Australasian Radiology (1 996) 40, 1 1 3-1 18

Computed tomography in miliary tuberculosis: Comparison with plain films, bronchoalveolar lavage, pulmonary functions and gas exchange SK Sharma,’ S Mukhopadhyay,* R Arora,’ K VarmaI3 JN Pandel and GC Khilnani’ Departments of ‘Medicine, %adiodiagnosis and 3Pathology, All India Institute of Medical Sciences, New Delhi, India

SUM MARY

Computed tomography (CT) of the chest, pulmonary function tests, bronchoalveolar lavage (BAL) and arterial blood gas analysis were performed in 26 patients with non-HIV miliary tuberculosis (MTB). CT was repeated after treatment in 11 patients. Nodular lesions were characteristically seen on CT. CT showed discrete and fine nodules in five patients in whom the lesions appeared to be larger than miliary on chest X-rays. Coalescing nodular lesions were noted on chest X-rays (n=7) and CT (n=18). Consolidation (n=6), cavitation (n=4), fibrosis (n=9) and air trapping (n=14) were detected on CT only. During follow up, air trapping increased (n = 14) and in some patients it appeared for the first time (n = 2). Lymph node enlargement and calcification were seen on both chest X-rays (n = 9 and n = 3, respectively) and CT (n= 12 and n=7, respectively). Pleural involvement was also seen in chest X-rays (n=4) and CT (n=5). Total lung capacity was higher in patients with a chest X-ray score > 10. Similarly a higher total cell count in BAL fluid was observed in patients with a CT score > 10. It is concluded from this study that CT is superior t o chest X-rays in detecting nodular lesions, lymphadenopathy and air trapping in patients with MTB.

Key words: bronchoalveolar lavage; chest; chest X-ray; computed tomography; miliary tuberculosis.

INTRODUCTION Miliary tuberculosis (MTB) is an illness produced by acute dif- fuse dissemination of acid-fast bacilli via the bloodstream. A high degree of suspicion is important as any delay in treat- ment is associated with an unfavourable prognosis. Without treatment, this illness is almost uniformly fatal. The roentgeno- graphic hallmark of acute disseminated tuberculosis is the mil- iary pattern on chest X-ray. The term ‘miliary’ refers to the ‘millet seed’ size of the nodules (<2mm) seen on classical chest films.

It is believed by many workers that the plain X-ray appear- ance of miliary nodules is produced by the summation of multiple tiny nodules. It is, therefore, possible that computed tomography (CT) could detect the diffuse tiny nodules of MTB before their size and number enabled detection by a chest X- ray. Unfortunately, some patients with MTB have normal chest X-rays when they first seek medical attention and some have patterns that are indistinguishable from those of acute inter- stitial A number of studies during the past few years have shown that CT can be useful in the assessment of

patients with chronic diffuse infiltrative lung disease (DILD).6.7 Furthermore, by eliminating the superimposition of structures, CT allows a better assessment of the distribution and severity of abnormalities.

Computed tomography findings of MTB in two patients with human immunodeficiency virus (HIV) infection have been reported re~ent ly ,~.~ but the CT findings of MTB in non-HIV patients have not been described previously. Moreover, repeat CT studies in these patients before and after chemotherapy have not been undertaken. The present study was undertaken to study the CT findings before and after chemotherapy and to compare these findings with X-ray, bronchoalveolar lavage (BAL) cellular, pulmonary function and gas exchange parameters.

METHODS Study population Twenty-six consecutive adult patients with MTB seen at the All India Institute of Medical Sciences Hospital, New Delhi, were studied. None of these patients was suffering from HIV infec-

SK Sharma MD, FCCP; S Mukhopadhyay MD; R Arora MD; K Varma MD; JN Pande MD; GC Khilnanl MD. Correspondence: Dr Sima Mukhopadhyay, Additional Professor, Department of Radiodiagnosis, All India Institute of Medical Sciences, New Delhi 110 029, India. Submitted 10 February 1995; resubmitted 13 June 1995; accepted 7 July 1995.

114 SK SHARMA ET AL.

tion. Criteria for the diagnosis included: (i) the presence of miliary nodules on chest X-ray which were defined as discrete radiographic densities measuring 1-2 mm; (ii) bacteriological proof of infection with Mycobacteriurn tuberculosis; (iii) biopsy specimens demonstrating caseating granulomas with acid-fast bacilli: and (iv) clinical presentation consistent with tuberculo- sis, with marked improvement after antituberculosis therapy. All patients fulfilled the first criterion and one or more of the others. However, five patients had nodular shadows measur- ing >2mm in their chest X-rays and tissue diagnosis was available for all these patients. Six patients fulfilled only the first and fourth criteria and X-ray findings in these patients did not differ from patients who had a proven diagnosis of MTB. A barium study in one of these six patients revealed findings compatible with the diagnosis of ileocaecal tuberculosis. In the remaining five patients other diagnostic tests were negative. Tissue biopsies (skin, lymph node, liver, bone and tonsil) in 11 patients revealed caseating granulomas, and staining for acid- fast bacilli was positive. M. tuberculosis was demonstrated in six patients in sputum/BAL fluid smears and culture was posi- tive in four of these. M. tuberculosis was isolated in pleural fluid of one patient and CSF in another. One patient had thoracoplasty for the treatment of pulmonary tuberculosis 30 years ago and at present had tuberculosis of the thoracic spine along with cold abscess. The mean age of the patients was 31.9 years (12.7 s.d.; range 17-64 years). There were 17 women. All but two were non-smokers. Their mean body-mass index (BMI) was 18.6kg/m2 (3.8 s.d.; range 13.3-26.8kg/m2). Fourteen patients had a BMI < 18. The average duration of symptoms was 3.4 months (3.2 s.d.; range 0.5-12 months).

Chest X-ray, conventional and high resolution computed tomography All 26 patients were subjected to chest X-ray and CT at first presentation. Chest X-rays were repeated in all patients at the end of antituberculosis treatment. However, CT was repeated in 11 patients after a variable period of antituberculosis therapy (mean 11 months; range 3-24 months). An X-ray of the chest was also taken in these 11 patients at the time of CT. Thus, 37 CT scans were performed in 26 patients. Of these, 14 were conventional CT (CCT) and 23 were both CCT and high resolution CT (HRCT). CT scans were performed on a 3rd generation Somatom DR-H scanner (Siemens). For CCT, 8mm thick sections were taken through the lungs at the end of full inspiration. High resolution CT was performed by taking two sections each of 2mm thickness at three levels (i.e. aortic arch, carina and 2cm above the dome of diaphragm). Intra- venous contrast was used in five patients as the possibility of mediastinal lymphadenopathy could not be clarified by non- enhanced scans in these patients. Scans were viewed with appropriate settings for the lungs (level 700HU, with 1000HU)

and the mediastinum (level 35HU, width 400HU). Chest films were assessed for the type of opacities (discrete nodularity or conglomeration), distribution of the nodules, lymph node enlargement and/or calcification and pleural involvement such as thickening, effusion, pneumothorax.

Each lung was divided into three zones (i.e. upper, middle and lower) by horizontal lines drawn at one-third and two-thirds of the vertical distance between the apex of the lungs to 1 cm below the domes of diaphragmlo The chest X-rays were scored by visual estimation of the degree of involvement of each zone: 0 = no involvement; 1 = mild; 2 = moderate; and 3 =severe involvement. The score given to each lung was added to give a total X-ray score.

CT scans were reviewed for the presence of nodular lesions, pattern of distribution of the lesion on vertical and cross-section basis, conglomeration of the lesions, fibrosis, cavitation, con- solidation, lymph node enlargement or calcification, pleural pathology and possible pathology in the visible part of the liver. Areas of lucency conforming to the polygonal shape of a secondary pulmonary lobule were defined as areas of air trapping.

Chest X-rays and CT scans were evaluated by two review- ers (SM and SKS) and scoring was done by consensus. In the case of a discrepancy, a reason was sought in other clinical and CT findings. The initial chest X-ray, CT scan, pulmonary function tests and bronchoalveolar lavage (BAL) were per- formed at approximately the same time. Plain radiographic and CT scores were correlated with each other and with functional parameters and BAL findings.

Pulmonary functions, gas exchange and bronchoalveolar lavage The lung volumes and their subdivisions were measured using a constant volume variable pressure body plethysmograph (P.K. Morgan, Chatham, UK) as described previously.” Airway resistance and thoracic gas volume were estimated with the patient panting inside the box at a frequency of 1 Hz. Pulmo- nary diffusing capacity (DLCO) was measured by the steady state technique as described previously.” The arterial blood gases were determined by a Radiometer ABL3 blood gas ana- lyser (Radiometer, Copenhagen, Denmark).

Treatment All patients were treated for 9 months with antituberculosis drugs, namely, streptomycin, ethambutol, rifampicin, isoniazid and pyrazinamide, in the combination streptomycin, rifampi- cin, isoniazid and pyrazinamide or ethambutol, rifampicin, isoniazid and pyrazinamide for the initial 2 months followed by rifampicin and isoniazid for a subsequent 7 months. None of the patients in this study received corticosteroids.

CT IN MlLlARY TUBERCULOSIS 115

Statistical analysis Pulmonary function, arterial blood gas and BAL data are expressed as mean t s.d. Chest X-ray and CT scores were correlated by applying Pearson’s correlation coefficient. The data for continuous variables were divided into two subgroups based on chest X-ray and CT scores of 10 or < 10. As appro- priate, either the unpaired t-test or the Wilcoxon rank-sum were applied to compare the levels of various parameters of pul- monary function, arterial blood gases and BAL in these two subgroups.

RESULTS Chest X-rays One patient with MTB had a few nodular lesions on chest X- ray (Fig. la). The remaining patients had extensive fine nodu- lar lesions in both lungs on chest X-ray. Five patients had nodules larger than classical miliary opacities (> 2mm). Lesions were uniformly distributed in 12 and non-uniformly in 14 patients. In the patients with non-uniformly distributed lesions mid and lower zones were involved to a greater extent. A con- glomeration of lesions was noted in seven cases. None of the patients had cavitating lesions. Lymph node enlargement was seen in nine patients (right paratracheal 5, hilar 4) and calcifi- cation in three. Pleural involvement (effusion 1, thickening 2, pneumothorax 1) was seen in four patients. The chest X-ray score varied from 3 to 18. Chest X-rays showed complete clearance of lesions in all 26 patients at the end of treatment.

Computed tomography scan Lung nodules were mostly discrete and fine in 17 patients (Fig. lb) and very fine in nine patients. In 18 patients there was a variable degree of conglomeraton of the nodules (Fig. 2). All five patients showing larger nodules on plain X-ray films

showed fine nodules on CT. The distribution of nodules was uniform from the apex to the base in eight patients and non- uniform in 18 patients. A striking feature of relative sparing of the periphery of the lung was seen in 10 patients (Fig. 3). The CT score varied from 6 to 18. Figure 4 shows a thin walled cavity or focal area of bronchiectasis in the left lung along with scattered areas of air trapping in both lungs. Most of the lucen- cies in the images in Fig. 5a conformed to the polygonal shape of a secondary pulmonary lobule, and were less lucent than the typical lesion of emphysema and these lucencies repre- sented areas of air trapping. Air trapping was seen at the first presentation in 14 patients. In two patients areas of air trap- ping were seen on repeat CT following treatment with anti- tuberculosis drugs. In four patients there was aggravation of air trapping during follow-up studies (Fig. 5b, 6). Lymph node enlargement was seen in 12 patients (right paratracheal 10,

Fig. 2 High resolution CT scan at intermediate bronchus level shows non-uniform distribution of nodular lesions in both lungs with areas of conglom-eration,

Fig. 1 (a) The chest X-ray shows a few nodular lesions in lung fields. (b) High resolution CT scan of the same patient through the upper lobes shows extensive bilateral almost uniformly distributed nodular lesions.

116 SK SHARMA ET AL.

hilar 2). Some of these patients also had subcarinal, pre- tracheal and azygoesophageal lymphadenopathy. Calcified nodes were seen in seven patients (hilar 4, paratracheal 3). Some of them also had calcified azygoesophageal and aorto- pulmonary window lymph nodes. Six patients had associated

consolidation, nine had fibrosis and four had cavitation. Pleural involvement was noted in five patients (bilateral pleural effu- sion 1, pneumothorax 1, thickening 3). There was spinal tuber- culosis in two patients and diffuse low attenuation lesions were seen in the liver in one patient on both non-enhanced and contrast-enhanced CT scans. On repeat CT, the lesions in the liver had cleared completely after antituberculosis therapy.

Pulmonary function and gas exchange Pulmonary function, except airways resistance and specific airways conductance, have been expressed as per cent predicted. Patients with MTB had low forced vital capacity (NC) (53.16? 14.8), forced expiratory volume in 1 s (FEV,) (54.17 k 12.24), peak expiratory flow rate (PEFR) (62.8 ? 21.6), flow at 50% of the vital capacity (v 50) (51 t 20.5), flow at 75% of the vital capacity (v75) (29.47 t 15.62) and flow during the middle half of vital capacity (FEF 25-75) (49.9 k 23.12). Their FEV,/FVC (%) (84.27 -t 9.9). functional residual capacitv (FRC) . , . . _ . .

(122.37 ? 38), residual volumehotal lung capacity (RVRLC) (%)

(51.68 ? 10.55) were increased. Their pulmonary diffusing

Fig. 3 CT scan through upper lobes shows extensive fine nodules in both lungs with relative peripheral sparing.

Fig. 4 High resolution CT at carina shows extensive nodular lesions with confluence mainly in right lung and focal area of bronchiectasis or a thin walled cavity in left lung along with scattered areas of air trapping.

Fig. 6 Follow-up CT scan of a patient at the end of antituberculosis therapy. The nodules have disappeared but air trapping has increased.

Fig. 5 (a) CT scan through upper lobes showing non-uniform distribution of nodules with areas of air trapping in both lungs. (b) Follow-up CT scan of the same patient after 7 months of antituberculosis therapy. The nodules have disappeared and air trapping has increased significantly.

CT IN MlLlARY TUBERCULOSIS 117

capacity (DLCO) (55.12 k 22), P,o, (69.75 ? 16.62) and Pacop (34.3 ? 10.96) were low. They had high airways resistance (Raw) (1.73 t 0.91) (cm H,O/Us) and normal specific airways conductance (SGaw) (0.20 ? 0.08) (cm s-l cm H,O-'). Six of 14 patients with areas of air trapping at the first presentation had low FVC despite low CT score. All these patients had elevated RVTTLC(%) suggesting hyperinflation of lung volumes thus explaining their lower FVC. Various parameters of pul- monary functions in patients with areas of air trapping did not differ significantly when compared with patients who did not have air trapping on CT.

Bronchoalveolar lavage Bronchoalveolar lavage in MTB patients revealed increased total cell count (2.6 5 1.8 x lo6 cells/mL) and proportion of lymphocytes (52.3 2 19.6%).

Correlations Computed tomography and chest X-ray scores correlated sig- nificantly (r = 0.9, n = 25, P < 0.001) with each other. Patients with a chest X-ray score > 10 had a higher total lung capacity (P < 0.05). Similarly patients with a CT score > 10 had a higher total cell count (P < 0.05) in the BAL fluid.

DISCUSSION Early diagnosis of MTB is important because of its high com- municability and potential for effective treatment, if adminis- tered early in the course of the disease. The diagnosis of MTB can be difficult for the following reasons. Clinical symptoms are sometimes subacute and non-specific. Specific respiratory symptoms are relatively unimpressive until late in the disease. The plain film may be normal at presentation and the abnor- malities may appear relatively late in the course of disease. The early chest X-ray findings may be suggestive of non- specific interstitial disease and late in the course of disease an X-ray may reveal coalescent opacities, making recognition of miliary shadows more difficult. Recent descriptions of the CT appearance in interstitial lung disease have suggested that CT may complement chest radiography in the diagnosis of inter- stitial lung disease.I2-l4 CT offers the advantage of decreased superimposition of parenchymal opacities.

Fortunately, all patients in the present study had abnormal chest X-rays. Compared with chest X-ray, CT was superior in the diagnosis of MTB in one patient in whom there were subtle abnormalities on chest X-ray. As on the plain films, the most characteristic feature on CT was the presence of discrete and fine nodules. Ten of the 26 patients with MTB had relative peripheral sparing of the lungs of nodular densities on CT. Nodular densities on CT may be seen in various interstitial lung diseases and these include silicosis, coal workers' pneu- moconioses, sarcoidosis, histiocytosis X, extrinsic allergic alveolitis and lymphangitis carcinornat~sis.~~~~ A posterior pre-

dominance of nodules is seen on CT in patients with sil ic~sis.~ Nodular opacities on CT in sarcoidosis are distributed along with bronchovascular bundles, interlobular septae, major fis- sures and subpleural regions. In the present study, by decreasing the superimposition of opacities, the CT was better in defining the morphology of miliary opacities compared with the chest X-rays. This was demonstrated by the presence of discrete and fine nodules on CT in five patients in whom chest X-rays revealed the opacities to be larger than miliary. Further- more, compared with chest X-ray, CT appeared to detect more conglomeration of nodules. The superiority of CT over conven- tional radiography has been reported by various authors in the assessment of disease extent and distribution in patients with diffuse infiltrative lung disease, including sarcoidosis.16-m In the present study CT demonstrated lymph nodal enlargement, calcification and pleural lesions in more patients than plain films. Areas of lucency conforming to the polygonal shape of a secondary pulmonary nodule were appreciated only on chest CT. These areas were less lucent than typical lesions of emphysema. Areas of emphysema, in contrast, are usually round and poorly defined. These lucencies represented areas of air trapping rather than true emphysema and are perhaps related to tuberculous bronchiolitis. Similar areas of patchy lucency have been seen in bronchiolitis obliterans.21,n This diagnosis of air trapping would not be made on the plain films in any of the patients in the present study.

In the present study a significant correlation was observed between chest X-ray score and total lung capacity, and total cell count in the BAL fluid and CT score. Remaining param- eters of pulmonary function, gas exchange and BAL did not correlate significantly either with chest X-ray or CT scores. Poor correlation between the severity of chest X-ray changes and clinical and functional impairment has been described in sar- c o i d o ~ i s . ~ ~ ~ ~ ~ In a previously reported study in patients with sarc~idosis,~~ CT grades (as indicated by the measurement of percentage of the volume judged to be abnormal) did not cor- relate with gallium scanning results and the percentage of BAL lymphocytes recovered. Poor correlation between roentgeno- graphic severity of disease and the functional impairment in MTB in the present study may be due to the fact that, although easily seen and quantified, the nodular lesions cause minimal dysfunction. This situation may be similar to that in silicosis and sarcoidosis in which the severity of interstitial fibrosis rather than the number or size of the nodules is responsible for the impaired function.26 As a group, patients with silicosis who have predominantly nodular opacities have less functional impairment than those with predominantly irregular opacities.

Thus, we conclude from this study that CT may be superior in detecting various lesions associated with MTB compared with chest X-ray. It is also useful in detecting air trapping at initial presentation and during the follow-up period. We do not know the clinical significance of various observed findings on

118 SK SHARMA ET AL.

CT at the present time. Long-term studies with large numbers of patients and good follow-up are required in order to find out the significance of these CT findings. We also do not advocate the routine use of CT for the diagnosis of miliary TB in clear- cut cases. However, CT can be useful in difficult cases with a normal or early nonspecific plain radiograph.

REFERENCES 1. Munt PW. Millary tuberculosis in the chemotherapy era: With a

cllnical review in 69 American adults. Medicine 1971 ; 51 : 139-55. 2. Linell F, Ostberg G. Tuberculosis in an autopsy material. With spe-

cial reference to cases not discovered until necropsy. Scand. J. Respir. Dis. 1966; 47: 200-8.

3. Berger HW, Samortin TG. Miliary tuberculosis: Diagnostic meth- ods with emphasis on the chest roentgenogram. Chest 1970; 58:

4. Proudfoot AT, Akhtar AJ, Dougias AC et a/. Miliary tuberculosis in adults. BMJ 1969; 1: 273-6.

5. Khatua SP. Tuberculosis meningitis in children: Analysis of 231 cases. J. lndian. Med. Assoc. 1961 ; 37: 332-7.

6. Muller NL, Miller RR. State of the art. Computed tomography of chronic diffuse infiltrative lung disease. Part I. Am. Rev. Respir. Dis.

7. Muller NL, Miller RR. State of the art. Computed tomography of chronlc dlffuse infiltrative lung disease. Part Ii. Am. Rev. Respir. Dis. 1990; 142: 1440-8.

8. Optican RJ, Oat A, Ravin CE. High resolution computed tomogra- phy in the diagnosls of miliary tuberculosis. Chest 1992; 102:

9. McGuinness G, Naidich DP, Jagirdar J, Leitman B, McCauiey DI. High resolution CT findings in miliary lung disease. J. Comput. Assist. Tomogr. 1992; 16: 384-90.

10. Muller NL, Mawson JB, Mathieson JR, Abbound R, Ostrow DN, Champion P. Sarcoidosis: Correlation of extent of disease at CT with cliniciai, functional and radiographic findings. Thoracic Radi-

11. Sharma SK, Pande JN, Verma K. Effect of prednisolone treatment in chronic silicosis.Am. Rev. Respir. Dis. 1991; 143: 814-21.

12. Bergin CJ. Muiier NL. CT in the diagnosis of interstitial lung dis- ease. AIR 1985; 145: 505-10.

586-9.

1990; 142: 1206-15.

941 -3.

ology 1989; 171: 613-18.

13. Steinberg DL, Webb WR. CT appearance of rheumatoid lung dis- ease. J. Comput. Assist. Tomogr. 1984; 8: 881 -4.

14. Yoshimura H, Hatakeyama M, Otsuji H eta/. Pulmonary asbesto- sis. CT study of subpleural curvilinear shadow. Work in progress. Radiology 1986; 158: 653-8.

15. Aberie DR, Gamsu G, Ray CS. High resolution CT of benign as- bestos related diseases: Clinical and radiographic correlation. AJR

16. Naidich DP, Zerhouni EA, Hutchins GB, McCauiey DL, Siegellman SS. Computed tomography of the pulmonary parenchyma. I. Dis- tal airspace diseases. J. Thorac. lmaging 1985; 1: 39-53.

17. Zerhouni EA, Naidich DP, Stitik FP, Khouri NF, Siegelman SS. Computed tomography of the pulmonary parenchyma. II. Intersti- tial disease. J. Thorac. Imaging 1985; 1 : 54-64.

18. Bergin CJ, Muller NL. CT of interstitial lung disease: A diagnostic approach. AJR 1987; 148: 8-15.

19. Soloman A, Kreel L. McNicol M, Johnsons N. Computed tomog raphy in pulmonary sarcoidosis. J. Comput. Assist. Tomogr. 1979;

20. Hamper WM, Fishman EK, Khouri NF, Johns CJ, Wang KP, Sie- gelman SS. Typical and atypical manifestations of pulmonary sarcoidosis. J. Comput. Assist. Tomogr. 1986; 10: 928-36.

21. Sweatman MC, Millar AB. Strickland B, Turner-Warwick M. Com- puted tomography in adult obliterative bronchiolitis. Clin. Radiol.

22. Lynch DA, Brasch RC, Hardy KA, Webb WR. Pediatric pulmonary disease: Assessment with high-resolution ultrafast CT. Radiology

23. Young RL, Krumholtz RA, Harkleroad LE. A physiologic roentgen- ographic disparity in sarcoidosis. Dis. Chest 1966; 50: 81-6.

24. Renzi G, Dutton RE. Pulmonary function in diffuse sarcoidosis. Respiration 1974; 31 : 124-36.

25. Bergin CJ, Bell DY, Coblentz CL, Chiles C, Gamsu G, Maclntyew NR, Coleman RE, Putman CE. Sarcoidosis: correlation of pulmo- nary parenchymal pattern at CT with results of pulmonary function tests. Radiology 1989; 171: 619-24.

26. Carrington CB, Gaensler EA, Minkus JP. Schachter AW, Burke GW, Goff AM. Structure and function in sarcoidosis. Ann. NYAcad.

1988; 151 : 683-91.

3: 754-8.

1990; 40: 116-19.

1990; 176: 243-8.

Sci. 1976; 278: 265-83.