Diagnosis of Tuberculous Pleurisy Using the Biologic Parameters Adenosine Deaminase, Lysozyme, and Interferon Gamma

Diagnosis of Tuberculous Pleurisy Using the Biologic Parameters Adenosine Deaminase, Lysozyme, and Interferon Gamma* Luis \&Ides, M.D.; Esther San jose, M.D.; David Alvarez, M.D.; Alfredo Sarandeses, M.D.; Antonio Pose, M.D.; Benilda ChomOn, Ph.D.; jose Manuel Alvarn::.-Dobano, M.D.; Marcelino Salgueiro, Ph.D.; and jose Ram6n Rodriguez Suarez, Ph.D. We compared the parameters pleural adenosine deaminase (PADA, determined in 405 patients), the PADA/serum ADA ratio (P/SADA; !76 cases), pleural lysozyme (PLYS, !76 cases), the PLYS/serum LYS ratio (P/SLYS; !76 cases), and pleural interferon gamma (IFN, 145 cases) regarding their ability to difFerentiate tuberculous pleural effusions from others. The 405 pleural effusions were classi6ed by previously established criteria as tuberculous (91), neoplastic (110), parapneumonic (58), empyemas (10), transudates (88), or miscellaneous (48). The intermean difFerences between the tuberculous group and each of the others were statisticaJiy sipi&cant for aD 6ve parameters (p<0.01 for PLYS and P/SLYS with respect to the empyema group; p<0.001 otherwise), except for PADA and P/SADA with respect to the empyema group. AD the tuberculous pleurisy cases had PADA values of 47 UIL or more, as compared to only 5 percent of the other cases (sensitivity, 100 percent; speci6city, 95 percent). P/SADA was above 1.5 in 85.7 percent of tuberculous effusions and ll percent of the others (sensitivity, 85.7 percent; speci6city, 89 percent). PLYS, with a diagnostic threshold of 15 g!ml, had a sensitivity of 85.7 percent and a speci&city of 61.6 percent; P/SLYS, with a

T

he incidence of pleural tuberculosis is linked to the local prevalence of tuberculosis in general. In the United States, pleural tuberculosis accounts for 23.4 percent of extrapulmonary tuberculoses and 3.8 percent of all tuberculoses; 1 in Spain, the pleura is affected in 23.3 percent of all patients with tuberculosis.2 The reliability ofearly pleural tuberculosis diagnosis has in recent years been greatly improved by the use of biochemical markers such as adenosine deaminase (ADA),3-7 lysozyme,s- 11 and interferon gamma. 12 Adenosine deaminase is found in most cells in the organism, but its chief role concerns the proliferation and differentiation of lymphocytes, especially T lymphocytes; the concentration of ADA in these cells is

*From the Unidad de Neumolog(a, Hospital Provincial (Drs. Valdes, Alvarez, and Pose); Servicio ae Bioquimic.1, Hospital Provincial (Drs. San jose, Sarandeses, and Chom6n); and the Servicio de Neumolog{a, Hospital Gener.d de Galicia (Drs. Alvarez-Dobaiio, Salgueiru, and Suarez), Santio1go de ComposteLl, Spain. Manuscript received March 2; re• ision accepted June 16. Reprint requests: Dr. \aides, Urbanizacion lbrque Montouto, c/ Jo;milla lbrdo Ba:.an 26, 15885 TEO LiJ Coruna, Spain 458

threshold of 1.1, had a sensitivity of 67.3 percent and a speci6city of 90.3 percent; and IFN, with a threshold of 140 pg!ml, had a sensitivity of94.! percent and a speci&city of 91.8 percent. The lowest misdassification rate was achieved by PADA, with statisticaDy signi&cant difFerences (p<0.001) with respect to P/SADA, PLYS, and P/SLYS, but not with respect to IFN. The only signi6cant pairwise correlations among these parameters were between P/SLYS and PADA and between P/SLYS and P/SADA. We conclude that PADA and IFN are useful parameters for early diagnosis of tuberculous pleurisy, and that the other parameters considered have no advantages over PADA and IFN for this purpose (though the high specificity of P/SLYS may be noted). (Chat 1~ 103:4S8-65) ADA= ade-me deaminase; INF =pleural ftuid interferon pmma; NLR neptive lilcelihood ratio: NPV = aeptive predictive value; PA~_A =pleural ftuid adenosine deaminase; PLR =positive lilcelibood ratio; PLYS =pleural ftuid lysozyme; PPV=positive predictive value; P/SADA=pleural adenosine deaminase/serum adenosine deaminase ratio; P/SLYS =pleural lysozyme/serum lySOL)'Dle ratio; SADA =serum adenosine deamiDase; SLE =systemic lupus erythematosus; SLYS =serum ly.ozyme

=

inversely proportional to their degree of differentiation. Lysozyme is a bacteriolytic enzyme produced by polymorphonuclear cells and macrophages; in recent years, the pleural lysozyme/serum lysozyme ratio, P/ SLYS, has been reported to be highly sensitive and specific for tuberculous plemal effusions,s- 11 though these findings have also been questioned. 13 Interferon gamma is a lymphokine produced by activated T lymphocytes; it plays a fundamental role in the immune response to tuberculosis, and high levels in tuberculous pleural effusions have recently been attributed to in situ stimulation of T 4 lymphoc-ytes by tuberculosis autigens. 12 As far as we know, however, no previous study has simultaneously detern1ined the pleural concentrations of ADA (PADA), lysozyme (PLYS), and interferon gamma (INF) so as to determine which of these parameters may be most useful for the diagnosis of this type of pleurisy. We determined PADA, PLYS, IFN, serum ADA (SADA), and serum lysozyme (SLYS) in patients with pleural effusions of various origins, evaluated the utility of PADA, the PADA/serum AIJA ratio (P/ Diagnosis of Tuberculous Pleurisy (ValdN et al)

SADA), PLYS, P/SLYS, and IFN for diagnosis of tuberculous pleural effusion, and investigated pairwise correlations among these parameters.

Table 1-Etiology of Pleural E.Jfuaioru*

MATERIAL AND METHODS . ·.· . ~ considered 456 patients admitted to our hospitals because of pleural ell'usions. F~ne were excluded from this study because no single diagnosis was arrived at. The remaining 405 were classified in the following six diagnosis groups. (I) Tuben:ulous pleurisy. Ninety-one patients (56 male and 35 female) aged 32.3± 17.1 years were assigned to this group on the basis of pleural 8uid and/or biopsy cultures or on observation of caseous granulomas in pleural biopsy tissue. (2) Neoplastic effusions. One hundred ten patients (69 male and 41 female) aged 61.8± 15.9 years were fOund to have neoplastic tissue in the pleural cavity ('Jllble 1). (3) Parapneumonic effusions. Fifty-eight patients (42 male and 16 female) aged 62.3± 18.1 years were classified as parapneumouic cases on the basis of compatible clinical and radiologic findings in the absence of any signs of cardiac insuBiciency. (4) Empyemas. In ten patients (8 male and 2 female) aged 49.3± 18.5 years, empyema was diagnosed in accordance with its de&nition as parapneumonic ell'usion with positive culture results.•• (5) Transudates. This group was comprised of 88 patients aged 70.2± 11.5years(62 maleand26female). Fbr69patients, congestive cardiac insuBiclency was diagnosed on the basis of clinical and radiologic &ndings, the disappearance of the effusion on suitable treatment, and the absence of pulmonary infiltrates or malignancy. Other causes of transudates are listed in 'lhble 1. (6) Miscellaneous. Fbrty-eight patients aged 57.3± 18.4 years (31 male and 17 female) had pleural effusions associated with pulmonary thromboembolism (detected by arteriography or perfusion pulmonary gammagraphy; 15 21 cases), postsurgical effects (9 cases), or other miscellaneous causes ('Jllble 1). Pleural and serum lysozyme and serum ADA activity were measured in 276 of the 405 cases: 49 patients with tuberculous pleurisy(aged 30.6± 17.7 years); 74 patients with neoplastic disease (aged 64.6 ± 13.9 years); 37 patients with parapneumonic effusions (aged 65.6± 17.7 years); 10 patients with empyemas (aged 49.3 ± 18.5 years); 70 patients with transudates (aged 70.7 ± 10.8 years); and 36 patients with miscellaneous disorders (aged 58.6 ± 17.5 years) ('Jllble 1~ Interferon gamma in pleural ftuid (IFN) was determined in 145 cases: 35 patients with tuberculous pleurisy (aged 30.5±18.9 years); 36 patients with neoplastic disease (aged 63.8± 13.5 years); 21 patients with parapneumonic effusions (aged 65.3± 18.0 years); 8 patients with empyemas (aged 51.0±20.5 years); 29 patients with transudates (aged 74.0±9.9 years); and 16 patients with miscellaneous disorders (aged 57.1 ± 17.1 years) ('Thble

1). Pleural ftuid and blood samples were taken at the same time from Casting patients. Both samples were centrifuged for 10 min at 3,000 rpm, and the supernatants were stored at - 40"C pending assay. Pleural biopsies were performed with a Cope needle,•• except when a transudate was suspected. Adenosine deaminase activity (UIL at 37"C) was determined colorimetrically by the method of Giusti; 17 the NH • 4 released by deamination of adenosine added to the samples was quantified by incubation with phenol nitroprusside in an alkaline medium, fOllowed by measurement of absorbance at 628 nm. Lysozyme was determined by radial immunodiffusion using a kit Samples were incubated fOr 6 h at 37"C (Quantiplate kit, Kallestad~ on agarose plates loaded with a suspension of Micrococcus lysodeikUcus. A calibration curve (lysis plaque diameter versus log lysozyme concentration)• was constructed using standards of known concentration. Values are given below in grams per milliliter. Pleural fluid interferon gamma was determined using a kit (Intertest-gamma Human Interferon-Gamma Kit, Genzyme), a

PADA, No. Tuben:ulous Neoplastic Lung Lymphoma

91

no

Ovary

Breast Stomach Uterus Colon Mesothelioma Chronic lymphatic leukemia Liver Kaposi's sarcoma Myeloma Neuroblastoma Kidney Pancreas Prostate Thymoma Seminoma Esophagus Undetermined Parapneumonics Empyema Miscellaneous Pulmonary thromboembolism Postsurgery Pancreatitis Systemic lupus erythematosus Hemothorax Chylotorax Benign asbestosis Hypothyroidism Yellow nail syndrome Transudates Heart failure Nephrotic syndrome Liver cirrhosis Hypoalbuminemia Superior vena cava syndrome Massive atelectasis Total

29 12 983 93 63 5 4 43 333 322 2 22 21

PADA P/SADA PLYS P/SLYS, No.

PADA PIS ADA PLYS P/SLYS INF, No.

49 74 17 10 98

35 36

2

4 1

2

1 1 1 1 1 1 14 58 10 10 48 21 99 541 44 3

3

11 14 37 36 13

3 2 31

11 11 11

88 69 851 43 33 33 111 405

3 21 8 16 6

70 55

1 29 25 2

276

145

*See text for explanation of abbreviations. solid-phase enzyme-immunoassay employing the multiple antibody sandwich principle. First, a microtiter plate coated with a monoclonal antibody specific for human interferon gamma is used to capture interferon gamma present in the samples. Then, a polyclonal antibody that binds to multiple epitopes on the Interferon gamma is added. Next, an amplifying antibody, biotin-labeled anti-immunoglobulin is applied, followed by streptavidin-peroxidase that binds to the biotin. Finally, o-phenylendiamine Is added as substrate for the peroxidase, and absorbance at 492 nm is read; the increase in this parameter is proportional to the concentration of interferon gamma in the sample. The results are expressed in picograms per milliliter (I P~Vml=2.5X IO-• U/ml).

Statistical Analysis The prevalence of tuberculous pleurisy among these patients was CHEST I 100 I 2 I FEBRUARY, 1993

-

Table 2-Mean Voluea (X± SD) of the lbrameten Studied in the DijJermt 1flpea ofEJfuaiona•

PADA P/SADA PLYS P/SLYS IFN

Tuberculous

Neoplastic

Parapneumonics

Empyema

107.5±37.9 2.6±1.1 26.6±14.4 1.2±0.3 440.6±375.2

24.8±22.3t 0.8±0.4t 16.6±10.9t 0.7±0.2t 94±30.1t

22.8±11.1t 0.9±0.6t 18.6±10.1t 0.8±0.3t 97.5±42.1t

159.7 ± 158.8:j: 8.1±9.4:j: 169.9±204.5§ 6.7±7.5§ 104.1±35t

Miscellaneous

Transudates

17.7±7.7t 11±7.4t 0.7±0.3t 0.3±0.2t 14.3±7.1t 11.9±5.9t 0.7±0.1t 0.6±0.1t 100.7 ± 43. 7t 70.8±21.3t

*See text for explanation of abbreviations. tp<0.001 with respect to the tuberculous group. :j:Not significant with respect to the tuberculous group. §p<0.01 with respect to the tuberculous group. calculated by dividing the number of tuberculous effusions by the total number of effusions. Distributions were deemed normal or not normal on the basis of their skew and kurtosis. The statistical significance of differences between means was estimated using Student's t test for normal distributions and Wilcoxon's ranked sums test otherwise. Correlations were quantified using Pearson's correlation coefficient after normalizalion of any nonnormally distributed variables by log or extended log transformation. The diagnostic value of the parameters studied was assessed in terms of sensitivity (defined as TPI[TP + FN]), specificity (TN[TN + FP]), positive predictive value PPV (TPI[TP + FP]), negative predic-

...

ADA Ull•

tive value NPV (TNI[TN + FN]), efficiency ([TP + TN]I[1btal No. of cases tested]), positive likelihood ratio PLR (sensitivity/ [1- specificity]), and negative likelihood ratio NLR ((1- sensitivity]I specificity), where TN indicates the number of true negative diagnoses; TP, bue positives; FN, false negatives; and FP, false positives. The positive and negative predictive values of PADA were calculated for various hypothetical values of the prevalence of tuberculous pleurisy in a given population, as follows: PPV = (Prevalence x Sensitivity)I(Prevalence x Sensitivity) + [(1 - Prevalence)(1 - Specificity)] NPV = [(1 - Prevalence) x Specificity]/[(1 - Prevalence) x IFN...,.,

700

500

300

70

*

II • II

*

~

* *

•• a•

47

200

ir_

140

100

30 50 10

* TuberculOus

480

Neoplasllcs

Parapneumonocs

£:~a

Mtsoellaneuub

Transudates

FIGURE 1. Adenosine deaminase (ADA, UIL) and Interferon gamma (IFN, pglml) levels in the pleural Huid of the various groups of patients studied. Diagnosis of lllben:ulous Pleurisy (\Wd88 et el)

'*' P/SLYS

FIISADA'•

5

! 4

I

• I.

3

~•

2.1

I

** *

•

t

*

*

* * ** *

*

2.5 2.3 2.1 1.9 1. 7 1.5 1.3 1.1

• I•• • I ••• • •

***

•

•• • ••

0.7 0.5

•

0.3

2. Pleural ADA/serum ADA ratio (P/SADA) and pleural lysozyme/serum lysozyme ratio (P/ 0.1 SLYS) levels in the various groups of patients studied.

** •

FIGURE

Tubefculous

Specificity] + [Prevalence X (1 - Sensitivity)] The diagnostic thresholds for each diagnostic parameter considered were defined as the values affording greatest diagnostic efficiency, which were determined using a computer program (GW BASIC)••; sensitivity, specificity, PPY, and NPV were likewise calculated with this program.

REsuLTS Table 2 lists the means and standard deviations of PADA, P/SADA, PLYS, P/SLYS, and IFN for each group of patients studied. The intermean differences between the tuberculous group and each of the others are statistically significant for all five parameters (p
*

• • • •

I

* * •

2

t ••

**

•

* •

!fi

• ••• -··-- . **

** ** ••

1.8

~.

••

It... ••• •• •

Neopldsto"s

*t 0.11

I

.

* •

1.1

--

- f. ••

** **t

*

I **** • : • * I * * * * •• •

PardllneurTo.Jilk.S

Empyema

•

I

~

••

••

Moscelilneous

Transudates

cases had PADA values below this level. In 94.2 percent of tuberculous pleurisy cases (33135), IFN was above 140 pg/ml, whereas 91.6 percent of cases of neoplastic disease (33136), 85.7 percent of parapneumonic effusions (18121), 81.2 percent of miscellaneous cases (13/16), and all the empyema and transudate cases had IFN values below this threshold. In 85.7 percent of tuberculous effusions (42/49), PLYS was above a threshold of 15 g/ml. However, only 56.7 percent of patients with neoplastic disease (42/ 74), 48.6 percent of parapneumonic cases (18137), 10 percent of empyema cases (1110), 77.1 percent of transudate cases (54170), and 69.4 percent of miscellaneous cases (25136) had PLYS values below this threshold (data not shown). Figure 2 shows the values of P/SADA and P/SLYS for each group of patients studied. In 85.7 percent of tuberculous cases (42149), P/SADA was 1.5 or more; in 87.8 percent of neoplastic cases (65174), 83.7 percent of parapneumonic effusions (31137), 20 percent of empyema cases (2110), 98.5 percent of transudate cases (69f70), and 97.2 percent of miscellaneous cases (35/36), P/SADA was below this level. For a P/SLYS CHEST 1103 I 2 I FEBRUARY, 1993

481

Table 3-Number of Miaclauijied Pleural Effuaiona in Each Group for Eoery lbrameter Studied*

Tuberculous Neoplastic Parapneumonics Empyema Miscellaneous Transudates Total

PADA

P/SADA

PLYS

P/SLYS

IFN

MH UV110

7/49 Q/74 6137 8110 lfJ6 1170 32/276 (11.5%)

7/49 32174 19137 9110 1l!J6 16170 941276 (34%) NS

16/49 5174 5137 lWlO lfJ6 1170 381276 (13.7%)

2135 3136 3121

1158 5.'10 W48

MIS 16/405 (3.9%)

(W

3/16 0129 111145 (7.5%)

p
•See text for explanation of abbreviations. NS =not significant.

threshold of 1.1, 67.3 percent of tuberculous effusions were positive (33149}, while 93.2 percent of cases of neoplastic disease (69174), 86.4 percent of parapneumonic effusions (32137}, 98.5 percent of transudate cases (69170}, and 97.2 percent of the miscellaneous cases (35136} were negative; however, 100 percent of the empyema cases had P/SLYS ratios above 1.1. Table 3 shows that the parameter with fewest misclassifications (3.9 percent) was PADA, with statistically significant differences from P/SADA, PLYS, and P/SLYS (p<0.001 in each case} but not from IFN. The etiologies of the nontuberculous effusions with high PADA, P/SLYS, or IFN are listed in Thble 4. Thble 5lists performance figures for each parameter. The prevalence of tuberculous pleurisy among our patients was 22.4 percent. The PADA had sensitivity of 100 percent, a specificity of94.9 percent, a positive predictive value (PPV} of 85 percent, a negative predictive value (NPV) of 100 percent, an efficiency of 96 percent, a positive likelihood ratio (PLR) of 20, and a negative likelihood ratio (NLR} of zero; all these indices are better than those of any other single parameter considered. In only one nontuberculous effusion (a case of chronic lymphatic leukemia} were the diagnostic thresholds for both PADA and INF simultaneously exceeded. The sensitivity of this diagnostic criterion was 94 percent (two tuberculous effusions had INF levels below the threshold}, and its specificity was 99 percent. Thble 6 lists the PPV and NPV values obtained as functions of the prevalence of tuberculous pleurisy in any given population. The only significant pairwise correlations among the parameters studied in the tuberculous pleurisy group were between P/SLYS and PADA and between P/SLYS and P/SADA (p
DISCUSSION

Mycobacterium tuberculosis invades the pleural cavity chiefty through rupture of subpleural caseous lesions. Bacillus protein antigens appear to induce a delayed hypersensitivity reaction that stimulates lymphocytes, which in tum release certain lymphokines that (1) activate macrophages against the mycobacterium, and (2} alter the permeability of pleural vessels and inftuence the formation of granulomas. Adenosine deaminase is considered to be an indicator ofcell-mediated immunity. 3 Since T lymphocytes play an important role in the immune response to tuberculosis, m possible correlations have therefore been sought between the high ADA levels in tuberculous pleural effusions and either the total number ofT lymphocytes• or the various lymphocyte subpopulations;21 however, no statistically significant results have been forthcoming. The ADA activity is contributed to by two main isoenzymes, ADA-1 and ADA-2, where ADA-2 is found only in macrophages, which release it when stimulated by the presence of live microorganisms within them, thus increasing the ADA content of body 8uids. 22 Since the initial proposal of Piras et al, 3 many studies have confirmed the utility of ADA for diagnosis of tuberculous pleurisy.+7 In our work too, ADA levels were significantly higher in tuberculous pleural effusions than in all other kinds except empyemas, which are easily distinguished, and our findings are similar to those of other studies23 regarding the sensitivity, specificity, PPY, and NPV of pleural ADA for differentiation of tuberculous pleurisy from other pleural effusions. Our PLR shows that the probability that tuberculous pleurisy gives rise to a pleural ADA Table 4- Etiology of Nontuberculou. Pleural E.ffuaiona With High PADA, PISLYS or IFN* Etiology Empyema Parapneumonics Neoplastics Lymphoma Lung Chronic lymphatic leukemia

Ovary Uterus Mesothelioma Thymoma Neuroblastoma Undetermined Pulmonary thromboembolism Postsurgery Heart failure

PADA, No.(%)

P/SLYS, No.(%)

(50) 1158 (1.7)

5.'10

(9) lWllO 2/12 (16.6) 2129 (6.9) 113 (33.3)

IWIO (100) 5137 (13.5) 5174 (6.7)

119 (11.1) 115 (20) 114 (25) Ill (100)

3117 (17.6) 113 (33.3)

IFN, No.(%) 3121 (14.2) 3136 (8.3) 118 (12.5) 113 (33.3)

118 (12.5)

Ill (100) 1114

(7)

•See text fur explanation of abbreviations.

316

119 1170

(11) (1.4)

(50)

Table 5-Senntivity, Specifo;ily, Poaitive Predictive \blue (PPV), Negtllive Predictive \blue (NPV), Efficiency, Positive Likelihood &tio (Pl.B), and Negative Likelihood llalio (NI..B) ofEvery lbrtmwtt~r Studied for DiagnoliB of Tuberculoua Pleun.y•

Test

Sensitivity, %

PADA P/SADA PLYS P/SLYS IFN

100.0 85.7 85.7 67.3 94.2

Spedficity, 4 ' · ···PPv.% 94.9 88.9 61.6 90.3 91.8

85.0 62.6 ~.5

60.0 78.5

NPY,·IJJ

Efficiency, 111

100.0 96.6 95.2 92.7 98.0

96.0 88.4 65.9 86.2 92.4

PLR

NLR

20

0 15.7 22.5 36.6 6.5

7.7 2.1 6.7 11.4

*See text for explanation of abbreviations.

activity of over 47 UIL is 20 times greater than the probability that any other pleural effusion produces such a level of ADA activity. Our NLR shows that a PADA level below this threshold rules out tuberculous pleurisy. The false-positive diagnoses given by PADA with this threshold were 5110 empyemas (50 percent), 2/12lymphomas (16.6 percent), 8198 neoplastic cases (8.1 pen:ent), and 1/58 parapneumonic processes (1.7 percent). In empyema cases, high PADA is due to large numbers of polymorphonuclear cells, and in lymphoma cases, it is due to the proliferation of relatively undifferentiated lymphocytes. Diagnostic specificity is increased by using PADA in conjunction with the 2-deoxyadenosine/ADA ratio, which is below 0.45 for tuberculous pleurisy and above this level for empyema and neoplastic disease. 24 Since sensitivity and specificity depend on the prevalence of the target condition in the target population, our results for these diagnostic quality parameters, and for the positive and negative predictive values of our diagnostic parameters, can only be generalized to regions in which the prevalence of tuberculous pleurisy among pleural effusion patients is similar to its value in our hospitals (PLR and NRL, however, are unaffected by prevalence and therefore, in principle, universally generalizable). As the prevalence of tuberculous effusions declines, the positive predictive value of PADA does too (Thble 6). 23 ·l111 It is clear, however, that for regions in which the prevalence of tuberculous pleurisy is at least as high as in ours, measurement of PADA is a simple test of great Table 6-lnjiUflftCflof the Prevalence of Tuberculou. Pleun.y Among lbtients With Pleural E.lfuaion on the Po.itive and Negative Predictive Value. of PADAfor DiagnoliB of Thia Conditima* Prevalence,% 5 10 15 20 25

Positive Predictive Value,%

Negative Predictive Value,%

51

100 100 100 100 100 100 100

69 78

83 87

30

89

35

91

*Sensitivity= 100 perL-ent; specificity= 95 percent.

diagnostic value. Little attention has been paid to the P/SADA ratio for diagnosis of tuberculous pleural effusions. Pettersson et al5 reported that a value above 2.5 always indicates tuberculosis, empyema, or rheumatoid arthritis. Moriwaki et al11 reported that the sensitivity of this parameter was 100 percent for a cut-off of 1.1 (though specificity was only 47 percent). For our data, a threshold of 1.5 gives a sensitivity of 95.9 percent with a specificity of81.4 percent, and there was little chance that a value below 1.5 could belong to a patient with tuberculous pleural effusions (NPV 96.6 percent). Lysozyme is present in the epithelioid cells of granulomas, in activated macrophages, and in granulocytes,8 and in spite of also being found in certain tumour cells, 10 has therefore been used for diagnosis of tuberculous pleural effusions.s-u Results have differed, however: Asseo et aJ9 found a sensitivity of only 66 percent, instead of the 100 percent reported by Moriwaki et al." For our patients, PLYS performed . poorly with a cut-off of 15 glml {sensitivity, 85.7 percent; specificity, 61.6 percent), though it could be used fairly confidently for ruling out tuberculosis (NPV 95.2 percent). The P/SLYS ratio performed rather better than PLYS. Biologically, a P/SLYS ratio greater than unity suggests the production of lysozyme in the pleural cavity. Asseo et al9 stress that all their tuberculous pleurisy patients satisfied this criterion (sensitivity, 100 percent; specificity, 93.5 percent); subsequently, Verea et al10 found a sensitivity of 100 percent and a specificity of 94 percent with a cut-off of 1.2 (though empyema cases were excluded from consideration), and Moriwaki et al" found a sensitivity of 100 percent Table 7 -Correlation Coej]identa Between the .ftarametera Studied in Tuberculou. Pleural E.IJ'U8ioru*

IFN PLYS P/SLYS

PADA

P/SADA

PLYS

P/SLYS

r:0.250 p:NS r:0.253 p:NS r:0.469 p<0.01

r:0.186 p:NS r:0.166 p:NS r:0.461 p
r:0.094 p:NS

r:0.261 p:NS

*See text for explanation of abbreviations. NS =not significant. CHEST 1103 I 2 I FEBRUARY. 1993

483

and a specificity of 87.5 percent. For our patients, the specificity of this parameter, 90.3 percent, was similar to that found in these earlier studies, but its sensitivity was very low, 67.3 percent. This may be due to the method used to determine lysozyme, since it has been reported that for patients with high P/SIXS ratios, but not for those with low P/SIXS ratios, the turbidimetric method gives higher values than the lysoplate method.17 The false positives we found for P/SLYS comprised all the empyema cases (10110), &ve parapneumonic effusions {13.5 percent), &ve neoplastic cases {6. 7 percent), a postsurgical effusion, and a case of congestive cardiac insufficiency (Thble 4). We were unable to study any pleural effusions secondary to rheumatoid arthritis or sarcoidosis, both of them situations that have produced high P/SLYS ratios in other studies.8 •9 .ss Some authors have reported high P/ SIXS in systemic lupus erythematosus (SLE~ but our four cases, like the two reported by Fonbin et al, 7 had P/SLYS values below the diagnostic threshold. In vitro treabnent of T4 lymphocytes from tuberculous pleural effusions with PPD stimulates production of interferon gamma, 211•30 the lymphokine that most increases the mycobactericidal activity of macrophages. A wide range ofhigh concentrations ofiFN has in recent years been found in tuberculous pleurisy cases, and IFN has been reported to have both 100 percent sensitivity and 100 percent speci&city for tuberculous pleurisy. 12.3l In these studies, this parameter was therefore more speci&c than PADA, chie8y due to the low levels of IFN in lymphoma and empyema patients with high PADA. Our results differ from these studies regarding both the sensitivity and specificity of IFN (94.2 percent and 91.8 percent, respectively), two of our tuberculous pleurisy patients having law IFN, and high IFN being found in nine nontuberculous patients (three parapneumonic cases, three cases.of pulmonary thromboembolism, one of lymphoma, one of chronic lymphatic leukemia, and one of neuroblastoma; see 'Thble 4). In the two tuber. culous pleurisy cases with low IFN (who had PADA values of 106 and 160 U/L and P/SLYS ratios of 1.82 and 1.25, respectively), it would seem that lymphokine production by sensitized T4 lymphocytes must have been low. In both cases, the amount of pleural effusion was rather small, and it has been reported previously that the mean IFN level for small effusions is lower than for large effusions (though in the study in questionll- the differences were not statistically significant and all the "low" values were above the diagnostic threshold used). Regarding false positives, high interferon gamma levels in nontuberculous patients have hitherto been found only in the pleural 8uid of a case of carcinomatous pleurisy,32 in the blood serum of SLE patientSD (though our two cases of this disease had low IF~ and in a few cases of infection by intracel-

lular pathogens. 34 The latter cause may possibly account for the high IFN of our three false positives with parapneumonic effusions; we have no idea why IFN was high in the other six, in all of whom pleural 8uid cells were over 80 percent lymphocytes and (except in the chronic lymphatic leukemia case, with PADA of74 UIL) PADA and P/SLYS were low. The misclassi&cation rate with IFN (111145, 7.5 percent) was greater than with PADA (161405, 3.9 percent), though the difference is not statistically significant. As in other studies, no signi&cant correlation was found between IFN and PADA,31 .31 nor between IFN and PIXS (Thble 7). There was slight but signi&cant correlation between P/SLYS and PADA (r= 0.469) and between P/SLYS and P/SADA (r = 0.461). Unlike some other authors,5 •11 we found no correlation between PADA and PIXS. It would seem that the rises in PADA, IFN, and PLYS in tuberculous pleurisy occur at different stages of the abnormality, being due respectively to infected macrophages, the release of lymphokines by sensitized T4 lymphocytes, and activated granulomas. We conclude that PADA and IFN are both useful for early diagnosis of tuberculous pleural effusions (note that determining PADA is much less expensive). More research is called for in order to clarify the cause or causes of high IFN in nontuberculous pleural effusions. None of the other parameters considered appears to offer any advantage over IFN or PADA, though the high speci&city of P/SLYS may be borne in mind. REFERENCES

1 W!g JG. Clinical forms of mycobacterial diseases. In: Fishman AP, eel. Pulmooary diseases and disorders. 2nd eel. New York: McGraw-Hill Book Co, 1988; 1843-62 2 Vidal R, de Gracia J, Ruiz J, Fite E, Mons6 E, Martin N. Estudio controlado de 637 pacientes con tuberculosis: dian6stico y resultados terapeuticos con esquemas de 9 y 6 meses. Meel Clio (Bare) 1986:87:368-70 3 Piras MA, Gakis C, Budroni M, Andreoni G. Adenosine deaminase activity in pleural effiuions: an aid to difFerential diagnosis. BMJ 1978; 2:1751-52 4 Ocana I, Martfnez-Vazquez JM, Segura RM, Fernandez de Sevilla T, Capdevila JA. Adenosine deaminase in pleural Ooids: test for diagnosis of tuberculous pleural effiuion. Chest 1983; 84:51-3 5 Pettersson T, Ojala K, Weber TH. Adenosine deaminase in the diagnosis of pleural effiuion. Acta Meel Scand 1984; 215:299304 6 Valdes L, Masa L, Alvarez-Calder6n P, Rodriguez-Segade S, Rabufial R, Ventosa JJ, et al. Detenninaci6n de adenosina desaminasa (ADA) en liquido pleural [abstract]. II Reuni6n Ordinaria de Ia Sociedad Gallega de Medicina lnterna, La Coruiia, 1985 7 Fbnbin J, ~ H, Perez J, Dominguez L, Martin MT, Montero MC. Diagnostic value of simultaneous determination of pleural adenosine deaminase and pleural lysozyme/serum lysozyme ratio in pleural efiUsions. Chest 1988: 93:303-07 8 Kloclcars M, Pettersson T, Riska H, Hellstrom P, Norbagen A. Pleural ftuid lysozyme in human disease. Arch Intern Meel

1979; 139:73-7 9 Asseo P, Traoopoulos G, Kotsovoulou-Fbusbki V. Lysozyme (muramidase) in pleural eiiUsions and serum. Am J Clin Pathol 1982; 78:763-67 10 Verea H, Masa J, Domfnguez L, Pmez J, Martm -MT, Fbnt4n J. Meaning and diagnostic value of determining the ly-rme lewl of pleural ftuid. Chest 1987; 91:342-45 11 Moriwaki Y, Kohjiro N, Itoh M, Nalallsuji Y, Okada M, Ishihara H, et al. Discrimination of tubereulous from carcinomatous pleural effusion by biochemical marJcen: adenosine deaminase, lysozyme, llbronectin and carcinoembryonic antigen. Jpn J Med 1989; 28:478-84 12 Ribera E, Ocaiia I, Martmez-Vazquez JM, RosseR M, Espaiiol T, Ruibal A. High level of interferon gamma in tubereulous pleural effusion. Chest 1988; 93:308-11 13 Light Rw. Clinical manifestations and useful test. In: Light Rw, ed. Pleural diseases. Philadelphia: Lea &: Febiger, 1983; 33-00 14 Light Rw. Parapneumonic elfusions and infections of the pleural space. In: Pleural diseases. Philadelphia: Lea &: Febiger; 1983; 101-18 15 McNeill BJ. Ventilation-perfusion studies and the diagnosis of pulmonary embolism: concise communication. J Nucl Med 1980; 21:319-23 16 Cope C. New pleural biopsy needle. JAMA 1958; 167:1107-08 17 Guisti G. Adenosine deaminase. In: Bergmeyer HU, ed. Methods of enzymate analysis. New York: Academic Press, 1974;

23

24

25

26

27 28

29

30

1092-99 18 Osserman EF, Lawlor DP. Serum and urinary lysozyme (muramidase) in monocytic and monomyelocytic leulcemia. J Exp Med 1966; 124:921-51 19 Sarandeses A, San Jose E, Chom6n 8, Areses J. C4lculo por ordenador del valor critico discriminante [abstract]. XXXVIII Congreso Nacional de Ia AEBC, Oviedo, Mayo, 1991 20 Maclcaness GB. The immunology of antitubereulous immunity. Am Rev Respir Dis 1968; 97:337-44 21 Ocaiia I, Ribera E, Ruiz I, Bejarano E, Fernandez de Sevilla T, Espaiiol T, et al. Subpoblaciones linfucitarias en los derrames pleurales: relaci6n con Ia actividad de ADA. An Med Intern (Madrid) 1988; 5:274-78 22 Galds C, Calia G, NaitanaA, Pirino D, Seru G. Serum adenosine

31

32

33

34

deaminase activity in HIV positive subjects: a hypothesis on the signi6caDce of ADA-2. Panminerva Med 1989; 31:107-13 Eoa J, \Wis V, Pmez de Oteyza C, Enriquez de Salamanca R. Utilidad y limitaciooes de Ia adenosina desaminasa en el diagia6Stico de Ia pleuresfa tubereulosa: estudio metaanalftico. Med Clin (Bare) 1990; 95-.333-35 Galds C, Calia GM, Naitana ACV, Ortu AR, Contu A. Serum and pleural adenosine deaminase activity: correct interpretation of the findings. Chest 1991; 99:1555-56 Maritz FJ, Malan C, Le 11om I. Adenosine deaminase estimations in the difFerentiation of pleural elfusions. S Afr J Med 1982; 62:55EHS8 Stranldnga WFM, Nauta JJ, Straub JP, Stam J. Adenosine deaminase activity in tubereulous pleural effusions: a diagnostic test. Tubercle 1987; 68:137-40 Pettenson T, IOoclcars M, Hellstr6m PE, Frfiseth B. Lysozyme in pleural elfusions. Chest 1988; 92:220-21 Verea H. AniUisis bioqurmico dellfquido pleural. Arch Bronconeumol1991; 27:115-20 Shimokata K, Kishimoto H, 1ilbgi E, Tsunebwa H. Determination ofT-ceR subset producing gamma-interferon in tuberculous pleural eiiUsion. Microbiol Immunol1986; 30:353-61 Ribera E, Espaiiol T, Martmez-V6zquez JM, Ocaiia I, Encabo G. Lymphocyte proliferation and gamma-interferon production after 'm vitro' stimulation with PPD: difFerences between tubereulous and nontubereulous pleurisy in patients with positive tubereulin skin test. Chest 1990; 97:1381-85 Ribera E, Martmez-V6zquez JM, Ocaiia I, Ruiz I, Segura RM, Encabo G, et al. Interferon-gamma y adenosina desaminasa en las pleuritis. Med Clin (Bare) 1990; 94:364-67 Shimokata K, Sab H, Murate T, Hasepwa Y, Hasepwa T. Cytokine content in pleural eiiUsion: comparison between tubereulous and carcinomatous pleurisy. Chest 1991; 99:110307 Hooks JJ, Moutsopoulos HM, Geis SA, Stahl NI, Declcer JL, Notkins AL. Immune interferon in the cireulation of patients with autoimmune disease. N Eogl J Med 1979; 301:5-8 Ennis FA,IIeager A. Immune interferon produced to high lewis by antigenic stimulation of human lymphocytes with inftuenza virus. J Exp Med 1981; 154:1279-84

Q£ST I 109 I 2 I FEBRUAR"( 1983

415