Differentiating between malignant and idiopathic pleural effusions: the value of diagnostic procedures
From the Departments of Internal Medicine, 1Thoracic Surgery, 2Radiology, 3Oncology and 4Preventive Medicine and Epidemiology, Hospital General Vall d’Hebron, Autonomous University, Barcelona, Spain
Address correspondence to Dra. Carmen Alemán, Hospital Vall d’Hebron. Department of Internal Medicine, Passeig Vall d’hebron 119–129, Barcelona 08035, Spain. email: 29261cal{at}comb.es
Received 26 July 2006 and in revised form 29 January 2007
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Background: Which diagnostic procedures should be used to differentiate between idiopathic and malignant pleural effusions, is not well established.
Aim: To identify which parameters allow differentiation between idiopathic and malignant pleural effusions.
Design: Case-note review.
Methods: Over a 12-year period, we treated 1014 consecutive pleural effusion patients, of whom 346 were diagnosed as having an idiopathic or malignant aetiology. We analysed medical history, chest X-ray, pleural fluid analysis (biochemical, microbiological and cytological), chest CT scan and additional examinations that were used according to clinical findings, and compared them with the eventual diagnosis and outcome.
Results: Eighty-three patients with idiopathic effusions and 263 with malignant effusions were included. Idiopathic pleural effusion resolved in 47 patients, improved in 20 and persisted in 16. Biochemical pleural fluid analysis did not predict these outcomes. A history of neoplasm, chest X-ray and CT features, as well as additional examinations according to clinical findings, established a diagnosis or suspicion of malignancy in 256 (97.7%) of the 263 patients who received a diagnosis of malignant effusion. Diagnostic thoracoscopy was helpful in seven patients in whom malignant disease was strongly suspected, despite the absence of other pathological findings.
Discussion: Non-invasive complementary examinations generally allowed accurate differentiation between malignant and idiopathic effusions. Patients with idiopathic pleural effusions generally had favourable outcomes.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Idiopathic effusion is usually defined as any effusion that remains undiagnosed after biochemical study, cytology studies and pleural biopsy,1–3 without taking into account the use of other examinations. However, it may be more appropriate to classify as idiopathic or indeterminate those pleural effusions that remain undiagnosed after exhaustive assessment, including additional examinations according to the clinical presentation.4–7 This concept has already been used in some studies.8
When the diagnosis remains uncertain even after chest CT study, there is no consensus regarding which additional tests should be performed. Recent reviews propose a conservative approach, with clinical and radiological observations if the patient improves, and further study of the pleural space by thoracoscopy if pleural effusion persists.9
One of the main problems in establishing the diagnosis in patients with exudative pleural effusion is differentiating between those secondary to neoplastic disease and those with an idiopathic cause.2,10 The factors associated with resolution of idiopathic pleural effusions during follow-up, and whether it is possible to establish a diagnosis of malignant disease using currently available diagnostic examinations, both remain uncertain.8
Since differentiation between malignant and idiopathic pleural effusions has particular clinical relevance, we reviewed our clinical experience to identify factors that might allow differentiation between idiopathic and malignant pleural effusions, and to investigate the prognostic factors predicting the outcome of idiopathic pleural effusions.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
From November 1992 to September 2004, 1014 consecutive patients with pleural effusions presented to our hospital, and were classified on the basis of clinical symptoms, pleural fluid characteristics and additional examinations, as follows.9,11–13
(i) (n = 162) Transudates: transudative effusion according to the Light criteria, or pleural effusion secondary to transudate, despite exudative criteria.9 (ii) (n = 84) Uncomplicated parapneumonic effusion: pneumonia and pleural fluid with non-purulent appearance, negative on Gram stain and culture, pH >7.2 and/or glucose >40 mg/dl. (iii) (n = 139) Complicated parapneumonic effusion/empyema: pneumonia and pleural fluid with one or more of: macroscopic pus, positive on Gram stain or culture, pH <7.2, glucose <40 mg/dl. (iv) (n = 150) Neoplastic effusion: malignancy on pleural cytological or histological examination. (v) (n = 113) Probable malignant effusion: active neoplastic disease in any location, one or more pleural cytological or histological examinations negative for malignant disease, but no other cause detected, hence attribution to the underlying neoplasm. (vi) (n = 129) Tuberculous effusion: pleural fluid positive on Ziehl-Nielsen stain or Löwenstein-Jensen culture, or pleural biopsy specimen showing granulomas or positive Ziehl-Neelsen stain or positive Löwenstein-Jensen culture, or pleural fluid adenosine deaminase (ADA) levels >43 UI/l with consistent clinical signs and symptoms, pleural fluid lymphocyte predominance and good response to anti-tuberculosis treatment. (vii) (n = 128) Miscellaneous: pleural fluid diagnosed as having other causes. (viii) (n = 83) Idiopathic: patients who could not be classified into any of the other diagnostic categories. (ix) (n = 26) Not classified: the pleural effusion could be attributed to several possible aetiologies.
We analysed demographic data, medical history, clinical symptoms, complete physical examination, routine blood testing, radiological features on chest X-ray (unilateral or bilateral effusion, size of effusion and presence of a single mass or multiple nodules), pleural fluid analysis (glucose, proteins, lactate dehydrogenase, pH, ADA, cells and differential cell count), pleural culture and pleural cytological examination. Other examinations were performed at the discretion of the attending physician: chest computed tomography (CT) scanning (unilateral or bilateral, size of the effusion, single mass, mediastinal lymphadenopathy, multiple nodules, atelectasis, nodular pleural enlargement or any combination of these features), closed-needle pleural biopsy and thoracoscopy (when malignancy was strongly suspected from non-invasive procedures, but with no definitive diagnosis, and in the absence of clinical or radiological improvement, malignancy needed to be ruled out), and additional examinations performed according to the clinical findings, including bronchoscopy, mammography, gynaecological examination, digestive tract endoscopy, abdominal ultrasonography, abdominal CT scanning, bone scan, bone-marrow biopsy, fine-needle aspiration of the pulmonary mass, liver or lymph node, and lung scintigraphy.
In patients with neoplastic or probable malignant effusion, although medical data and complementary examinations were recorded during hospital stay, data about clinical course were obtained from medical history, recording the date of death when available or the date of transferral to a palliative care centre as the final follow-up visit for those receiving only symptomatic treatment.
In effusions of unknown origin, follow-up was recorded with clinical, physical and chest-X-ray examination at each visit, and, where relevant, the establishment of an alternative diagnosis was recorded.
Statistical analysis
The 
2 test or Fisher's exact test, as appropriate, was used to assess the relationships between prognostic factors and spontaneous resolution of idiopathic pleural effusions at 6 months follow-up. The 2 test was used to compare data for neoplastic vs. probable malignant effusions, and subsequently between malignant (neoplastic and probable malignant) vs. idiopathic effusions. Because of the presence of extreme values, Kruskal-Wallis analysis of variance was used to compare distribution of biochemical variables between neoplastic vs. probable malignant and subsequently between malignant vs. idiopathic effusions.14
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Among the 1014 patients with pleural effusions, 263 patients had a diagnosis or suspicion of malignant pleural effusion (150 neoplastic and 113 probable malignant), mean age 70 ± 13 years. Ninety-six had could not be classified into any of the diagnostic categories at discharge, but in 11 of these a diagnosis explaining the pleural effusion was established at follow-up: five with various types of collagen disease or vasculitis (two with rheumatoid arthritis); three with exudative effusions who were found to have heart failure (dilated myocardiopathy with impaired cardiac function), the effusion resolving with appropriate therapy; one with hepatitis C virus cirrhosis; and two cases initially diagnosed with idiopathic effusions in whom relevant abdominal diseases (pancreatitis and infected hepatic hydatid cyst) had been present a few weeks before, and the pleural effusion resolved spontaneously without recurrence (hence these cases were attributed to the abdominal disease). In two further patients, no diagnosis could be established, but pleural effusion resolved spontaneously during the first month. Since follow-up in these patients was only 1 and 5 months respectively, they were not included in the definitive diagnosis of idiopathic pleural effusion. Hence a definitive diagnosis of idiopathic pleural effusion was established in only 83 (9.7%) patients, mean age 69 ± 16 years.
Analysis of the differences between the malignant effusion groups showed that patients with neoplastic effusion underwent pleuroscopy and required a chest drain (with or without pleurodesis) more often (p < 0.001), while prior history of neoplasm (p = 0.036) and metastatic dissemination on chest X-ray (p = 0.021) and abdominal ultrasound study (p = 0.03) was documented more often in patients with probable malignant effusion. Pulmonary, hepatic, bone or peripheral lymph node metastasis was documented in 68/150 patients with neoplastic effusion but in 69/113 patients with probable malignant pleural effusion. No significant differences were found for the other clinical parameters, survival time, pleural fluid analysis, chest CT results or other additional tests performed. Subsequently, the malignant effusion groups were combined for purposes of comparison with the idiopathic group.
Baseline clinical characteristics and significant differences between groups are summarized in Table 1. Results of biochemical analyses and cell counts in the patients with pleural effusion are shown in Table 2, and the data from additional examinations in the various effusion groups, with their significant differences, are summarized in Table 3.
|
|
|
Follow-up data were available for all the patients with idiopathic effusion, 135 of those with neoplastic pleural effusion, and 91 of those with probable malignant effusion.
Analysing patients with negative cytology and evidence of pleural effusion alone on the chest X-ray, the following variables were significantly different between malignant and idiopathic effusions: prior history of neoplasm (p < 0.001), constitutional syndrome (p < 0.001), effusion size greater than one-third of the hemithorax (p < 0.001), and findings suggesting malignant disease on chest CT (p < 0.001), abdominal ultrasound or CT (p < 0.001), mammography (p = 0.046), gynaecological examination, including pelvic sonography (p = 0.033), broncoscopy (p = 0.005) or bone scan (p = 0.028). Table 4 shows the sensitivity and specificity of these variables for the diagnosis of malignancy in this group of patients.
|
As shown in Figure 1, after undergoing chest CT, abdominal CT, gynaecological examination and other additional tests in accordance with the patient's signs and symptoms and taking into consideration the prior history of neoplasm, the diagnosis of suspected malignant disease was established in 256 of the 263 pleural effusions ultimately classified as malignant. The seven patients in whom malignant disease was not suspected after the additional examinations included five with mesothelioma, one with adenocarcinoma of unknown origin, and one with large-cell lung cancer. Five of these seven patients presented with associated constitutional syndrome, and in all cases pleural effusion size was greater than one-third of the hemithorax (in three of them, greater than two-thirds).
|
Follow-up lasted a mean of 1100 days (range 215–3520 days). Idiopathic pleural effusion resolved in 47 patients, improved in 20 and persisted without relevant symptoms in 16. Six patients had a history of neoplastic disease (Table 5). No relationship was found between the neoplasm and the effusion; hence, these patients were classified in the idiopathic effusion group. During follow-up, six additional neoplasms were diagnosed, all at least 12 months after the onset of pleural effusion (Table 6).
|
|
The following clinical factors were statistically associated with spontaneous resolution at 6 months: female sex (p = 0.013), age <65 years (p = 0.021), presence of fever (p = 0.003) or chest pain (p = 0.002) at clinical onset, and absence of COPD (p = 0.02).
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Idiopathic pleural effusion currently accounts for 5% to 25% of the aetiologies of pleural effusion, in reported series.1,8,9,15–17 The few reports on patients in whom the cause of pleural effusion has remained unclear after biochemical, microbiological, and cytological analysis of pleural fluid have mainly included effusions associated with neoplasms (neoplastic and probable malignant effusions) and tuberculous effusions. Although the study of these effusions has classically included closed-needle pleural biopsy or thoracoscopy to obtain specimens for histological study,10,18–20 pleural fluid markers such as adenosine deaminase (ADA) and interferon-
are now available, with a sensitivity and specificity close to 100% for the diagnosis of tuberculous pleuritis in areas with high prevalence of this disease.21–24 In addition, in both classic and more recent series of patients with idiopathic pleural effusion, despite a positive tuberculin test, many patients do not develop pleural tuberculosis during follow-up.8,15,25–28 Many of our patients underwent additional tests, guided by the symptoms and physical examination. Our criteria to define idiopathic effusion included pleural fluid ADA concentration and none of our patients categorized in this way were diagnosed with tuberculosis at any site after an adequate follow-up period. Differentiation between malignant and idiopathic pleural effusions has particular clinical relevance.8,9,15,29,30 In our series, 54% of the patients with malignant effusions died within 6 months of the diagnosis of pleural effusion, whereas only one patient with a diagnosis of idiopathic pleural effusion within 6 months of diagnosis, and the cause of death was not related to the effusion.
The combination of malignant pleural fluid cytology (114 patients) and the presence of a lung mass or multiple nodules on chest X-ray (76 patients) established the diagnosis or the suspicion of malignancy in 162 patients. It is currently accepted that when a diagnosis cannot be made after the initial assessment, helical chest CT is justified.2,31,32 After chest CT, the diagnosis of suspected malignant disease was established in 212/263 patients with malignant effusions, whereas atelectasis suggestive of endobronchial obstruction was detected in only 2/83 patients with idiopathic pleural effusion.
Forty-four of the 51 patients whose diagnosis was unclear after cytology and chest X-ray or CT scanning had a history of neoplasm, or findings consistent with malignant disease on abdominal CT or gynaecological examination, or in other additional tests. In six of the remaining seven patients with a definitive diagnosis of malignant effusion, this diagnosis was obtained by pleuroscopy. In the patient with large-cell carcinoma, the diagnosis was established by closed-needle pleural biopsy. Thoracoscopy is mainly diagnostic in neoplastic pleural effusions and tuberculous pleurisy; its utility in diagnosing other benign effusions is limited.20,33–37 In patients with malignant pleural effusion, diagnostic thoracoscopy is used to thoroughly investigate the presence of a neoplasm when there is an elevated clinical suspicion34,35 and no evidence of the disease in other additional tests. In a study of 93 patients referred for thoracoscopy, which included patients evaluated in the Departments of Internal Medicine or the Emergency Medicine, who are also included in our study, the presence of four criteria (symptomatic period >1 month, absence of fever, blood-tinged pleural fluid and chest CT scan) established an adequate classification of malignancy in 95% of the patients. Thoracoscopy is also used to determine the extension of a previously diagnosed neoplasm,38,39 and to permit the therapeutic application of pleurodesis.9
After a mean follow-up of 1100 days, certain clinical characteristics of patients with idiopathic pleural effusion were found to correlate with better prognosis and indicated a higher chance of resolution of the effusion: female sex, age <65 years, fever or chest pain at onset and absence of COPD. This series of idiopathic pleural effusion included six patients who developed neoplastic disease during follow-up, consistent with the incidence of neoplastic disease in the European population during the same period.40,41 The interval between the diagnosis of pleural effusion and the onset of neoplasm, together with the clinical findings, suggest that the effusion had no relationship with the cancer. The only inconclusive case is the patient who developed mesothelioma in the same hemithorax 18 months after the onset of the effusion. Nevertheless, thoracoscopic examination and pleural biopsy performed in the first month after the onset of effusion showed no evidence of the disease, highlighting the difficulty in diagnosing mesothelioma.
There are a great number of diseases that affect the pleural cavity. Patients with effusion whose cause remains uncertain after cytological examination and chest CT study, in whom neoplastic effusion is suspected, benefit from abdominal CT and gynaecological examination, or other complementary examinations as indicated by the clinical features, before using invasive investigations. In patients in whom a malignant process remains strongly suspected but unconfirmed after such examinations, thoracoscopy has a high diagnostic yield. Finally, patients whose pleural effusion remains undiagnosed despite these examinations, and whose medical history has no suggestion of an underlying neoplastic process, usually have a benign course. We advocate a conservative approach with clinical and radiological follow-up for these patients.
|
Acknowledgements |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The authors thank Maria Murias, nursing supervisor, and Susana Espinosa and Veronica Florez, attending nurse, for their invaluable participation, as well as Celine Cavallo for linguistic advice.
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
1. Bartter T, Santarelli R, Akers SM, Pratter MR. The evaluation of pleural effusion. Chest (1994) 106:1209–14.[ISI][Medline]
2. Rubins JB, Colice GL. Evaluating pleural effusions. Postgraduate Medicine (1999) 105:39–48.[Medline]
3. Ansari T, Idell D. Management of undiagnosed persistent pleural effusions. Clin Chest Med (1998) 19:407–17.[CrossRef][ISI][Medline]
4. Light RW. The undiagnosed pleural effusion. Clin Chest Med (2006) 27:309–19.[CrossRef][ISI][Medline]
5. Feinsilver SH, Barrows AA, Braman SS. Fiberoptic bronchoscopy and pleural effusion of unknown origin. Chest (1986) 90:516–19.[ISI][Medline]
6. Chang SH, Perng RP. The role of fiberoptic bronchoscopy in evaluating the causes of pleural effusions. Arch Inter Med (1989) 149:855–7.[Abstract]
7. Poe RH, Levy PC, Israel RH, Ortiz CR, Kallay MC. Use of fiberoptic bronchoscopy in the diagnosis of bronchogenic carcinoma. A study in patients with idiopathic pleural effusions. Chest (1994) 105:1663–7.[ISI][Medline]
8. Ferrer JS, Muñoz XG, Orriols RM, Light RW, Morell FB. Evolution of Idiopathic Pleural Effusion. Chest (1996) 109:1508–13.[ISI][Medline]
9. Light RW. Pleural effusion. N Engl J Med (2002) 346:1971–7.
10. Gunnels J. Perplexing pleural effusions. Chest (1978) 74:390–3.[ISI][Medline]
11. Antony V, Loddenkemper R, Astoul P, Boutin C, Goldstraw P, Hott J, Rodriguez Panadero F, Sahn SA. Management of malignant pleural effusions. Eur Respir J (2001) 18:402–19.
12. American Thoracic Society. Management of malignant pleural effusions. Am J Respir Crit Care Med (2000) 162:1987–2001.
13. Alemán C, Alegre J, Monasterio J, Segura RM, Armadans L, Angles A, Varela E, Ruiz E, Fernández de Sevilla T. Association between inflammatory mediators and the fibrinolysis system in infectious pleural effusions. Clin Sci (2003) 105:601–7.[CrossRef][ISI][Medline]
14. SPSS for Windows, release 6.0. (1993) Chicago IL: SPSS.
15. El Solh AA, Abdo T, Pineda L, Ramadan F, Berbary E. A longitudinal study of idiopathic exudative lymphocytic pleural efusión in older people. J Am Geriatr Soc (2005) 53:1957–60.[CrossRef][ISI][Medline]
16. Villena V, López A, Echave-Sustaeta J, Álvarez C, Martín P. Estudio prospectivo de 1000 pacientes consecutivos con derrame pleural. Etiología del derrame pleural y carácterísticas de los pacientes. Arch Bronconeumol (2002) 38:21–6.[ISI][Medline]
17. Porcel-Pérez JM. Manejo práctico del derrame pleural. An Med Interna (2002) 19:202–8.[Medline]
18. Storey D, Dines D, Coles D. Pleural effusion. A diagnostic dilemma. JAMA (1976) 236:2183–6.[Abstract]
19. Leslie W, Kinasewitz G. Clinical characteristics of the patient with nonspecific pleuritis. Chest (1988) 94:603–8.[ISI][Medline]
20. Boutin C, Astoul P, Seitz B. The role of thoracoscopy in the evaluation and management of pleural effusions. Lung (1990) 168(Suppl.):1113–21.[ISI][Medline]
21. Ocaña I, Martinez-Vazquez JM, Segura RM, Fernández de Sevilla T, Capdevila JA. Adenosine deaminase in pleural fluids: test for diagnosis of tuberculous pleural effusion. Chest (1983) 84:51–3.[ISI][Medline]
22. Valdes L, Alvárez D, San Jose E, Penela P, Valle JM, Garcia-Pazos JM, Suarez J, Pose A. Tuberculous pleurisy: a study of 254 patients. Chest (1998) 158:2017–21.
23. Lee YCG, Rogers JT, Rodriguez RM, Miller KD, Light RW. Adenosine deaminase levels in nontuberculous lymphocytic pleural effusions. Chest (2001) 120:356–61.[CrossRef][ISI][Medline]
24. Hiraki A, Aoe K, Eda R, Maeda T, Murakami T, Sugi K, Takeyama H. Comparison of six biological markers for the diagnosis of tuberculous pleuritis. Chest (2004) 125:987–9.[CrossRef][ISI][Medline]
25. Arrington CW, Hawkins JA, Richert JH, Hopeman AR. Management of undiagnosed pleural effusions in positive tuberculin reactors. Am Rev Respir Dis (1966) 93:587–92.[ISI][Medline]
26. Alcaide J, Altet MN, Canela-Solerc J, Pina JM, Milà C, de Souza ML, Jiménez A, Solsona J, Ventosa J, Salleras L. A study of the tuberculous infection in adults. Rev Clin Esp (2003) 203:321–8.[ISI][Medline]
27. Salgueiro M, Gonzalez J, Zamarron C, Pombo M, Ricoy J, Presedo MB, Calvo U, Pérez ML, Antúnez J, Durán JL, Rodríguez JR. Tuberculosis in Santiago de Compostela from 1999 to 2002. An epidemiological study. An Med Interna (2004) 21:215–22.[Medline]
28. Ribera E, Martinez-Vazquez JM, Ocana I, Ruiz I, Segura RM, Encabo G, Pascual C. Gamma interferon and adenosine deaminase in pleuritis. Med Clin (1990) 94:364–7.
29. Ferrer J, Roldan J, Teixidor J, Pallisa E, Gich I, Morell F. Predictors of pleural malignancy in patients with pleural effusion undergoing thoracoscopy. Chest (2005) 127:1017–22.[CrossRef][ISI][Medline]
30. Venekamp LN, Velkeniers B, Noppen M. Does ‘idioapathic pleuritis’ exist? Natural history of nonspecific pleuritis diagnosed after thoracoscopy. Respiration (2005) 72:74–8.[CrossRef][ISI][Medline]
31. Branch WT, McNeil BJ. Analysis of the differential diagnosis and assessment of pleuritic chest pain in young adults. Am J Med (1983) 75:671–9.[CrossRef][ISI][Medline]
32. Light RW. Pleural effusion due to pulmonary emboli. Curr Opin Pulm Med (2001) 7:198–201.[CrossRef][Medline]
33. Emad A, Rezaian GR. Diagnostic value of closed percutaneous pleural biopsy vs. pleuroscopy in suspected malignant pleural effusion or tuberculous pleurisy in a region with a high incidence of tuberculosis: a comparative, age-dependent study. Respir Med (1998) 92:488–92.[CrossRef][ISI][Medline]
34. Menzies R, Charbonneau M. Thoracoscopy for the diagnosis of pleural disease. Ann Intern Med (1991) 114:271–6.[ISI][Medline]
35. Hucker J, Bhatnagar NK, al-Jilaihawi AN, Forrester-Wood CP. Thoracoscopy in the diagnosis and management of recurrent pleural effusions. Ann Thorac Surg (1991) 52:1145–7.[Abstract]
36. Harris R, Kavuru M, Mehta A, Medendorp S, Wiedemann H, Kirby T, Rice T. The impact of thoracoscopy on the management of pleural disease. Chest (1995) 107:845–52.[ISI][Medline]
37. Celik M, Halezeroglu S, Senol C, Keles M, Yalcin Z, Urek S, Kiral H, Arman B. Video-assisted thoracoscopy surgery: experience with 341 cases. Eur J Cardiothorac Surg (1998) 14:113–16.[Medline]
38. Rodriguez Panadero F. Lung cancer and ipsilateral pleural effusion. Ann Oncol (1995) 6:S25–7.[Medline]
39. Chi DS, Abu Rustum NR, Sonoda Y, Chen SW, Flores RM, Downey R, Aghajanian C, Barakat RR. The benefit of video-assisted thoracoscopic surgery before planned abdominal exploration in patients with suspected advanced ovarian cancer and moderate to large pleural effusions. Gynecol Oncol (2004) 94:307–11.[CrossRef][ISI][Medline]
40. Gonzalez CA, Navarro C, Martinez C, et al. The European prospective investigation into cancer and nutrition (EPIC). Rev Esp Salud Publica (2004) 78:167–76.[Medline]
41. Riboli E, Kaaks R. The EPIC project: rationale and study design. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol (1997) 26(Suppl. 1):S6–14.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||