with other causes of pleural effusion. In addition, three studies examined the diagnostic value of C-reactive protein (CRP) comparing TB and bacterial pneumonia. The cut-off values varied among the included studies, ranging from 0.081 to 3 ng/ml (Table 1). We used the QUADAS tool for study quality assessment. Figure 2 provides an overall impression of the methodological quality of the studies. All blood samples were taken in close proximity to the confirmation diagnosis. All patients were verified by the same reference standards in all studies. None of the included studies explained the withdrawals or reported results that could be interpreted, and the physicians were blinded to the index test while verifying outcomes by reference standards in only one study. We could not exclude the possibility of incorporation bias.

Table 1 Characteristics of the nine included studies that used biomarkers to assess patients with suspected pulmonary tuberculosis

Prevalence (n)

Age group years



Polzin, 2003, Germany11

0.26 (104)



PCT ¼ 0.25

Schleicher, 2005, South Africa13

0.51 (67)



PCT ¼ 3 CRP ¼ 24.6

Nyamande, 2005, South Africa10

0.26 (169)



PCT ¼ 1.75

Cakir, 2005, Turkey15

0.64 (28)



PCT ¼ 0.081



PCT ¼ 0.5

Author, year, country, reference

Baylan, 2006, Turkey7 8

0.5 (150)

Kang, 2009, Korea

0.35 (87)



PCT ¼ 0.1 CRP ¼ 10

Naderi, 2009, Iran9

0.5 (92)



PCT ¼ 0.5

0.56 (90)


0.62 (164)



PCT ¼ 0.25 CRP ¼ 20 PCT ¼ 0.1 CRP ¼ 10

Porcel, 2009, Spain12 14

Ugajin, 2011, Japan

Reference test Positive sputum ZN staining and infiltration/cavitation on CXR ZN stain of sputum or lavage fluid, TB culture and compatible findings on CXR or lung tissue biopsy Positive sputum ZN stain or culture Sputum stain, TB culture and pleural biopsy showing caseating granuloma with positive staining ZN stain and TB culture TB culture, adenosine deaminase .65 IU/dl in lymphocyte-predominant exudative pleural effusions and CXR Clinical, radiological, sputum ZN stain, culture and response to anti-tuberculosis chemotherapy TB culture and pleural biopsy ZN stain, TB culture and CXR

*PCT in ng/ml; CRP in mg/l. † In ng/ml. PCT ¼ procalcitonin; CRP ¼ C-reactive protein; TB ¼ tuberculosis; SD ¼ standard deviation; ZN ¼ Ziehl-Neelsen; CXR ¼ chest X-ray; NA ¼ not available; AECB ¼ acute exacerbation of chronic bronchitis; CAP ¼ community-acquired pneumonia; HAP ¼ hospital-acquired pneumonia; IU ¼ international unit.

PCT for diagnosis of PTB


Figure 2 Quality Assessment of Diagnostic Accuracy Studies criteria for included studies. The consensus judgment of quality criteria is shown as cumulative percentages across the nine studies included.

Diagnostic accuracy indices The results of the meta-analysis indicate that PCT testing has a low degree of accuracy with regard to differentiating pulmonary TB from other causes of pneumonia. The pooled sensitivity and specificity

estimates were respectively 0.42 (95%Cl 0.30–0.56) and 0.87 (95%CI 0.63–0.96). The low positive likelihood ratio (LRþ 3.20, 95%CI 0.93–10.9) of PCT was not sufficient for a rule-in test, and the high negative likelihood ratio (LR- 0.67, 95%CI 0.50–

Table 1 (continued)

PCT Sensitivity 58.0; Specificity 91.0 Sensitivity 81.8; Specificity 82.4


Mean serum PCT concentration in TB patients mean 6 SD†

Comparison group

Pleural fluid PCT

Study design


0.14 6-0.02




Sensitivity 78.8; Specificity 82.3

1.03 (IQR3.17)

Pneumococcal CAP



Sensitivity 30.0; Specificity 82.0 Sensitivity 72.0; Specificity 80.0


4.16 6 7.94




0.12 60.07

Non-pulmonary TB, other causes of pneumonia Non-tuberculous pleurisy

Sensitivity 66.7; Specificity 90.0

Case control

Sensitivity 41.3%; Specificity 100% Sensitivity 86.2; Specificity 78.9



Non-TB, healthy controls



Sensitivity 83.3%; Specificity 75.0%

0.29 (range, 0.01 to 0.87)

Bacterial CAP



Sensitivity 36.9; Specificity 63.1



Non-tuberculous pneumonia




Sensitivity 74.0%; Specificity 77.0% Sensitivity 76.5%; Specificity 72.6%

Pleural fluid 0.10 6 0.05 0.21 6 0.49

Non-tuberculous pleurisy

Sensitivity 25.0; Specificity 76.0 NA


Sensitivity 78.4; Specificity 85.4;




The International Journal of Tuberculosis and Lung Disease

0.90) could not reduce the post-test probability to a level that TB could be safely excluded. In contrast to the low capacity of PCT to distinguish TB from miscellaneous causes of non-tuberculous pulmonary disease, PCT has a moderate to high degree of accuracy with regard to distinguishing TB from bacterial pneumonia. The pooled sensitivity and specificity estimates were respectively 0.78 (95%CI 0.67–0.86) and 0.85 (95%CI 0.78–0.90). The high LRþ (5.19, 95%CI 3.62–7.46) of PCT was sufficient for the rule-in diagnosis of bacterial pneumonia; the low LR- (0.26, 95%CI 0.17–0.39) also provided important information about the diagnosis of pulmonary TB. Three studies also examined the accuracy of the CRP test in distinguishing TB from bacterial pneumonia. The pooled sensitivity and specificity estimates were respectively 0.76 (95%Cl 0.68–0.82) and 0.78 (95%Cl 0.71–0.84). The LRþwas 3.21 (95%CI 2.40– 4.30) and the LR- was 0.29 (95%CI 0.21–0.40). We constructed summary ROCs for both PCT. The area under the ROC curve (AUC) for PCT was 0.89 (95%CI 0.86–0.91) for studies comparing TB and bacterial pneumonia (Figure 3A), and an AUC of 0.57 (95%CI 0.53–0.61) for studies comparing TB and miscellaneous pulmonary infections (Figure 3B). The diagnostic OR for PCT was 19.9 (95%CI 11.7–34.0) when compared with bacterial pneumonia (Figure 4A), and 4.25 (95%CI 1.1–16.9) when compared with various causes of pneumonia (Figure 4B). The diagnostic OR for CRP was 11.3 (95%CI 6.6–19.6) for CRP (Figure 4C). Overall, PCT has a higher discriminative capability than CRP in differentiating pulmonary TB from bacterial pneumonia. We did not observe a substantial degree of heterogeneity for PCT (I2 ¼ 0.0%, 95%CI 0.0–84.7) or CRP (I2 ¼ 0.0%, 95%CI 0.0–89.6). We performed a sensitivity analysis by excluding the study by Polzin et al., which used only smear and radiographic findings as the reference standard. The pooled OR, excluding Polzin et al.’s study, was 22.00 (95%CI 11.96–40.50), slightly higher than the overall estimate (dOR 19.9, 95%CI 11.7–34.0). The Galbraith plots did not indicate a potential source of heterogeneity in either PCT or CRP meta-analysis. The results of the puortitative analyses are summarised in Table 2.

DISCUSSION PCT has been shown to be a valuable diagnostic tool in differentiating bacterial from other causes of lower respiratory tract infections; however, there are only limited data regarding its usefulness in diagnosing active pulmonary TB. In the current study, we showed that PCT may not be a useful indicator for the identification of TB in a mixed population with miscellaneous causes of pulmonary infection. Although the specificity is reasonable (0.87, 95%CI

Figure 3 Summary receiver operating characteristic curve (solid line) and the bivariate summary estimate (solid square), together with the corresponding 95% confidence ellipse (inner dashed line) and 95% prediction ellipse (outer dotted line). The symbol size for each study is proportional to the study size. HSROC ¼ hierarchical summary receiver-operating characteristic curves.

0.63–0.96), the sensitivity is unacceptably low (0.42, 95%CI 0.30–0.56). However, PCT can be used to differentiate pulmonary TB from typical bacterial pneumonia with reasonable accuracy. High heterogeneity was observed in the subgroup where PCT was used to differentiate TB from other causes of pulmonary infection. The high heterogene-

PCT for diagnosis of PTB


Figure 4 Forest plot of the diagnostic ORs of studies that used PCT compared with A) bacterial pneumonia; B) various causes of pneumonia; and C) CRP for the diagnosis of pulmonary tuberculosis patients. OR ¼ odds ratio; CI ¼ confidence interval; PCT ¼ procalcitonin; CRP ¼ C-reactive protein.

ity is likely to have resulted from differences in the nature of the comparison populations. In the subgroup where PCT was used to differentiate between TB disease and bacterial pneumonia, the pooled estimate showed a high degree of agreement (I2 ¼ 0). In actual clinical practice, physicians are usually faced with various possible aetiologies of pulmonary conditions in patients presenting with fever and pulmonary infiltrates. For frontline physicians, PCT is therefore of little value for the rule-in diagnosis of TB in febrile patients presenting with undifferentiated respiratory

symptoms. However, for patients with typical presentation and chest X-ray findings suggestive of bacterial pneumonia, an elevated PCT (.0.5 ng/ml) usually assists clinicians in ruling out active TB. Of note, two studies on HIV-positive patients showed higher serum PCT levels than non-HIVinfected patients.10,13 Mean serum PCT levels in patients with pulmonary TB in these two series were 4.16 ng/ml (SEM 1.197) and 1.03 ng/ml (SEM 3.16). Several possible explanations for this phenomenon include unrecognised concomitant bacterial infection,


The International Journal of Tuberculosis and Lung Disease

Figure 4


higher proportion of patients with extensive or disseminated TB and elevated circulating levels of inflammatory cytokines. Other than in HIV-infected patients, PCT is only mildly elevated in patients with active TB. The cut-off values used in the included studies were between 0.10 and 0.25 ng/ml, lower than the standard cut-off value (0.50 ng/ml) recommended for critically ill patients. Lower cut-off values may improve the sensitivity of the test, at the cost of specificity. The determination of the cut-off value should take into consideration the severity and timing of the disease of interest in the target population. In the emergency department setting, patients tend to present with undifferentiated symptoms in the early stages of disease; a lower cut-off value, e.g., 0.1 ng/ ml, may thus be more appropriate. In critical care unit settings, patients are more ill and the prevalence of bacterial infection is higher; a higher cut-off value (0.25 ng/ml) is therefore recommended. Recently, a new desktop molecular diagnostic machine, the GeneXpertw MTB/RIF assay (Cepheid, Table 2

Sunnyvale, CA, USA), has dramatically reduced the turnaround time for TB detection from 3–4 weeks to ,2 h. However, Xpert has an intermediate sensitivity between smear microscopy and TB culture, and the Xpert test alone may therefore not be sufficient for excluding TB.23 In addition, Xpert is costly and will not be used for routine screening in the near future. Urinary lipopolysaccharide antigen lipoarabinomannan (LAM) has also emerged as a promising assay for TB detection. However, recent data have shown that this assay has acceptable sensitivity only in HIV-infected patients with low CD4 cell counts.24 Although results for these two newly developed assays were promising, we believe PCT could still play a role in the diagnosis of pulmonary TB. In patients presenting with right upper lung infiltrates, a negative PCT test may raise the pre-test probability of pulmonary TB, and further testing with Xpert may help confirm the diagnosis. Our review should be interpreted in the light of several limitations. First, heterogeneous prevalence of

Summary of accuracy indicators of the eight studies included

Variable PCT, discrimination between TB and pneumonia with various aetiologies PCT, discrimination between TB and bacterial pneumonia CRP, discrimination between TB and bacterial pneumonia

Studies n

Sensitivity (95%CI)

Specificity (95%CI)



AUC (95%CI)

Diagnostic OR (95%CI)

I2 (95%CI)


0.42 (0.30–0.56)

0.87 (0.63–0.96)

3.20 (0.93–10.9)

0.67 (0.50–0.90)

0.57 (0.53–0.61)

4.25 (1.1–16.9)

80.8 (49.5–93)


0.78 (0.67–0.86)

0.85 (0.78–0.90)

5.19 (3.62–7.46)

0.26 (0.17–0.39)

0.89 (0.86–0.91)


0.76 (0.68–0.82)

0.78 (0.71–0.84)

3.21 (2.40–4.30)

0.29 (0.21–0.40)

0.85 (0.75–0.99)

19.9 (11.7–34) 11.3 (6.6–19.6)

0.0 (0.0–84.7) 0.0 (0.0–89.6)

CI ¼ confidence interval; AUC ¼ area under receiver operating characteristic; OR ¼ odds ratio; PCT ¼ procalcitonin; TB ¼ tuberculosis; CRP ¼ C–reactive protein; LR ¼ likelihood ratio.

PCT for diagnosis of PTB

pulmonary TB and heterogeneous comparison groups may explain part of the observed heterogeneity. Lack of clinical criteria in defining the clinical severity of pulmonary TB and the varying severity of TB disease in different studies were also a source of heterogeneity. Sources of heterogeneity are better evaluated with individual patient data or explored with metaregression analysis. Lack of access to these data, and the limited number of studies that met our eligibility criteria, limited our capacity to evaluate the underlying sources of heterogeneity. Second, selected studies in our meta-analysis used a wide range of cut-off values; sensitivities and specificities therefore varied across studies. Sensitivities and specificities may vary across studies and negatively correlate with each other due to the threshold effect. To cope with this problem, we constructed the HSROC curve that summarises pairs of sensitivities and specificities from each study included. The area under the SROC curve can be interpreted independently of the threshold effect. In addition, we used a bivariate modelling approach that allows for the calculation of mean sensitivity and specificity, adjusting for the inherent dependency in the paired sensitivity and specificity.19 The bivariate model and the hierarchical SROC model are mathematically equivalent; however, different parameterisations are used that lead to the different parameter summaries. A third limitation may be the relatively small number of studies and pooled sample size. The small number of studies available for our review prevented us from performing more extensive subgroup or meta-regression analysis. Studies with small sample sizes may be susceptible to Type II error and wide CIs. The less precise estimates deriving from the pooled effect make a definitive conclusion difficult. In conclusion, PCT values were mildly elevated in HIV-negative patients with pulmonary TB disease, particularly in those with disseminated TB. In HIVpositive patients with pulmonary TB, PCT was markedly elevated. Our data suggest that serum PCT is a useful biomarker for differentiating between pulmonary TB and bacterial pneumonia. It is important to realise that PCT is of little diagnostic value for differentiating pulmonary TB from other infections in patients with undifferentiated lower respiratory symptoms. Given the limited number of reports on this topic, large, prospective studies are needed to determine the diagnostic value of PCT in pulmonary TB.


4 5

6 7















Acknowledgements The authors thank P-S Hsieh and M C Lee from Medical Wisdom Inc. Aurora, CO, USA, for their help in statistical analysis and critical review. Conflict of interest: None declared.

References 1 Lawn S D, Zumla A I. Tuberculosis. Lancet 2011; 378: 57–72. 2 Dye C, Scheele S, Dolin P, Pathania V, Raviglione M C. Global burden of tuberculosis: estimated incidence, prevalence, and





mortality by country. World Health Organization Global Surveillance and Monitoring Project. JAMA 1999; 282: 677– 686. Assicot M, Gendrel D, Carsin H, Raymond J, Guilbaud J, Bohuon C. High serum procalcitonin concentrations in patients with sepsis and infection. Lancet 1993; 341: 515–518. Maruna P, Nedeln´ıkova´ K, Gurlich R. Physiology and genetics of procalcitonin. Physiol Res 2000; 49 Suppl 1: S57–61. Dandona P, Nix D, Wilson M F, et al. Procalcitonin increase after endotoxin injection in normal subjects. J Clin Endocrinol Metab 1994; 79: 1605–1608. Brunkhorst F M, Heinz U, Forycki Z F. Kinetics of procalcitonin in iatrogenic sepsis. Intensive Care Med 1998; 24: 888–889. Baylan O, Balkan A, Inal A, et al. The predictive value of serum procalcitonin levels in adult patients with active pulmonary tuberculosis. Jpn J Infect Dis. 2006; 59: 164–167. Kang Y A, Kwon S Y, Yoon H I, Lee J H, Lee C T. Role of Creactive protein and procalcitonin in differentiation of tuberculosis from bacterial community acquired pneumonia. Korean J Intern Med 2009; 24: 337–342. Naderi M, Hashemi M, Kouhpayeh H, Ahmadi R. The status of serum procalcitonin in pulmonary tuberculosis and nontuberculosis pulmonary disease. J Pak Med Assoc 2009; 59: 647–648. Nyamande K, Lalloo U G. Serum procalcitonin distinguishes CAP due to bacteria, Mycobacterium tuberculosis and PJP. Int J Tuberc Lung Dis 2006; 10: 510–515. Polzin A, Pletz M, Erbes R, et al. Procalcitonin as a diagnostic tool in lower respiratory tract infections and tuberculosis. Eur Respir J 2003; 21: 939–943. Porcel J M, Vives M, Cao G, et al. Biomarkers of infection for the differential diagnosis of pleural effusions. Eur Respir J 2009; 34: 1383–1389. Schleicher G K, Herbert V, Brink A, et al. Procalcitonin and Creactive protein levels in HIV-positive subjects with tuberculosis and pneumonia. Eur Respir J 2005; 25: 688–692. Ugajin M, Miwa S, Shirai M, et al. Usefulness of serum procalcitonin levels in pulmonary tuberculosis. Eur Respir J 2011; 37: 371–375. Cakir E, Deniz O, Ozcan O, . Pleural fluid and serum procalcitonin as diagnostic tools in tuberculous pleurisy. Clin Biochem 2005; 38: 234–238. Gatsonis C, Paliwal P. Meta-analysis of diagnostic and screening test accuracy evaluations: methodologic primer. AJR Am J Roentgenol 2006; 187: 271–281. Leeflang M M, Deeks J J, Gatsonis C, Bossuyt P M. Systematic reviews of diagnostic test accuracy. Ann Intern Med 2008; 149: 889–897. Whiting P, Rutjes AW, Reitsma J B, Bossuyt P M, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 2003; 3: 25. Reitsma J B, Glas A S, Rutjes A W, Scholten R J, Bossuyt P M, Zwinderman A H. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol 2005; 58: 982–990. Harbord R M, Whiting P, Sterne J A, et al. An empirical comparison of methods for meta-analysis of diagnostic accuracy showed hierarchical models are necessary. J Clin Epidemiol 2008; 61: 1095–1103. Lijmer J G, Bossuyt P M, Heisterkamp S H. Exploring sources of heterogeneity in systematic reviews of diagnostic tests. Stat Med 2002; 21: 1525–1537. Galbraith R F. A note on graphical presentation of estimated odds ratios from several clinical trials. Stat Med 1988; 7: 889– 894. Boehme C C, Nabeta P, Hillemann D, et al. Rapid molecular detection of tuberculosis and rifampin resistance. N Engl J Med 2010; 363: 1005–1015. Minion J, Leung E, Talbot E, Dheda K, Pai M, Menzies D. Diagnosing tuberculosis with urine lipoarabinomannan: systematic review and meta-analysis. Eur Respir J 2011; 38: 1398–1405.

PCT for diagnosis of PTB



R´esumer de fa¸con syst´ematique et quantifi´ee les preuves actuelles de la valeur de la calcitonine (PCT) dans la diff´erenciation de la tuberculose (TB) pulmonaire d’autres infections pulmonaires. M E´ T H O D E S : Nous avons cherche´ sur MEDLINE, EMBASE et la base Cochrane jusqu’a` aout ˆ 2013 des articles notifiant exclusivement la performance diagnostique du test de PCT ou le comparant aux autres biomarqueurs afin de diagnostiquer la TB pulmonaire. Les r´esultats sont sch´ematis´es dans les parcelles foresti`eres, les courbes caract´eristiques de fonctionnement du r e´ cepteur hi e´ rarchiques et les mod`eles a` deux variables a` effets al´eatoires. R E´ S U L T A T S : Nous avons trouv e´ neuf e´ tudes qui r´epondaient aux conditions et examin´e 951 e´ pisodes de suspicion clinique de TB ainsi que plus de 426 cas OBJECTIF :

confirm´es de TB. Selon une analyse a` deux variables, la sensibilit´e cumul´ee du test de PCT de diff´erencier les cas de TB d’autres infections e´ tait de 42% (IC95% 30–56) et la sp´ecificit´e e´ tait de 87% (IC95% 63–96). La sensibilit´e cumul´ee pour diff´erencier les cas de TB et la pneumonie bact´erienne e´ tait de 78% (IC95% 67–86) et la sp´ecificit´e e´ tait de 85% (IC95% 78–90). On a trouv´e une faible h´et´erog´en´eit´e dans les e´ tudes comparant les patients atteints de TB et la pneumonie bact´erienne. C O N C L U S I O N : Le r´esultat sugg`ere que la PCT a une sensibilit´e et une sp´ecificit´e g´en´eralement acceptables en termes de distinction entre une infection pulmonaire bact´erienne et une TB. Cependant, en raison de la sensibilit´e et de la sp´ecificit´e imparfaites du test, les d´ecisions m´edicales devraient eˆ tre bas´ees a` la fois sur les r´esultats du test a` la PCT et sur les signes cliniques. RESUMEN

O B J E T I V O : Recapitular de manera sistema´ tica y cuantitativa los datos fidedignos existentes sobre la utilidad de la prueba de la procalcitonina (PCT) con el fin de diferenciar entre la tuberculosis (TB) y las dema´s infecciones pulmonares. M E´ T O D O S : Se realizo ´ una busqueda ´ en las bases de datos MEDLINE, EMBASE, Cochrane y otras, hasta agosto del 2013. Se escogieron art´ıculos que notificaban exclusivamente el rendimiento diagnostico ´ de la prueba de la PCT o que la comparaban con otros marcadores biologicos, ´ con el fin de diagnosticar la TB pulmonar. Los resultados se esquematizaron en diagramas de bosque, curvas jera´rquicas de eficacia diagnostica ´ y modelos de efectos aleatorios bifactoriales. R E S U L T A D O S : Se encontraron nueve estudios que cumpl´ıan con los requisitos y examinaban 951 episodios de presunci on ´ cl´ınica de infecci on ´ tuberculosa adema´s de 426 casos confirmados de TB.

Segun ´ el ana´lisis bifactorial, la sensibilidad acumulada de la prueba de la PCT para diferenciar los casos de TB de otras infecciones fue 42% (IC95% 30–56) y la especificidad fue 87% (IC95% 63–96). La sensibilidad acumulada para diferenciar los casos de TB, de la neumon´ıa bacteriana fue 78% (IC95% 67–86) y la especificidad acumulada fue 85% (IC95% 78–90). Se observo´ una baja heterogeneidad en los estudios que comparaban los pacientes con TB y neumon´ıa bacteriana. ´ N : Los resultados de la busqueda CONCLUSIO ´ ponen de manifiesto una sensibilidad y especificidad aceptables de manera constante en los estudios, con respecto a la capacidad de la prueba de la PCT para diferenciar la TB de la neumon´ıa bacteriana. Sin embargo, dada la insuficiente sensibilidad y especificidad de la prueba, las decisiones m´edicas se deben basar en la prueba de la PCT y tambi´en en los hallazgos cl´ınicos.

Value of procalcitonin in differentiating pulmonary tuberculosis from other pulmonary infections: a meta-analysis.

To systematically and quantitatively summarise the current evidence on the utility of the procalcitonin test (PCT) in discriminating pulmonary tubercu...
432KB Sizes 0 Downloads 3 Views