- Citado por SciELO
versión impresa ISSN 0102-3586
J. Pneumologia v.29 n.4 São Paulo jul./ago. 2003
The value of cytology and pleural biopsy in the differential diagnostic of nonspecific pleural effusions*
Vera Luiza CapelozziI; Danieli Cheke da RosaII; Aloísio S. Felipe da SilvaII
IAssociate Professot of Pulmonary Pathology at the Department of Pathology and Pulmonary Pathology Consultant to the Discipline of Pneumology
IIResident of Pathology
A significant percentage of pleural effusions remains without a diagnostic explanation. In such circumstances, the anatomical-pathological result of nonspecific chronic pleuritis should be revised. This is an important issue, not only for the pathologist, but mainly for the pneumologist, who usually makes the clinical-pathological correlation during the routine practice. Although the existence of established criteria is accepted, they are certainly subjective and make the communication by means of a common language difficult, mainly among pathologists. The authors recently studied 311 pleural biopsies with histo-pathological diagnoses of chronic nonspecific pleuritis. All specimens were reviewed and the histo-pathological parameters quantified by stereology. The patients were stratified according to the final diagnosis of their disease, including cases of chronic renal failure, vasculitis, pancreatitis, tuberculosis, cancer, and congestive cardiac failure, in this case considering the pleura as normal. This procedure allowed us to obtain a discriminating model, whose morphological subcriteria classified almost 90% of the nonspecific chronic pleuritis biopsies, according to their final clinical diagnoses, which included true nonspecific chronic pleuritis or tuberculosis pleuritis, paraneoplastic pleuritis, or even normal pleura. By adding to this model the biochemistry or differential cytology of the pleural liquid, its classificatory power reaches 99% of correctness. This study represents the result of the experience acquired over several years in the histo-pathological interpretation of pleural biopsies, based on the correlation between morphology and biochemistry and cytology of the pleural fluid. (J Pneumol 2003;29(4):225-34)
Key words: Biopsy, needle. Pleura/pathology. Pleura/cytology. Pleural effusion.
Abbreviations used in this study
DHL Dehidrogenase lactic
LE Lupus erythematosus
LES Systemic Lupus Erythematosus
This study reports on years of experience in the histological pathological interpretation of pleural biopsies based on the correlation of morphology with biochemistry and cytology of the pleural fluid.
The role of needle biopsy for diagnosis of pleural effusion is well defined. (1-6) Histological evidence of granulomatous pleuritis (Figure 1) or malignant effusion (Figure 2) gives an accurate diagnosis, however significance of a biopsy that shows only non-specific chronic pleuritis remains questionable. (Figure 3). In pleural specimens, considered normal or non-specific (Figure 3), there is a 77% possibility of absence of disease. (7) However, tumors or tuberculosis can eventually be detected in up to 40% of the patients previously submitted to biopsies that disclosed non-specific chronic pleuritis. (2) Multiple pleural biopsies (closed or open) increase the diagnostic potential and subsequent investigations will disclose if these patients are bearers of malignant or granulomatous pleuritis. (1, 6, 7) This will only be achieved with a marked increase in morbidity and costs. Even taking into account that certain clinical data might identify patients that should be submitted to complementary procedures, many will continue undiagnosed. (8)
There are few detailed studies on the morphological substrate for non-specific chronic pleuritis (9-11) and in those published there are no diagnostic criteria, even when the pleura are normal or was not reached by the biopsy. Generally, the histopathological spectrum is restricted to diagnoses of acute or chronic pleuritis. Even in the remodeling phases after inflammatory reaction, this diagnosis is part of the diagnosis of non-specific chronic pleuritis. Despite the acceptance of diagnostic criteria which are unquestionably subjective and although using a common language the relationship among pathologists continues to be difficult.
A study on 311 pleural biopsies was recently carried out (12) by the Discipline of Pulmonary Diseases (HC-FMUSP) whose histological pathological findings had disclosed non-specific chronic pleuritis. All specimens were reviewed and the histopathological parameters were quantified by stereology. (13) (Figure 4). Briefly, stereology is defined as one of the methods to quantify structures based on their spatial projection on a bi-dimensional plane. With a grid of 50 lines and of 100 points, mounted on a common light microscope lens it is possible to determine an area or volume that a given structure occupies in the tissue compartment of reference (for instance, fractional vascular area in the pleural sub-mesothelial screen compartment, fractional area of inflammation in the pleural sub-mesothelial adipose compartment, etc...). Thus, the fraction of the area of a structure (for example, vessels), is directly proportional to the number points which are found on a given structure divided by the total number of points that cover the histological field in analysis, that is the tissue reference compartment (for instance, the pleural sub-mesothelial screen compartment). Next, according to their final diagnosis patients were divided into strata, such as: uremia, vasculitis, pancreatitis, tuberculosis, cancer and congestive heart failure, in this case considering the pleura as normal. Such procedure provided a discriminating model, whose morphological sub-criteria, in non-specific chronic pleuritis, classified almost 90% of the biopsies according to the accurate clinical diagnosis. This includes truly non-specific chronic pleuritis, or tuberculous pleuritis, para-neoplasias or normal pleura. As stated earlier when biochemistry or differential cytology of the pleural fluid is included, the model for classification will be correct in 99% of cases.
For the purpose of this study histopathological analysis was systematized providing improved experience and knowledge to enhance diagnostic precision in clinical practice.
Role of the pathologist
Guidelines for diagnosis
The final histopathological diagnosis, resulting from analysis of a pleural biopsy rests upon six important requirements:
1) Pleura stratification into 3 distinct compartments (Figure 5):
- Visceral-parietal compartment
- Submesothelial screen compartment
- Submesothelial adipose tissue compartment
2) Definition of satisfactory pleural biopsy: A pleural biopsy is considered to be satisfactory when at least 2 out of the 3 histo-anatomic compartments defined above are present in a 3:1 proportion, in terms of submesothelial screen compartment and submesothelial adipose tissue compartment, respectively. This proportion requires a biopsy made with a Cope needle of at least 5 mm. The visceral-parietal compartment is virtual, but often is represented by a fibrino-leukocytic or recent hemorrhage pattern.
3) Receptivity by the general practice pathologists: The general practice pathologists should decide and inform in the report if the biopsy is representative of the parietal pleura or not, if it contains only skeletal muscle, or if it reached the liver, or if, in addition to parietal pleura it also included lung tissue and the corresponding visceral pleura. Often, the biopsy is representative of the parietal pleura, with presence of all the histoanatomic compartments, however from a histopathological point of view it is still normal. The pathologist should not hesitate to report that the pleura is normal. This situation may have clinical implications when the effusion is an exudate and the patient has, for instance, congestive heart failure. However, experience shows that most of those cases are diagnosed as non-specific chronic pleuritis, which has no meaning for the physician. It would be comparable to a diagnosis of chronic cervicitis in a cervix biopsy from a sexually active woman.
4) Identification of the histopathological changes found stratified by:
- type of tissue
- normal connective tissue
- granulation tissue
- reparatory fibrosis
- inflammatory reaction components
- polymorphonuclear cells
- mononuclear cells
- extension and distribution of the above mentioned changes throughout the pleural compartments
5) Determination of histopathological pattern resulting from the above mentioned changes, such as:
- granulomatous (Figure 1)
- neoplastic (Figure 2)
- granulation tissue (Figure 3)
- mononuclear cells in band like infiltrates
- truly non-specific (Figure 3)
- normal (Figure 6)
- purulent pleurisy (empyema) (Figure 7)
- fibrotic pleurisy(Figure 8)
6) Correlation to pleural fluid biochemistry and differential cytology:
Granulation tissue tuberculosis
The histopathological pattern in granulation tissues is the result of the onset of repair of an acute inflammatory reaction. The fibrin pseudo membrane and polymorphonuclear neutrophils permeated by small vessels and young fibroblasts is involved and progressively becomes incorporated into the pleural submesothelial screen compartment (Figure 3). The change from submesothelial screen to granulation tissue may be the outcome of infectious and non-infectious aggressions to the pleura. However, we demonstrated, (12) by quantitative analysis of pleural biopsies previously classified as non-specific chronic pleuritis, that this histopathological pattern is associated to tuberculous pleurisy in 100% of the cases. So, tuberculous pleurisy may be the fibrino-exudative inflammatory response of the submesothelial screen in the absence of granulomas.
Pathologists experienced in pleural biopsies may consider this association unrealistic and they may argue that fibrin is found in most biopsies obtained for diagnostic elucidation. However, classical experimental models (14), which reproduce the pathogenesis of tuberculous pleurisy in clinical practice (15), support our findings. (12) Thus, Patterson (14) and Stead et al. (15) described, in their studies on tuberculous pleurisy, the presence of sparse granulomas along with an extensive transformation of the submesothelial screen into granulation tissue. Theses findings may justify the larger amount of granulation tissue recovered by the Cope needle when compared to granulomas. Another important subject for discussion is that even when no intact mesothelium is found, pathologists may identify fibrin in the granulation tissue that modifies the pleural mesothelial screen compartment (specific site for tuberculosis) and fibrin that occupies the visceral-parietal compartment (specific for truly non-specific pleuritis). Indeed non-specific chronic pleuritis fibrin occupies the visceral-parietal compartment as a pattern or an isolated pseudo membrane, which has not become incorporated to the pleural submesothelial screen, as in the case of tuberculous pleuritis. Furthermore, this compartment shows weak signs of chronic inflammatory reaction and reparatory fibrosis instead of exudative inflammation. In these cases analysis of the pleural fluid usually supports the final diagnosis of tuberculous pleuritis. (17) It is a serous or bloody serous exudate (<10%) with a total concentration of proteins usually higher than 4.0 g/dL. The total leukocyte count is usually lower than 5,000/:l, with prevalence of lymphocytes. (Figure 9).
Absence of mesothelial cells as well as few eosinophils is also characteristic of tuberculous pleurisy. Low levels of glucose (<60 mg/dl) and low pH (<7.30) in the pleural fluid are found in up to 20% of patients with tuberculous pleurisy. Presence of adenosine deaminase > 40 mg/dL also complements diagnosis of tuberculous pleurisy.
Band-like mononuclear cell infiltrates
The histopathological pattern of band-like mononuclear cells infiltrates is characterized by inflammatory reaction in the pleural submesothelial adipose tissue compartment (Figure 10). The infiltrate is comprised of mononuclear cells with predominance of lymphocytes and plasma cells, distributed along the fatty lobes in a characteristic band-like arrangement. The band-like infiltrates are associated to mild changes in the vascular component of inflammatory reaction, that is, there is a slight proliferation of small vessels as granulation tissue. Furthermore, the band-like infiltrate cellularity can be so intense that it may cause pinching, frequently resulting in incorrect diagnosis of oat cell cancer or questionable biopsy by a pathologist not familiar with this event. In our experience (12) finding of band-like infiltrates histopathological pattern, modifying the histological architecture of the pleural submesothelial adipose tissue compartment is detected in 93.1% of the patients with malignant neoplasias (local or distant cancers and lymphoproliferative diseases lymphomas and leukemias). The remaining 6.9% were diagnosed with truly non-specific chronic pleuritis (one case) and normal pleura (3 cases). In those patients, further clinical investigations demonstrated viral infections of influenza / para-influenza type.
Paramalignant pleurisy, characterized by band-like mononuclear cell infiltrates, represents a pleural morphologic reaction in patients with malignancies, nevertheless without presence of malignant cells in the pleura at the moment of biopsy. Paramalignant pleurisy can be a response to several different malignant conditions which have not yet involved the pleura. (16, 17) Inflammatory changes in the submesothelial adipose tissue compartment represent an immune pleural response to tumor produced antibodies. A similar situation occurs in the post-cardiac injury syndrome, (18) in which an autoimmune phenomenon produces antibodies that trigger an exuberant pleuropericarditis with abundant mononuclear cells. Another exuberant mononuclear cells infiltrate is well documented as an immune phenomenon, characterizing peripheral reactions to tumors, as is the case of gastric cancer. (19) Also, the band-like mononuclear infiltrate in the submesothelial adipose tissue compartment may be the result of lymphatic obstruction due to presence of local tumors. A similar infiltrate, associated to non-tuberculous pleurisy was described by Nagata et al. (9) Irrespective of the events related to the relationship pleura-tumor in cases of paraneoplastic pleurisy a band-like mononuclear cell infiltrate points to a common characteristic. Thus, this lymphoproliferative response of the pleura to a distant or local tumor may require thoracoscopy or other diagnostic procedures.
Pathologists may argue that chronic inflammation and its sequels can interfere with recognition of limits between the pleural submesothelial screen compartment and the submesothelial adipose tissue compartment. We have observed (12) that chronic inflammation and its sequels to the submesothelial screen compartment are characterized by a mononuclear cell infiltrate around the vessels, with a thin layer of collagen tissue (fibrocellular proliferation), modifying the screen, tending to expand in depth from the mesothelial screen, that is, from the external elastic lamina (Figure 5). The submesothelial screen can, thus, be easily detached from the submesothelial adipose tissue, probably because the pleural compartments have not been destroyed by the original inflammatory reaction and also, because the internal elastic lamina remains intact. (16) Thus, cellular density of the band-like infiltrate in the submesothelial adipose tissue permits to recognize the inter-compartment limits.
Biochemistry of the pleural fluid shows low levels of glucose and low pH, with an increase in DHL. The differential cell count shows an increase in lymphocytes with irregularly outlined nuclei (Figure 11) and presence of immunoblasts.
Truly non-specific chronic pleurisy
In this histopathological pattern of pleural involvement, the basic reaction is characterized by a mild to moderate infiltrate around the vessels in the submesothelial pleura, without granulation tissue (Figure 3). As we have demonstrated, the discriminating model classified 84.2% of the pleural biopsies in the histopathological pattern of truly non-specific chronic pleurisy or non-tuberculous benign pleurisy, that is, a histopahtological substrate common to several conditions. (7-9) The other 15.8% biopsies were erroneously classified as normal pleura. Under the generic name of truly non-specific chronic pleurisy, infections (viral and Mycoplasma pneumonia), collagenosis (SLE, rheumatoid arthritis, Sjörgen syndrome), vasculitis (Wegeners granulomatosis, Churg-Strauss syndrome), post cardiac injury syndrome, embolism, uremia and pancreatitis, can be found. The definitive diagnosis will be established after further clinical investigation, together with minor histopathological and cytological criteria that will be discussed below.
Truly non-specific chronic pleurisy
The histopathological exam discloses changes in the pleural submesothelial screen compartment (Figure 5). The inflammatory reaction causes lymphocytes, plasma cells and hystiocytes recruitment, which tend to spread around the vessels, bringing about endothelial tumefaction. There is a mesothelial cell hyperplasia, showing characteristic cytopathic effects (Figure 12), suggesting tests for viral antigens by immunohistochemistry or electronic microscopy. In our study, (12) adenovirus and influenza/para-influenza represented 5.2% of the infectious causes leading to effusions due to truly non-specific chronic pleurisy.
In such cases biochemistry of the pleural fluid helped to reach diagnosis, excluding tuberculous and paraneoplastic pleurisy. In tuberculous effusions, serous or bloody serous (<10%) exudate is usually found with a total protein concentration often higher than 4.0 g/dL. The total leukocyte count is usually lower than 5,000/:l, with prevalence of lymphocytes. Absence of mesothelial cells is another characteristic of tuberculous pleurisy, as well as few eosinophils. Low glucose levels (<60 mg/dl) and low pH (<7.30) in the pleural fluid are present in up to 20% of patients with tuberculous pleurisy. Presence of adenosine deaminase (> 40g/dL) can also complete diagnosis of tuberculous pleurisy. In malignant effusions, biochemistry of the pleural fluid shows low glucose levels and low pH with increase of DHL. The differential cell count shows an increase in lymphocytes, with irregularly outlined nuclei, presence of immunoblasts and intensely dyskaryotic cells leading to a diagnosis of malignancy.
However, the fluid cytology may show presence of atypical mesothelial cells (Figure 13), stressing the previously mentioned cytopathic viral effect.
Truly non-specific chronic pleurisy
Collagenosis may cause pleural effusions. Study of pleural biopsies will reveal a histopathological pattern centered on truly non-specific chronic pleuritis. Therefore, in addition to the perivascular infiltrate at the submesothelial screen, the pathologist will be able to identify leukocytoclastic vasculitis and capillaritis (Figure 14).Thus, a histopathological pattern possibly associated to systemic lupus erythematosus will be characterized. In other situations, a necrotic nodule, with histiocytes in a palisade arrangement, together with histopathological signs of truly non-specific chronic pleuritis, might suggest the characteristic pleural involvement pattern of rheumatoid arthritis.
In both situations, the biochemical study of the pleural fluid will show low glucose levels, low pH and presence of rheumatoid factor, whereas the cytological study will show LE cells. Another differential characteristic is the presence of elongated histiocytes (Figure 15) and giant multinucleated cells (Figure 16). However, one should bear in mind that mostly, findings in the pleural fluid are infrequent and rarely characterize the disease.
Truly non-specific chronic pleurisy
Some systemic or lung vasculitis, such as Wegeners granulomatosis and Churg Strauss syndrome may present with pleural effusion and non-tuberculous benign pleurisy. If the pathologist identifies necrotizing granulomatous vasculitis along the small vessels of the pleural submesothelial screen added (Figure 17) to the histological pattern of basal truly non specific chronic pleurisy, possibility of Wegeners granulomatosis should be considered. Finding of eosinophilic vasculitis might also be suggestive of a possible pleural involvement in the Churg-Strauss syndrome. (Figure 18) together with the histopathological pattern of truly non-specific chronic pleurisy (Figure 3). In both situations, the pleural fluid will show increase of glucose and pH and decrease in HDL. At cytological exam of the pleural fluid, multinucleated giant cells together with smear rich in eosinophilic debris, or the finding of eosinofilia (Figure 19) will represent additional information supporting the two conditions. However, in diagnostic practice it must be considered that findings of the pleural fluid are not characteristic and do not contribute to diagnosis.
Truly non-specific chronic pleurisy
Post cardiac injury syndrome
In some situations of injury to the heart the myocardium reacts specifically with an auto-immune reaction, producing auto-antibodies which provoke a pleuropericarditis with abundant mononuclear cells liable to form a cuff around the vessels (Figure 20).
Truly non-specific chronic pleurisy
Embolism, uremia, pancreatitis
The histopathological pattern of truly non-specific chronic pleurisy can form the morphological substrate resulting from the pleural response in pulmonary embolism, uremia or pancreatitis, together with hemosiderosis (Figure 21). This would be indicative of chronic bleeding at the pleural submesothelial screen compartment. Biochemical analysis of the pleural fluid will disclose increase of Glucose, pH, LDH and amylase, together with hemosiderin rich histiocytes (Figure 22) also signs of erythrophagocytosis at differential cytology exam.
As demonstrated in our study, histological finding of normal pleura may be the anatomical substrate of patients bearers of clinical conditions associated to pleural effusion transudates. (12) It was shown that 72.7% of the 22 cases of non-specific chronic pleuritis, re-classified according to quantitative criteria as normal pleura were associated to congestive heart failure, liver cirrhosis and nephrotic syndrome (Figure 6).
In such cases, diagnosis is achieved by adding a differential cytology exam of the pleural fluid showing lymphocytes with evident nucleoli (Figure 23). Evident nucleoli indicate nuclear activity, that is, proliferating activity, attesting that cells are immature (need of intense cell division, with appearance of young or immature cells). In the case of congestive heart failure, liver cirrhosis and nephrotic syndrome, the immune defense system is recruited with such an intensity that time is too short for maturation of its elements, mainly the lymphocytes, which will comprise a pleural fluid with exuberant evidence of their nucleoli.
Purulent pleurisy (empyema)
The characteristic empyema substrate is an acute suppurative inflammatory reaction (Figure 7). Whole and degenerated neutrophils are meshed into fibrin nets, together with necrotic cell wastes, with extensive and intense modification of the histological architecture of all pleural compartments. In a logical temporal sequence, that once identified will produce a more reliable histological parameter for differential diagnosis with desmoplastic mesothelioma, the stages of inflammation, stratified by the suppurative acute phase, followed by organization into granulation tissue and finally, a fibrotic-hyalinizating repair response were recognized.
The main and predominant component of pleural fluid analysis will be degenerating neutrophils (Figure 24).
Fibrotic pleurisy composes the histopathological pattern for two situations: the fibrous pleural plaque, when localized, and diffuse fibrous pleurisy, in case of a more extensive involvement of the pleural surface (Figure 8). Both situations are histopathological markers of exposure to asbestos. (20)
Systematic approach of the parietal pleura in needle biopsy specimens, according to the guidelines proposed above and resulting from a formerly carried out quantitative analysis, associated to information brought forth by differential biochemical and cytological exams of the pleural fluid, will increase diagnostic precision in cases that, otherwise, would continue to be generically classified as non-specific chronic pleuritis.
1. Abrams LD. A pleural punch biopsy. Lancet 1958;1:30-1. [ Links ]
2. Cope C. New pleural biopsy needle: preliminary study. JAMA 1958; 167:1107-8. [ Links ]
3. Donohoe RF, Katz S, Matthews MJ. Pleural biopsy as an aid in the etiologic diagnosis of pleural effusion: review of the literature and report of 132 biopsies. Ann Intern Med 1958;48:334-62. [ Links ]
4. Levine H, Cugell DW. Blunt-end needle biopsy of the pleura and rib. Arch Intern Med 1962;109:516-25. [ Links ]
5. Prakash UBS, Reiman HM. Comparison of needle biopsy with cytologic analysis of 414 cases. Mayo Clin Proc 1985;60:158-64. [ Links ]
6. Sahn AS. The pleura. Am Rev Respir Dis 1988;138:184-234. [ Links ]
7. Poe RH, Israel RH, Utell MJ, Greenblatt DW, Kallay MC. Sensitivity, specificity and predictive values of closed pleural biopsy. Ann Intern Med 1984;144:325-8. [ Links ]
8. Leslie WK, Kinasewitz GT. Clinical characteristics of the patient with nonspecific pleuritis. Chest 1985;94:603-8. [ Links ]
9. Nagata N, Kawarada Y, Shigematsu N, Ishibashi T. Subpleural mononuclear cells infiltration: significance in the differential diagnosis of pleuritis showing non-specific histologic findings. Chest 1990;98:1116-20. [ Links ]
10. Von Hoff DD, LiVolsi V. Diagnostic reability of needle biopsy of the parietal pleural: a review of 272 biopsies. Am J Clin Pathol 1975;64: 200-3. [ Links ]
11. Mestiz P, Purves MJ, Plloard AC. Pleural biopsy in the diagnosis of pleural effusion. Lancet 1958;2:1349-53. [ Links ]
12. Capelozzi VL, Saldiva PHN, Antonângelo L, de Carvalho TS, Logulo A, de Carvalho CR, et al. Quantitation in inflammatory pleural disease to distinguish tuberculous and paramalignant from chronic non-specific pleuritis. J Clin Pathol 1997;50:935-49. [ Links ]
13. Gundersen HJG, Bendtesen TF, Korbo L, Marcussen N, Moller A, Nielsen K, et al. Some new, simple and efficient stereological methods and their use in pathological research and diagnosis. APMIS 1988; 96:379-94. [ Links ]
14. Paterson RC. The pleural reaction to inoculation with tubercle bacilli in vaccinated and normal guinea pigs. Am Rev Tuberc 1917;1:353-71. [ Links ]
15. Stead WE, Eichenholz A, Stauss H-K. Operative and pathologic findings in twenty-four patients with syndrome of idiopathic pleurisy with effusion presumably tuberculous. Am Rev Tuberc 1955;71:413-502. [ Links ]
16. Hebert A. Pathogenesis of pleurisy, pleural fibrosis, and mesothelial proliferation. Thorax 1986;41:176-89. [ Links ]
17. Sahn AS. Malignant pleural effusions. Semin Respir Med 1987;9:43-53. [ Links ]
18. Stelzner TJ, King Jr TE, Antony VB, Sahn SA. The pleuropulmonary manifestation of the postcardiac injury syndrome. Chest 1983;84:383-7. [ Links ]
19. Watanabe H, Enjoji M, Imai T. Gastric carcinoma with lymphoid stroma. Its morphologic characteristics and prognostic correlations. Cancer 1976;38:232-43. [ Links ]
20. Capelozzi VL. Asbesto, asbestose e câncer: critérios diagnósticos. J Pneumol 2001,27:206-18. [ Links ]
Vera Luiza Capelozzi
Professora Associada do Departamento de Patologia, Faculdade de Medicina da USP
Av. Dr. Arnaldo, 455, 1º andar, sala 1.143
01246-903 São Paulo, SP
Tel. (11) 3066-7427
Fax (11) 5096-0761
Received for publicaion on the 6th of November, 2002
Approved, after review on the 10th of April, 2003
* This research was done at the Department of Pathology of Faculdade de Medicina da Universidade de São Paulo-USP, São Paulo, SP. Scientific support: Fapesp Fundação de Amparo à Pesquisa do Estado de São Paulo. LIM05-HCFMUSP Laboratório de Investigação Médica Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo. CNPq Conselho Nacional de Desenvolvimento Científico e Tecnológico.