Pleural Effusion Lipoproteins Measured by NMR Spectroscopy for

Article pubs.acs.org/jpr

Pleural Effusion Lipoproteins Measured by NMR Spectroscopy for Diagnosis of Exudative Pleural Effusions: A Novel Tool for Pore-Size Estimation Ching-Wan Lam* and Chun-Yiu Law Department of Pathology, The University of Hong Kong, Pokfulam Road, Hong Kong, China ABSTRACT: High-resolution proton nuclear magnetic resonance (NMR) spectrometry of biofluids has been increasingly used in laboratory diagnosis of various diseases. In this study, we extended the use of 1H NMR spectroscopy for laboratory diagnosis of exudative pleural effusions using pleural fluids. We compared this new NMR-based test with Light’s criteria, the current gold standard for laboratory diagnosis of exudative pleural effusions. We analyzed 67 samples of pleural effusions from patients with pulmonary malignancy (N = 32), pulmonary tuberculosis (N = 18), and congestive heart failure (N = 17). The metabolomes of pleural effusions were analyzed using 1H NMR spectroscopy on a Bruker 600 MHz spectrometer. Through a metabolome-wide association approach with filtering of insignificant markers (p value 0.5; ratio of pleural fluid lactate dehydrogenase (LDH) to serum LDH > 0.6; or pleural fluid LDH > 0.6 or 2/3 times the upper limit of normal in serum.17 However, in a study of 2115 subjects, this test had low specificities of 74% at a sensitivity of 98%.18 Others reported that the specificity ranged from 67 to 74%, despite a sensitivity of 90−98%.19−21 Patients with transudative pleural effusion would be mislabeled as having exudative effusions.22−24 A transudative pleural effusion misclassified as exudate will subject a patient to unnecessary investigations and treatments. With knowledge of increase in pleural permeability in exudative effusions, we hypothesize that the metabolomes of exudative pleural effusions and transudative effusions are very different. Metabolites with larger particulate size, such as lipoproteins, would pass through the pleural membrane from the circulation and result in a higher concentration of exudative effusions. In this work, metabolomes of pleural effusions were first studied using proton NMR spectroscopy, and the differential markers for exudates were determined through a metabolome-wide association approach and multivariate analysis (principal component analysis (PCA) and orthogonal partial least-squares-discriminant analysis (OPLS-DA)).

■

clinical assessment. There were 17 cases of CHF, which were confirmed clinically after excluding infectious or malignant causes. The classification of transudates and exudates was based on the underlying etiology. There were 50 cases of exudates and 17 cases of transudates. All samples were collected into sterile plain bottles and transferred immediately to the laboratory, followed by immediate centrifugation for 10 min with 3000 rpm at 4 °C to obtain a clear supernatant. The supernatant was mixed with buffer solution containing 100 mM phosphate buffer in 100% D2O containing 0.1% trimethylsilyl propanoic acid (TSP) (Sigma: 269913) in a 1:2 ratio (v/v). The NMR buffer used in this study contains D2O for locking and TSP for reference. The mixed sample was transferred to a 5 mm NMR tube for NMR analysis. Untargeted Metabolomic Profiling of Pleural Effusions Using 1H NMR and Biomarker Identification

All of the high-resolution proton NMR experiments were performed using a Bruker Avance 600 MHz NMR spectrometer (Bruker Biospin, Rheinstetten, Germany) equipped with a 5 mm PABBI proton probe. Chemical shifts of the spectra were referenced to the internal standard TSP. Spectra were recorded with a spectral width of 12019.230 Hz, at 298 K, into 32K data points before Fourier transformation. All spectra were recorded using an automatic sample changer, SampleXpress (Bruker Biospin), and the ICON-NMR software (Bruker BioSpin). Automatic tuning, frequency-locking on D2O, and shimming were performed using ICON-NMR before data acquisition. A Carr−Purcell−Meiboom−Gill (CPMG) pulse program was applied for acquisition of all 1H NMR spectra with water presaturation using a “cpmgpr1d” (Bruker pulse

MATERIALS AND METHODS

Samples

Pleural effusion samples from 67 patients were recruited. There were 32 cases of malignant pleural effusions that were confirmed by pleural biopsy, cytology, and clinical assessment. There were 18 cases of pulmonary tuberculosis (TB), which were confirmed by pleural biopsy, culture, PCR analysis, and 4105

dx.doi.org/10.1021/pr5004856 | J. Proteome Res. 2014, 13, 4104−4112

Journal of Proteome Research

Article

Table 1. Demographic Summarya characteristics

exudate (N = 50)

transudate (N = 17)

age sex (male) PE protein, g/L serum protein, g/L protein ratioc PE LDH, IU/L serum LDH, IU/L LDH ratioc causes

68 (63−74) 23 47 (45−49) 68 (66−70) 0.68 (0.66−0.71) 438 (346−531) 260 (198−323) 1.79 (1.35−2.23) malignancy (N = 32) tuberculosis (N = 18)

78 (72−85) 10 26 (20−31) 63 (60−67) 0.39 (0.30−0.48) 100 (76−125) 245 (210−279) 0.39 (0.25−0.53) CHF (N = 17)

p valueb 0.02 0.6 or 2/3 times the upper limit of normal in serum (155 IU/L). Total protein concentration and LDH activity were measured by rate biuret method and enzymatic rate method, respectively.

■

RESULTS

Demographic Data

Demographic data are summarized in Table 1. Exudates and transudates are classified based on the underlying etiology; 32 cases of malignant pleural effusions were confirmed by pleural biopsy/cytology. One case was diagnosed clinically; the patient presented with a progressively enlarging lung mass over 6 months. However, the patient succumbed before further radiological and histological investigations; 18 cases of pulmonary TB were confirmed by pleural biopsy, positive mycobacteria culture, or positive PCR identification for TB. One patient was diagnosed clinically; the patient presented with fever, pneumonia, and pleural effusion. Pleural biopsy showed the presence of granulomas, and the patient responded well to anti-TB treatment. Seventeen cases of CHF were diagnosed clinically, and their pleural biopsy/cytology showed no evidence of malignancy or tuberculous infections. All samples were collected before anticancer or anti-TB treatments. The mean age was 68 years old (range 63−74) for exudative pleural effusion and 78 years old (range 72−85) for transudative pleural effusion. Using Light’s criteria, the sensitivity was 98.0% and the specificity was 64.7%. Sixty samples (89.6%) were correctly classified, including 49 cases of exudates (98.0%) and 11 cases of transudate (64.7%). Seven samples (11.7%) were misdiagnosed including one case of exudate (2.0%) and six cases of transudates (35.3%).

Lipoprotein As Specific Markers for Exudative Effusions

Combining the results from univariate and multivariate analysis, lipoprotein was found to be the most discriminatory biomarker for exudative pleural effusions. The representative 1H NMR spectra of all samples are shown in Figure 5A. The identities of the buckets at 0.90, 1.30, and 5.34 ppm were matched to proton signals of lipoproteins corresponding to CH3, CH2, and −CHCH−, respectively.9−11 A marked increase in proton resonance signals was observed for CH3, CH2, and −CH CH− in the 1H NMR spectrum of exudates compared with that of transudate, as shown in Figure 5B. 4107



Article

Figure 3. PCA score plot (A) and loading plot (B). The PCA figure shows an intrinsically good clustering separation, suggesting that there is already a good differentiation of the two classes of sample.

Performance of Lipoprotein Particles as Biomarkers for Exudative Pleural Effusion

The contributions of CH3, CH2, and −CHCH− lipoprotein resonance to the spectrum were measured by integrating their respective proton resonances and normalized to the total intensity of the spectrum. The CH3 signals showed an AUROC of 0.96 (95% confidence interval (CI) = 0.89 to 0.99) with a sensitivity of 98.0% (95% CI = 89.4−99.9) and a specificity of 88.2% (95% CI = 63.6−98.5) at the optimal cutoff >0.040. The CH2 signals showed an AUROC of 0.96 (95% CI = 0.88−0.99) with a sensitivity of 84.0% (95% CI = 70.9−92.8) and a specificity of 100.0% (95% CI = 80.5−100.0) at the optimal cutoff >0.046. The −CHCH−signals showed an AUROC of 0.96 (95% CI = 0.88−0.99) with a sensitivity of 84.0% (95% CI = 70.9−92.8) and a specificity of 100.0% (95% CI = 80.5− 100.0) at the optimal cutoff >0.008. Using the optimal cut-offs, the CH3, CH2, and −CHCH− lipoprotein resonances could correctly predict 49 (98%), 42 (84%), and 42 (84%) out of 50 cases of exudates and 15 (88.2%), 17 (100%), and 17 (100%) out of 17 cases of transudates with an accuracy of 95.5, 88.1, and 88.1%, respectively. The ROC curves and box plots of CH3, CH2, and −CHCH− lipoprotein resonances are shown in Figure 6A,B.

Figure 4. More robust resolution of the two clinical groups is shown in the OPLS-DA score plot (A), loading plot (B), and coefficient plot (C).

determined clinically. The first step in the evaluation of pleural effusions is to classify them as transudates for nonpleural causes or exudates for pleural causes using Light’s criteria.16 However, Light’s criteria had limited specificity ranging from 67 to 74%,18−21 leading to misidentification of pleural transudates as exudates. Through metabolomic profiling, our study identified that lipoproteins were diagnostic markers for exudative effusions, which have an AUROC of 0.96, sensitivity of 98%, and specificity of 88%; this method thus demonstrated a significant improvement in the diagnostic specificity. The increased concentrations of lipoproteins observed in exudates are results of increased pleural permeability in exudative effusion secondary to the inflammatory responses caused by malignancies or infections. For transudate, free fluids accumulate secondary to the change of oncotic and hydrostatic pressures; therefore, no excess of pleural effusion lipoproteins

■

DISCUSSION Pleural effusion is an abnormal accumulation of fluids in the pleural space. The etiology of pleural effusion is not easily 4108



Article

Figure 5. Representative 1H NMR spectra (0.5−5.5 ppm) of all samples after normalization with total intensity of the spectrum (A) and proton resonance signals of CH3, CH2, and CHCH of lipoproteins. (B) Black: exudates; blue: transudates.

example, the lactate and glucose signals may reflect cellular activities in pleural effusion. To the best of our knowledge, there is no clinical diagnostic method to measure the pleural pore-size. The current method measured the average of all lipoprotein signals, which provide a scientific estimation of average pore-size. Despite the fact that capillary pore size is likely to be nonhomogenous in pleural membrane, the indirect measurement of average pore-size by lipoprotein is potentially clinically useful. Like glycated hemoglobin (i.e., HbA1c) currently in use for diagnosis and monitoring of diabetes mellitus,30 the level of glycated hemoglobin in whole blood provides a very good estimation of glucose exposure over 3 months, although glycated hemoglobin levels in individual red blood cells are different. Using an untargeted whole metabolome approach, we identified that lipoproteins, lactates, and glucoses were three major markers that differentiate exudates from transudates. Nevertheless, lipoproteins are metabolically more stable than

was observed, and the majority of the lipoprotein in transudates is of the smaller particulate size. The physiological upper limit of blood capillary pore-size (diameter) normally ranges between 5 and 12 nm,27 thus allowing the smaller size lipoproteins such as HDL (7−15 nm) to pass through into the pleural space. Capillary leak secondary to inflammation could result in a larger pleural pore-size ranged from 24 to 60 nm, thus allowing larger lipoproteins such as VLDL (31 to 64 nm) to pass into the pleural space28,29 (Figure 7). Using the principle of size exclusion, only large-sized lipoproteins could be measured and were elevated in exudative effusion but not in transudates. We propose that the detailed differences in lipoprotein particle subclass could be further determined by NMR-based methods used for serum lipoprotein subfractioning.8,12 In addition, the NMR-based metabolic profiling is not limited to lipoprotein analysis, which carries an exploratory potential of metabolic profiles to understand outliers. For 4109



Article

Figure 6. ROC curves showed CH3, CH2, and −CHCH− lipoprotein as markers for exudative effusions (A). Box plots comparing CH3, CH2, and −CHCH− lipoprotein between exudative (triangle) and transudative (circle) effusions (B).

glucose and lactate and do not require vials with anticoagulants for collections to stop cellular metabolism. In addition, the levels of lactate and glucose provide no additional information on the pleural permeability. Therefore, we proposed that lipoproteins would be the best markers to classify exudates and transudates. The monitoring of pleural capillary pore-size using lipoprotein as surrogate marker has never been reported before. Not only does this NMR-based method outperform the current gold standard, the Light’s criteria in identification of exudative

pleural effusions, but also the measurement of average pleural pore-size also provides an objective mean to monitor the patient’s progress because the pleural pore size will closely reflect the underlying pathomechanism of inflammation, that is, the capillary leaks. In contrast, Light’s criteria can only be used for diagnosis but not monitoring because there is no clear reflection of the underlying pathomechanism. We believe this new test will change the current management of pleural effusions and will become a new standard for clinical practice. 4110



Article

(6) Jeyarajah, E. J.; Cromwell, W. C.; Otvos, J. D. Lipoprotein particle analysis by nuclear magnetic resonance spectroscopy. Clin. Lab. Med. 2006, 26 (4), 847−870. (7) Otvos, J. D. Measurement of lipoprotein subclass profiles by nuclear magnetic resonance spectroscopy. Clin. Lab. 2002, 48 (3−4), 171−180. (8) Oostendorp, M.; Engelke, U. F.; Willemsen, M. A.; Wevers, R. A. Diagnosing inborn errors of lipid metabolism with proton nuclear magnetic resonance spectroscopy. Clin. Chem. 2006, 52 (7), 1395− 1405. (9) Otvos, J. D.; Jeyarajah, E. J.; Bennett, D. W. Quantification of plasma lipoproteins by proton nuclear magnetic resonance spectroscopy. Clin. Chem. 1991, 37 (3), 377−386. (10) Otvos, J. D.; Jeyarajah, E. J.; Hayes, L. W.; Freedman, D. S.; Janjan, N. A.; Anderson, T. Relationships between the proton nuclear magnetic resonance properties of plasma lipoproteins and cancer. Clin. Chem. 1991, 37 (3), 369−376. (11) Petersen, M.; Dyrby, M.; Toubro, S.; Engelsen, S. B.; Norgaard, L.; Pedersen, H. T.; Dyerberg, J. Quantification of lipoprotein subclasses by proton nuclear magnetic resonance-based partial leastsquares regression models. Clin. Chem. 2005, 51 (8), 1457−1461. (12) Mora, S.; Otvos, J. D.; Rosenson, R. S.; Pradhan, A.; Buring, J. E.; Ridker, P. M. Lipoprotein particle size and concentration by nuclear magnetic resonance and incident type 2 diabetes in women. Diabetes 2010, 59 (5), 1153−1160. (13) Wishart, D. S.; Jewison, T.; Guo, A. C.; Wilson, M.; Knox, C.; Liu, Y.; Djoumbou, Y.; Mandal, R.; Aziat, F.; Dong, E.; Bouatra, S.; Sinelnikov, I.; Arndt, D.; Xia, J.; Liu, P.; Yallou, F.; Bjorndahl, T.; Perez-Pineiro, R.; Eisner, R.; Allen, F.; Neveu, V.; Greiner, R.; Scalbert, A. HMDB 3.0–The Human Metabolome Database in 2013. Nucleic Acids Res. 2013, 41 (Database issue), D801−D807. (14) Beckonert, O.; Keun, H. C.; Ebbels, T. M.; Bundy, J.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Metabolic profiling, metabolomic and metabonomic procedures for NMR spectroscopy of urine, plasma, serum and tissue extracts. Nat. Protoc. 2007, 2 (11), 2692−2703. (15) DePew, Z. S.; Maldonado, F. The role of interventional therapy for pleural diseases. Expert Rev. Respir. Med. 2014, 8 (4), 465−477. (16) Light, R. W. Clinical practice. Pleural effusion. N. Engl. J. Med. 2002, 346 (25), 1971−1977. (17) Light, R. W.; Macgregor, M. I.; Luchsinger, P. C.; Ball, W. C., Jr. Pleural effusions: the diagnostic separation of transudates and exudates. Ann. Intern. Med. 1972, 77 (4), 507−513. (18) Porcel, J. M. Pearls and myths in pleural fluid analysis. Respirology 2011, 16 (1), 44−52. (19) Paramothayan, N. S.; Barron, J. New criteria for the differentiation between transudates and exudates. J. Clin. Pathol. 2002, 55 (1), 69−71. (20) Gazquez, I.; Porcel, J. M.; Vives, M.; Vicente de Vera, M. C.; Rubio, M.; Rivas, M. C. Comparative analysis of Light’s criteria and other biochemical parameters for distinguishing transudates from exudates. Respir. Med. 1998, 92 (5), 762−765. (21) Heffner, J. E.; Brown, L. K.; Barbieri, C. A. Diagnostic value of tests that discriminate between exudative and transudative pleural effusions. Primary Study Investigators. Chest 1997, 111 (4), 970−980. (22) Porcel, J. M. Identifying transudates misclassified by Light’s criteria. Curr. Opin. Pulm. Med. 2013, 19 (4), 362−367. (23) Kummerfeldt, C. E.; Chiuzan, C. C.; Huggins, J. T.; DiVietro, M. L.; Nestor, J. E.; Sahn, S. A.; Doelken, P. Improving the predictive accuracy of identifying exudative effusions. Chest 2014, 145 (3), 586− 592. (24) Bielsa, S.; Porcel, J. M.; Castellote, J.; Mas, E.; Esquerda, A.; Light, R. W. Solving the Light’s criteria misclassification rate of cardiac and hepatic transudates. Respirology 2012, 17 (4), 721−726. (25) Xia, J.; Broadhurst, D. I.; Wilson, M.; Wishart, D. S. Translational biomarker discovery in clinical metabolomics: an introductory tutorial. Metabolomics 2013, 9 (2), 280−299. (26) Chadeau-Hyam, M.; Ebbels, T. M.; Brown, I. J.; Chan, Q.; Stamler, J.; Huang, C. C.; Daviglus, M. L.; Ueshima, H.; Zhao, L.; Holmes, E.; Nicholson, J. K.; Elliott, P.; De Iorio, M. Metabolic

Figure 7. Schematic representation of lipoproteins’ flow in transudative pleural effusions (left) and exudative (pleural effusions). In transudative pleural effusion, the capillary pore size (diameter) ranged between 5 and 12 nm, which allows smaller HDL (7−15 nm) but not larger VLDL (31 to 64 nm) to pass into the pleural fluid. In exudative pleural effusion (right), the increased capillary pore-size (24−60 nm) allows lipoproteins of larger size to pass through.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel: 852-2255 5655. Fax: 852-2255 9915. Author Contributions

The manuscript was written through contributions of all the authors and all authors have given their approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Research Fund for the Control of Infectious Diseases (RFCID) from the Food and Health Bureau, the Government of Hong Kong SAR.

■

REFERENCES

(1) Nicholson, J. K.; Holmes, E.; Kinross, J. M.; Darzi, A. W.; Takats, Z.; Lindon, J. C. Metabolic phenotyping in clinical and surgical environments. Nature 2012, 491 (7424), 384−392. (2) Nicholson, J. K.; Lindon, J. C. Systems biology: Metabonomics. Nature 2008, 455 (7216), 1054−1056. (3) Yap, I. K.; Angley, M.; Veselkov, K. A.; Holmes, E.; Lindon, J. C.; Nicholson, J. K. Urinary metabolic phenotyping differentiates children with autism from their unaffected siblings and age-matched controls. J. Proteome Res. 2010, 9 (6), 2996−3004. (4) Lauridsen, M. B.; Bliddal, H.; Christensen, R.; DanneskioldSamsoe, B.; Bennett, R.; Keun, H.; Lindon, J. C.; Nicholson, J. K.; Dorff, M. H.; Jaroszewski, J. W.; Hansen, S. H.; Cornett, C. 1H NMR spectroscopy-based interventional metabolic phenotyping: a cohort study of rheumatoid arthritis patients. J. Proteome Res. 2010, 9 (9), 4545−4553. (5) Bjerrum, J. T.; Nielsen, O. H.; Hao, F.; Tang, H.; Nicholson, J. K.; Wang, Y.; Olsen, J. Metabonomics in ulcerative colitis: diagnostics, biomarker identification, and insight into the pathophysiology. J. Proteome Res. 2010, 9 (2), 954−962. 4111



Article

profiling and the metabolome-wide association study: significance level for biomarker identification. J. Proteome Res. 2010, 9 (9), 4620−4627. (27) Sarin, H. Physiologic upper limits of pore size of different blood capillary types and another perspective on the dual pore theory of microvascular permeability. J. Angiogenesis Res. 2010, 2, 14. (28) Mihaleva, V. V.; van Schalkwijk, D. B.; de Graaf, A. A.; van Duynhoven, J.; van Dorsten, F. A.; Vervoort, J.; Smilde, A.; Westerhuis, J. A.; Jacobs, D. M. A systematic approach to obtain validated partial least square models for predicting lipoprotein subclasses from serum NMR spectra. Anal. Chem. 2014, 86 (1), 543−50. (29) Ala-Korpela, M.; Korhonen, A.; Keisala, J.; Horkko, S.; Korpi, P.; Ingman, L. P.; Jokisaari, J.; Savolainen, M. J.; Kesaniemi, Y. A. 1H NMR-based absolute quantitation of human lipoproteins and their lipid contents directly from plasma. J. Lipid Res. 1994, 35 (12), 2292− 2304. (30) The International Expert Committee. International Expert Committee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care 2009, 32 (7), 1327−1334.

4112


Pleural Effusion Lipoproteins Measured by NMR Spectroscopy for

Recommend Documents