INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Different studies have shown that cytokine concentrations (especially the proinflammatory modulators), measured in various biological fluids (serum, blister fluids, bioptic supernatants) (1), may be correlated with the extent of various diseases including psoriasis (1, 2), bullous dermatoses (5, 6), acquired immune deficiency syndrome (7, 8), multiple myeloma (9), and septic shock (10). This finding is not surprising because the cytokines regulate interaction between cells and, therefore, their concentrations are important for determining disease evolution. The selection of the cytokine studied was based on different aspects: three modulators represent the main T-helper type 1 (Th1) cytokines, namely, interleukin-2 (IL-2), IL-12, and interferon gamma (IFN-) (11); two represent the main T-helper type 2 (Th2) cytokines, namely IL-4 and IL-10; the remaining four represent the main proinflammatory mediators, namely, IL-1, IL-6, IL-8, and tumor necrosis factor-alpha (TNF-). In addition, previously published studies report that at least TNF-, IL-2, and IFN- exert antimycobacterial activities (14). Because cytokines are not specific for the different diseases, it is expected that different cytokines may be increased in various pathologies. In this study, a panel of nine cytokines was evaluated in bronchoalveolar lavage fluids (BALF) of 15 patients with pulmonary tuberculosis. BALFs were collected from both lungs, one lung designated more affected and the other less affected, on the basis of high-resolution computed tomography (HRCT) analysis. Thoracic HRCT represents an excellent technique for defining pulmonary tuberculous lesions (17). In fact, it permits reliable diagnostic interpretations based on the lesion types, disease activity, and disease extent that the conventional methods cannot explore satisfactorily. HRCT may detect tuberculous lesions characterized by a 0.2 mm size (18). Micro- and macronodules may also be well studied, as can cavities, exactly defining the wall thickness of those with even minimal dimensions. Given the reliability of the HRCT score, as shown in this study and in others for different diseases (19), a three-step methodological approach was followed: the first one was based on the comparison between the cytokine concentrations as well as the HRCT parameters observed in the more affected versus the less affected lung; the second was based on the analysis of the correlations between the HRCT abnormalities and the individual cytokine concentrations; and, finally, the third point was represented by the analysis of the reciprocal correlations calculated among the different cytokines and the different HRCT scores generated for various characteristic images.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Fifteen human immunodeficiency virus (HIV)-negative, newly diagnosed active pulmonary tuberculosis patients were enrolled in this study. Clinical presentation and chest radiographs were compatible with pulmonary tuberculosis and sputum was positive for acid-fast bacilli. All of them had positive tuberculous skin testing and no evidence of other concurrent diseases. Mycobacterium tuberculosis positivity in sputa was confirmed both by Bactec (Bactec 460 TB; Becton Dickinson, Milan, Italy) and conventional sputum culture. Demographic characteristics were as follows: median age 32, range 18 to 52; six males and nine females; four smokers. None of the patients were previously treated with antituberculous therapy. In addition, no corticosteroids or immunosuppressive agents had been administered. An accurate clinical examination and evaluation of vital signs, fever, and a complete panel of biochemical and hematological variables were registered. The study was approved by the hospital ethics committee, and all subjects gave an informed consent.

HRCT and Arbitrary Score Generation

All patients underwent chest HRCT analysis in the supine position. HRCT studies were performed on an A-TOM XR 6000 (Ansaldo Elettronica Biomedicale, Genoa, Italy). The parameters used for evaluation were a 1.5-mm section thickness at 10-mm intervals with a 512 × 512 reconstruction matrix, 130 kV and 175 mA, and a high-spatial-frequency algorithm. All images were obtained at window settings appropriate for lung parenchyma (level, 400 to 800 Hounsfield units [HU]; width, 1,800 to 1,600 HU). HRCT scans were evaluated for the presence, distribution, and extent of the following signs: (1) miliary nodules (1- to 2-mm nodules involving both the intralobular interstitium and interlobular septa); (2) nodules (< 10-mm nodules related to the terminal or respiratory bronchioles and separated from the pleural surface or interlobular septa by more than 2 mm); (3) consolidation (panlobular and polylobular consolidation); (4) ground glass; (5) cavity; (6) bronchial lesion (bronchial wall thickening, bronchiectasis); (7) fibrotic bands. Extent of involvement was assessed independently for each of the three lung zones. On the HRCT scans, the lungs were divided into six zones (upper, middle, and lower zones of the right and left lung). These areas of the lung were defined as the "upper zones" above the level of the carina; the "middle zones" between the level of the carina and the level of the inferior pulmonary veins; and the "lower zones" below the level of the inferior pulmonary veins. The HRCT score was determined by visually estimating the extent of disease in each zone. The HRCT scans were analyzed by two independent chest radiologists and final conclusions on the findings were reached by means of consensus. The arbitrary score was based on the percentage of lung parenchyma that showed evidence of each recorded abnormality: (1) involvement of less than 25% of the image; (2) 25 to 50%; (3) 50 to 75%; (4) more than 75%. A total score for each area was generated adding the partial scores of 6 of the 8 parameters considered, in particular those defined with the numbers 1-6 (theoretically ranging between 0 and 24). Altogether, the total score for each lung ranged between 0 and 72. The last total score permitted the classification of each lung of the same individual as either that with more affected lung or that with less affected lung. Throughout this report the total score represents the sum of partial score of findings by observing the more affected area for each lung.

Chest Radiology Score Generation

Using a Thoramat (Siemens, Erlangen, Germany) radiologic instrumentation and the same parameters as described, a radiologic total score for each area of the lung was generated to be compared with HRCT total score.

Bronchoalveolar Lavage (BAL)

Fiberoptic bronchoscopy was performed in all cases using the previously described method (20). Prebronchoscopy medications for each patient included intramuscular injections of 0.5 mg of atropine; 4% xylocaine was nebulized into the nasopharynx, a total of 4 ml of 1% lidocaine hydrochloride (Xylocaine) directly applied to the laryngeal area via the bronchoscope, and 2 ml of 1% Xylocaine directly applied to the trachea. Both of the lungs were found abnormal by HRCT. The BAL was performed in the more affected area for each lung. The bronchoscope was inserted near a segment of the affected lung as guided by the HRCT. Obviously, the less affected lung was lavaged first. A standard lavage protocol was performed by infusing five times 20-ml aliquots (a total of 100 ml) of sterile 0.9% saline, at body temperature, through the aspiration port. After each aliquot, lavage was collected via the same port, immediately by gentle hand suction into a plastic trap. The fluid recovered was filtered through sterile gauze and the volume measured. BAL fluid analysis was performed immediately after the bronchoscopy, and it was independently verified by one technician blind to the patient's condition. The epithelial lining fluid (ELF) volume was determined by the urea method (21). This approach permitted the correction of the concentrations of different molecules evaluated in BALF for the dilution factor, due to the technique adopted. A limit of this method is linked to possible urea alterations depending on inflammation. This fact determines underestimation of the more inflamed lesions. Therefore, the differences found could be even more evident than those observed.

Cytokine Concentrations

To determine the urea content of lavage fluid samples, a commercially available kit (Sigma 65 UV) was purchased from Sigma (Milan, Italy). Cytokine concentrations were measured in supernatants of BALF, after 20-fold concentration by means of AMICON concentrators (AMICON, Beverly, MA). Determinations of human IL-1, IL-2, IL-4, IL-6, IL-8, IL-10, IFN-, and TNF- were measured by enzyme linked immunosorbent assay (ELISA; R&D Systems, Minneapolis, MN). IL-12 was determined by means of an IL-12 + p40 EASIA kit (MEDGENIX Diagnostics, SA, Fleurus, Belgium). Kit sensitivities were as follows: IL-1 = 0.3 pg/ml; IL-2 = 2.5 pg/ml; IL-4 = 4.1 pg/ml; IL-6 = 0.7 pg/ml; IL-8 = 18 pg/ml; IL-10 = 2 pg/ml; IL-12 = 1.5 pg/ml; IFN- = 3 pg/ml; and TNF- = 4.4 pg/ml.

Statistical Analysis

The limited number of patients examined, the lack of knowledge of the data distribution, and the use of semiquantitative scores obliged a nonparametric analysis. Therefore the results were expressed as medians and ranges and comparisons were obtained by using the Wilcoxon rank paired test. Correlations were calculated by means of the Spearman rank correlation test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All 15 patients were analyzed by HRCT and chest radiology. The analysis of the correlations between these two methods, obtained using the total scores, showed a significant correlation coefficient (r = 0.74, p < 0.001). To establish which method was the more sensitive, a comparison of the frequency of the positive scores was done for each radiologic abnormality and for the total score (Table 1). Because the comparison showed a better HRCT performance, we used the data obtained with this technique throughout this report.

The more affected areas were the following: upper right zone in 10 cases, upper left in nine cases, middle left in six cases, and middle right in five cases. On the basis of the more or the less affected lung, we obtained two subgroups of 15 values for each cytokine and each HRCT finding. The observed total HRCT scores were a median of 10 (range 4 to 15) for the more affected lungs and a median of 4 (range 1 to 11) for the less affected lungs (p = 0.0004). Furthermore, a significant correlation (r = 0.69, p = 0.01) was found between the total HRCT scores obtained for the more and the less affected lungs of the same patients. To analyze the differences between cytokine concentrations of the more and the less affected lungs, the Wilcoxon rank paired test was employed, as well as for the HRCT scores (Tables 2 and 3). Significantly increased cytokine concentrations were observed for IL-1, IL-6, IL-8, IL-12, and TNF-, which are generally considered proinflammatory cytokines (Table 2). A similar analysis was obtained for the abnormality scores and the total HRCT score (Table 3). With the exception of bronchiectasis and miliary nodules, all comparisons were statistically significant. Total HRCT score yielded the greatest difference (0.0004 versus 0.004 obtained with the best individual HRCT abnormality).

The ELF levels of the cytokines were used to explore possible correlations with the scores for each finding and the total HRCT scores (Table 4). Generally, the correlations observed between the cytokine concentrations and the total HRCT score were better than those found with the individual findings. Significant direct correlations were observed between the total HRCT score and IL-6, IL-8, IL-12, and TNF-, whereas a negative correlation was found with IL-2 (Figure 1). Table 5 presents the correlation matrix between the concentrations of each cytokine and the others. Significant correlations were observed between TNF- and IL-1 or IL-6; IL-6 and IL-8 or IL-12; IL-8 and IL-12; IFN- and IL-2; and IL-4 and IL-1. Table 6 shows the correlation matrix for the HRCT scores. Interestingly, total score shows the higher correlation coefficients with the individual findings, with the exception of bronchiectasis and fibrosis, not included in the total score.

View larger version (18K): [in this window] [in a new window]   Figure 1.   Correlations between the cytokine concentrations with HRCT total score.

Very similar data, although less significant, were obtained when the 30 lungs were subdivided on the basis of the total score in the 15 lungs with the lower HRCT scores and the 15 lungs with the high HRCT scores (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There is very limited information in the literature concerning the concentrations of wide panel of cytokines, concomitantly determined in BALFs of patients affected with active pulmonary tuberculosis. To eliminate the variations caused by the dilution factor, all determinations were corrected using the urea method, as already described, even if this method understimated the possible differences. Therefore, the cytokine concentrations have been expressed as pg/ml of ELF for all cytokines except IL-2 for which U/ml of ELF was adopted. Generally, the other studies available report the cytokine concentrations measured in supernatants of unstimulated BAL cells, cultured for short periods (3, 4, 16). For this reason, no true comparison between our findings and those published may be done. Notwithstanding this, an analysis of these publications reveals increased releases of IL-1 and TNF- by unstimulated BAL cells from tuberculous patients as compared with normal subjects (3). In addition, IFN- and TNF- messenger ribonucleic acids (mRNAs) were increased in CD4⁺ T cells of mice infected with Mycobacterium avium, and these cytokines were important in early protection against this type of Mycobacterium (14). In another study, Lai and coworkers (22) showed that CD4⁺ T cells of tuberculous patients produced higher mRNA amounts of IL-2 than normal cells. In contrast, IL-5 mRNA was reduced, whereas IFN- and IL-4 mRNA concentrations remained unchanged. Finally, IFN- and IL-12 mRNA were found to be increased in BAL cells of patients with active tuberculosis as compared with those affected with inactive tuberculosis (11). These findings are in agreement with our results, with respect to the proinflammatory cytokines, including also IL-12 and IL-8.

HRCT is useful in evaluating radiologic images both from a quantitative and a qualitative point of view (23). Radiologic abnormalities were quantified by means of eight different scores evaluated by two expert chest radiologists by reading the conventional chest radiographs and HRCT images. Comparison between these two methods showed a better sensitivity of HRCT, the number of negative scores being more frequent in chest radiology evaluation. On the basis of several attempts, combining together various HRCT finding scores, we selected the total score, including all parameters except fibrosis and bronchiectasis (data not shown).

The use of an arbitrary score simplifies the comparison of the cytokine concentrations determined separately for the two lungs of the same patient. In fact, by means of the total score, we could generate two lung subgroups: one with the higher (more affected lung) and one with the lower score (less affected lung). Significant differences between the two groups were observed for IL-1, IL-6, IL-8, IL-12, and TNF-, suggesting that these mediators are associated with the amount of tuberculous disease as evaluated by HRCT. Evidently, the proinflammatory cytokines may relate to the intensity of inflammatory phenomena linked to tuberculosis in the two lungs.

A more in-depth analysis of this point was obtained by evaluating the correlations between each radiologic finding, total HRCT score, and cytokine concentrations. As reported, IL-6, IL-8, IL-12, and TNF- were significantly correlated to the total HRCT score, confirming their direct involvement in inflammatory phenomena. By contrast, IL-2 concentrations were inversely correlated with this total score. This finding may be interpreted in light of a report (16) identifying IL-2 as a protective agent against M. tuberculosis. TNF-, although considered as another protective mediator, has also been defined as the cause of cavitation processes (24). Significant correlations were also observed with each of the HRCT findings, although generally less significant, suggesting that total HRCT score could represent a useful biologically validated indicator of disease activity.

The utility of the arbitrary generated total HRCT score was also evident when a correlation was explored between the scores calculated for the more affected or less affected lungs of the same patients. In fact, a significant correlation between the total HRCT scores of the two lungs was observed, suggesting that the more affected one lung is, the more affected the second one is. Furthermore, the determination of more cytokine types permitted the calculation of reciprocal correlations between the different concentrations, in order to evaluate possible coordinated syntheses of these modulators. In fact, various correlations were observed between the proinflammatory cytokines, such as IL-8 versus IL-6 or IL-12 versus IL-8, generally produced by monocytes. Interestingly, the two Th1 cytokines, IL-2 and IFN-, produced by the same cells, were significantly correlated.

With regard to the correlations of each HRCT score, we underline that consolidation, bronchial wall thickening, and nodules 2 to 10 mm are strongly correlated with each other, whereas ground glass and miliary nodules are scarcely or not correlated at all with the other findings. In conclusion, the data presented in this report stress the concept that various cytokines are quantitatively associated with tuberculous disease as quantified by a HRCT score including six different radiologic abnormalities. This association may represent both a cause and effect of the pathogenetic mechanisms involved at the cellular level.

Cells participating in protection against M. tuberculosis are mainly monocyte/macrophages, granulocytes, CD4⁺/CD8⁺ T cells, and natural killer (NK) cells (12). These types of cells are notoriously activated by various proinflammatory cytokines including IL-1, IL-6, IL-8, TNF-, IL-2, IL-12, and IFN-. The concentrations of most of these cytokines were significantly different between the more and the less affected lungs. No significant differences were observed with regard to IL-2 and IFN-. Therefore, these two Th1 mediators were correlated with each other, and IL-2 showed a negative correlation with the total HRCT score. An intriguing finding is the significant correlation between IL-1 and IL-4. This result may be explained by the known activity exerted by IL-1 on some subsets of T lymphocytes that, as previously reported (25, 26), release more IL-4. Although this study provided indications that several cytokines are involved in the tuberculous processes, more studies are necessary to define their specific roles in pulmonary tuberculosis.