Disseminated Tuberculosis
A 10-Year Experience in a Medical Center
Jann-Yuan Wang, MD, Po-Ren Hsueh, MD, Shu-Kuan Wang, BS, I-Shiow Jan, MD, Li-Na Lee, MD, PhD,
Yuang-Shuang Liaw, MD, PhD, Pan-Chyr Yang, MD, PhD, and Kwen-Tay Luh, MD, PhD
Abstract: Disseminated tuberculosis remains a diagnostic challenge
because the presentations are nonspecific. In the current retrospective
study we describe the clinical characteristics and outcome of
disseminated tuberculosis. From January 1995 to December 2004,
patients with culture-confirmed tuberculosis who fulfilled the criteria
for disseminated tuberculosis were selected and their medical records
reviewed. Their clinical isolates were genotyped. Of the 3058
patients with culture-confirmed tuberculosis, 164 (5.4%) had
disseminated disease; 14.0% of patients had acquired immunodefi-
ciency syndrome. The most common radiographic finding was
miliary lung lesions (47.0%); 31.1% of patients died at the end of the
study. Poor prognostic factors included hypoalbuminemia, hyper-
bilirubinemia, renal insufficiency, and delayed antituberculosis
treatment. Clinical findings suggestive of disseminated tuberculosis
were miliary lung lesions, serum ferritin >1000 mg/L, infiltrative
liver disease, and adjusted calcium >2.6 mmol/L. Simultaneously
performing mycobacterial culture and histopathologic examination
of bone marrow biopsy was more sensitive and faster than just
performing mycobacterial blood culture in diagnosing disseminated
tuberculosis. Of the 64 preserved Mycobacterium tuberculosis
isolates, 47 (73.4%) were clustered and 27 (42.2%) were Beijing
family. Since prognosis was worse in patients with delayed treatment,
a high index of suspicion is required, especially in those with clinical
findings suggestive of disseminated tuberculosis.
(Medicine 2007;86:39–46)
Abbreviations: AIDS = acquired immunodeficiency syndrome, ALP =
alkaline phosphatase, ESRD = end-stage renal disease, GGT = gamma
glutamyltransferase, HIV = human immunodeficiency virus, TB =
tuberculosis.
INTRODUCTION
A lthough extrapulmonary tuberculosis (TB) had been
observed for many centuries, the exact incidence of
disseminated TB is still unclear. Disseminated TB is defined
as tuberculous infection involving the blood stream, bone
From Department of Internal Medicine (JYW, YSL, PCY) and Department
of Laboratory Medicine (PRH, SKW, ISJ, LNL, KTL), National Taiwan
University Hospital, Taipei, Taiwan.
Address reprint requests to: Li-Na Lee, MD, PhD, Department of Laboratory
Medicine, National Taiwan University Hospital, No. 7, Chun Shan
South Road, Taipei, 100, Taiwan. Fax: 886-2-2322-4263; e-mail:
linalee@ntu.edu.tw.
Copyright n 2007 by Lippincott Williams & Wilkins
ISSN: 0025-7974/07/8601-0039
DOI: 10.1097/MD.0b013e318030b605
Medicine  Volume 86, Number 1, January 2007
Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
marrow, liver, or 2 or more noncontiguous sites, or miliary
TB2,22. The symptoms are nonspecific and the duration of
symptoms before diagnosis is variable7,16,18,20,22,26. There-
fore, it mimics a variety of diseases and requires a high index
of suspicion. Of all cases of disseminated TB found at
autopsy, 33%–80% were missed antemortem28. Moreover,
previous studies revealed that misdiagnosis leading to a
treatment delay of a mere 1–8 days contributes to a high
mortality rate4,9,30. Therefore, we conducted the current
study to describe the clinical characteristics and outcome of
patients with disseminated TB.
PATIENTS AND METHODS
The current study was conducted in a 1500-bed,
tertiary-care referral center in northern Taiwan. Disseminat-
ed TB was diagnosed if the patient had any of the following
conditions:
1. Isolation of Mycobacterium tuberculosis from blood, bone
marrow, liver biopsy specimen, or
2 noncontiguous
organs5;
2. Isolation of M. tuberculosis from 1 organ and histologic
demonstration of caseating granulomatous inflammation
from the bone marrow, liver biopsy specimen, or another
noncontiguous organ;
3. Isolation of M. tuberculosis from 1 organ and radio-
graphic finding of miliary lung lesions2,22.
A patient with isolated tuberculous hepatic abscess or
tuberculoma rather than diffuse hepatic involvement was not
included in the study1. The dual sites involvement of cervical
lymph node and lung was regarded as a loco-regional disease
rather than disseminated disease34.
We searched the mycobacterial laboratory and his-
tology databases from January 1995 to December 2004.
Those who fulfilled the criteria for disseminated TB were
included and their medical records reviewed. Preserved
isolates of M. tuberculosis from those patients were sub-
cultured on Lowenstein-Jensen media, incubated at 37 8C in
an aerobic atmosphere with 5%–10% CO2, and genotyped
using spoligotyping14. Strains that only hybridized to the last
9 spacer oligonucleotides (spacers 35–43) were defined as
Beijing family12. A cluster was defined as a group of 2 or
more isolates of the same DNA fingerprint.
Routine mycobacteriologic studies, including acid-fast
smear, mycobacterial culture, and drug susceptibility test,
were performed as previously described31. The medium
for primary mycobacterial isolation was Lowenstein-Jensen
medium (BBL, Becton Dickinson, Sparks, MD) before 1996;
39
Wang et al
after 1996, Middlebrook 7H11 selective agar with anti-
microbials (Remel Inc., Lexena, KS) was added. The
fluorometric BACTEC technique (BACTEC Mycobacte-
rium Growth Indicator Tube [MGIT] 960 system, Becton-
Dickinson, Sparks, MD) was added after July 1998.
Serum calcium level was adjusted by adding 0.2 mmol/L
for every 1 g/dL decrease of serum albumin below 4 g/dL25.
Infiltrative liver disease was defined if serum alkaline
phosphatase (ALP) was elevated to >440 U/L (normal =
<220 U/L) with elevated serum gamma glutamyltransferase
(GGT) level (normal = <53 U/L) or GGT >106 U/L. Standard
antituberculosis treatment consisted of isoniazid, ethambutol,
rifampicin, and pyrazinamide (HERZ) in the 2-month induction
phase and HER in the maintenance phase, and was modified
according to the presence of hepatic and/or renal disease,
adverse effects, and the results of drug susceptibility testing
after it became available. In patients with liver disease,
pyrazinamide may be omitted. The maintenance phase was
4 months in general, but was prolonged to 7 months in patients
with skeletal involvement. Multidrug-resistant TB, defined as
TB simultaneously resistant to isoniazid and rifampicin, was
treated for at least 18 months. The antituberculosis treatment
was considered early if started within 14 days after the initial
visit, and was judged as completed if fulfilling the definition
of the World Health Organization33. All deaths were either
identified by reviewing the medical records in our hospital
or captured through the National Surveillance Network of
Communicable Disease (Centers for Disease Control, Taiwan).
All patients were followed-up until completely treated, died, or
until June 2005 (end of the study).
Intergroup differences were analyzed using either the
ANOVA (continuous variables) or chi-square test (categori-
cal variables). One-year survival curves for each variable
with possible prognostic significance were generated using
the Kaplan-Meier method and were compared using the log-
rank test. If a significant difference (p < 0.05) was reached,
the variables then entered the multivariate survival analysis
using the stepwise forward Cox regression to identify factors
independently associated with mortality.
RESULTS
From January 1995 to December 2004, a total of 3058
patients with culture-confirmed TB were identified. Of these,
164 (5.4%) fulfilled the diagnostic criteria for disseminated
TB. Laboratory tests for human immunodeficiency virus
(HIV) were performed in 157 (95.7%) patients. Eighty-seven
(53.0%) patients had underlying comorbid conditions, the
most common being acquired immunodeficiency syndrome
(AIDS) (23 patients, 14.0%) and diabetes mellitus (23
patients, 14.0%), followed by malignancy (16 patients,
9.8%), end-stage renal disease (ESRD) (14 patients, 8.5%),
liver cirrhosis (11 patients, 6.7%), status post-transplantation
(6 patients, 3.7%), autoimmune disease (5 patients, 3.0%),
and alcoholism (3 patients, 1.8%). In the 23 AIDS patients,
CD4 count was available for 18 (78.3%). The median CD4
count was 38 mL (range, 2–135 mL). Kaplan-Meier survival
analysis revealed that the underlying comorbid condition
(HIV/AIDS vs. other comorbidities vs. no comorbidities)
40
Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Medicine  Volume 86, Number 1, January 2007
was significantly associated with prognosis (p = 0.002)
(Figure 1). The most commonly involved organs were the
lungs (143 patients, 87.2%) and musculoskeleton (32 patients,
19.5%), followed by the urogenital system (28 patients,
17.1%). HIV/AIDS patients had a higher incidence of
mycobacteremia (30.4%) and bone marrow involvement
(39.1%) (Table 1). TB peritonitis or pericarditis was more
common in patients with underlying diseases other than HIV/
AIDS (23.4%).
Fourteen (8.5%) and 5 (3.0%) M. tuberculosis isolates
were resistant to isoniazid and rifampicin, respectively. All
rifampicin-resistant isolates were also multidrug resistant.
Significantly more patients in the HIV/AIDS group received
antituberculosis treatment at an early stage than patients in
the other 2 groups (p = 0.015, see Table 1). The most
common radiographic findings on chest images were miliary
lesions (47.0%) and consolidation (32.3%). The former was
more common in the HIV/AIDS (69.6%) group, whereas the
latter was usually seen in patients with other comorbid
conditions (42.2%). Only 4.9% of patients with disseminated
TB had mainly fibrotic change on their chest images.
Cavitation was also rare (2.4%).
The most common abnormal laboratory findings were
hypoalbuminemia (74.7%) (normal = >3.5 g/dL), elevated
GGT (71.0%) (normal = <53 U/L), and ALP (66.4%)
(normal = <220 U/L), followed by hyponatremia (58.9%)
(normal = <135 mmol/L) and anemia (43.9%) (normal =
>10 g/dL) (Table 2). Elevated aspartate aminotransferases
(normal =  40 U/L), alanine aminotransferases (normal =
 40 U/L), and total bilirubin (normal =  1.0 mg/dL) were
found in 37.0%, 20.8%, and 25.5%, respectively. Patients
with underlying comorbid conditions other than HIV/AIDS
had most abnormal results. Hypercalcemia (adjusted calcium
>2.6 mmol/L) was noted in 22 patients (13.4%), including
7 with ESRD and 5 with malignancy. None of them received
vitamin D supplementation. Of them, serum phosphate and
intact parathyroid hormone levels were available in 20 and 7,
respectively. The average phosphate level was 5.1 mg/dL
(range, 2.6–9.5; normal = 3–4.5 mg/dL), while intact
FIGURE 1. Survival curves for patients with different under-
lying comorbid conditions, plotted using the Kaplan-Meier
method. Black dots represent patients who were still alive at
the end of the current study.
n 2007 Lippincott Williams & Wilkins
Medicine  Volume 86, Number 1, January 2007 Disseminated Tuberculosis

TABLE 1. Clinical Characteristics of 164 Patients With Disseminated Tuberculosis

HIV/AIDS Other Comorbidities No Comorbidities
(n = 23) No. (%) (n = 64) No. (%) (n = 77) No. (%) p*
Age, in yr: mean (SD) 37.1 (9.4) 61.4 (14.9) 58.9 (23.8) < 0.001
Sex: M/F 22 (95.7) / 1 47 (73.4) / 17 50 (64.9) / 27 0.015
Presenting symptom
Fever 16 (69.6) 36 (56.3) 26 (33.8)
Malaise 4 (17.4) 9 (14.1) 4 (5.2)
Weight loss 6 (26.1) 2 (3.1) 4 (5.2)
Poor appetite 2 (8.7) 0 7 (9.1)
Night sweats 2 (8.7) 1 (1.6) 2 (2.6)
Cough 11 (47.8) 13 (20.3) 13 (16.9)
Dyspnea 1 (4.3) 8 (12.5) 9 (11.7)
Myalgia 0 2 (3.1) 7 (9.1)
Back or joint pain 0 3 (4.7) 17 (22.1)
Dysuria 0 1 (1.6) 6 (7.8)
Scrotal swelling 0 0 4 (5.2)
Consciousness change 0 8 (12.5) 7 (9.1)
Headache 0 0 4 (5.2)
Abdominal pain 1 (4.3) 5 (7.8) 4 (5.2)
Abdominal distention 0 2 (3.1) 5 (6.5)
Neck mass 2 (8.7) 0 4 (5.2)
Symptom duration >3 wk 18 (78.3) 32 (50.0) 53 (68.8) 0.018
Proved involvementy
Blood 7 (30.4) 6 (9.4) 2 (2.6)
Bone marrow 9 (39.1) 12 (18.8) 2 (2.6)
Liver 4 (17.4) 3 (4.7) 3 (3.9)
Lung 16 (69.6) 58 (90.6) 69 (89.6)
Lymph node 12 (52.2) 4 (6.3) 11 (14.3)
Urogenital 1 (4.3) 8 (12.5) 19 (24.7)
Central nervous system 3 (13.5) 6 (9.4) 9 (11.7)
Musculoskeletal 0 11 (17.2) 21 (27.3)
Peritoneum/pericardium 0 15 (23.4) 9 (11.7)
Sputum AFS positive 6 (26.1) 18 (28.1) 16 (20.8) 0.587
Isoniazid resistance 3 (13.0) 3 (4.7) 8 (10.4) 0.341
Rifampicin resistance 1 (4.3) 2 (3.1) 2 (2.6) 0.991
Any-drug resistance 3 (13.0) 5 (7.8) 13 (16.9) 0.276
Multi-drug resistance 1 (4.3) 2 (3.1) 2 (2.6) 0.991
Clusteringz
Yes, Beijing strain 4 (44.5) 12 (48.0) 11 (36.7)
Yes, non-Beijing strain 2 (22.2) 6 (24.0) 12 (40.0)
No 3 (33.3) 7 (28.0) 7 (23.3) 0.715
Pattern on chest image
Consolidation 5 (21.7) 27 (42.2) 21 (27.3) 0.132
Fibrotic change 0 4 (6.3) 4 (5.2)
Miliary lesion 16 (69.6) 24 (37.5) 37 (48.1)
Nodule 0 0 3 (3.9)
No parenchymal lesion 2 (8.7) 9 (14.1) 12 (15.6)
Early anti-TB treatmentx 18 (78.3) 28 (43.8) 37 (48.1) 0.015
*All factors were analyzed by the chi-square test, except age (using ANOVA).
y
Involvement was proved by either mycobacterial culture or histopathologic examination.
z
The results of genotyping were available in only 63 patients.
x
Antituberculosis treatment was considered early if started within 14 d after initial visit.

n 2007 Lippincott Williams & Wilkins 41

Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Wang et al Medicine  Volume 86, Number 1, January 2007

TABLE 2. Laboratory Findings of 164 Patients With Disseminated Tuberculosis

Other Comorbidities No Comorbidities
HIV/AIDS (n = 23) (n = 64) (n = 77)
p value
Case Mean (SD) Case Mean (SD) Case Mean (SD) in ANOVA
Hemoglobin (g/dL) 23 9.5 (1.6) 64 9.9 (2.3) 77 11.2 (2.0) <0.001
Leukocyte (109/L) 23 5411 (3686) 64 7609 (4831) 77 8109 (3225) 0.019
Platelet (109/L) 22 188 (100) 64 182 (124) 77 282 (130) <0.001
Albumin (g/dL) 21 3.0 (0.6) 59 2.7 (0.6) 66 3.2 (0.8) <0.001
Aspartate aminotransferase (U/L) 22 74 (64) 63 53 (48) 77 38 (45) 0.007
Alanine aminotransferase (U/L) 21 49 (52) 60 39 (64) 68 27 (36) 0.192
ALP (U/L) 21 320 (188) 59 504 (544) 60 322 (292) 0.036
GGT (U/L) 19 131 (148) 53 191 (177) 52 97 (106) 0.005
Total bilirubin (mg/dL) 21 0.9 (1.1) 64 1.9 (2.6) 76 0.7 (0.6) 0.001
Direct bilirubin (mg/dL) 20 0.4 (0.6) 57 1.2 (2.1) 58 0.4 (0.5) 0.006
Creatinine (mg/dL) 22 1.1 (0.9) 64 2.4 (2.4) 74 1.2 (1.2) <0.001
Sodium (mmol/L) 21 132 (7) 62 132 (6) 75 136 (5) <0.001
Adjusted calcium (mmol/L) 20 2.25 (0.12) 57 2.47 (0.41) 62 2.36 (0.23) 0.018
Ferritin (mg/L) 8 8243 (9748) 26 2418 (4442) 11 712 (507) 0.010

parathyroid hormone level was lower than normal in
6 patients and at the lower part of the normal range in the
remaining 1 (mean, 6.9 pg/mL; range, 1–18.9 pg/mL; normal
= 15–70 pg/mL). Of the 14 patients with ESRD, 1 had ever
received iron supplement, and his serum ferritin level was
normal (normal = male, 26.6–377 mg/L; female, 3.0–151 mg/L).
Within 1 year of follow-up, 99 (60.4%) patients were
completely treated, 51 (31.1%) died, 10 (6.1%) were under
treatment, and 4 (2.4%) were lost. Twenty-three (14.0%)
patients died of TB. The cause of death was M. tuberculosis
septic shock with multiorgan failure in 15 and respiratory
failure due to extensive pulmonary damage in 8. In the
remaining 28 mortalities, septic shock due to pathogens other
than M. tuberculosis was responsible for 14 mortalities.
Another 7 patients died of respiratory failure, including ex-
tensive pulmonary damage in 4 and sputum impaction in 3.
Another 3 patients, including 2 patients with intracerebral
hemorrhage and 1 with brain tumor, died of severe increased
intracranial pressure. Three patients died of massive upper
gastrointestinal bleeding. The cause of death in the re-
maining 1 was acute myocardial infarction. Serum levels of
albumin, total bilirubin, and creatinine, and timing of anti-
tuberculosis treatment were independent prognostic factors
in the multivariate Cox regression analysis (Table 3). The
clinical findings that suggested the diagnosis and could lead
to early treatment of TB are listed in Table 4. In fact, 73.5%
of patients with early antituberculosis treatment had miliary
pattern on chest radiograph, compared with only 19.8% of
those with delayed treatment (p < 0.001). In the 164 patients
with disseminated TB, at least 1 diagnostic clue was present
in 116 (70.7%).
Specimens submitted for mycobacteriologic studies
and the results are summarized in Table 5. Data were
aggregated by patient and sample type. For example, if 1 of
the sputum samples was positive, then the results of sputum
culture were recorded as positive. The culture yield rates of
sputum (97.1%), lymph node (100%), and synovial fluid
(92.3%) were the highest. The most common biopsy sites
were bone marrow, followed by lymph node (Table 6). The
highest were from the testis (100%), joint (100%), and lymph
nodes (92.0%). The yield rate of histopathologic examination
for bone marrow was 44.7%. Fifteen bone marrow samples
were submitted for both mycobacterial culture and histo-
pathologic examination. Five were positive in both tests,
another 6 were culture-positive, and another 3 were histo-
pathologically positive. The diagnostic rate of the combined
mycobacterial culture and histopathologic examination for
TB was 93.3%.
The clinical isolates of M. tuberculosis were available
in 64 patients, including 9 HIV/AIDS patients, 25 with other
underlying comorbid condition, and 30 previously healthy
patients. Genotyping revealed that 47 (73.4%) isolates
belonged to 5 clusters and the remaining 17 (26.6%) isolates
had unique patterns. Beijing family was the largest cluster
and consisted of 27 (42.2%) clinical isolates (see Table 1).
DISCUSSION
Disseminated TB is a potentially lethal form of TB
resulting from massive lymphohematogenous dissemina-
tion of M. tuberculosis bacilli. The emergence of the
HIV/AIDS pandemic and widespread use of immunosup-
pressive drugs has change the epidemiology of disseminated
TB8,23. Impaired cell-mediated immunity underlies the
disease’s development. Clinical manifestations are nonspe-
cific, and typical chest radiographic findings may not be
seen11,17. Systematic data collection and reporting to study
the epidemiology and prognosis of disseminated tuberculosis
is lacking. In addition, the diagnostic approach has not been
standardized24.

42 n 2007 Lippincott Williams & Wilkins

Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Medicine  Volume 86, Number 1, January 2007 Disseminated Tuberculosis

TABLE 3. Survival Analysis for 164 Patients With Disseminated Tuberculosis

Hazard Ratio

No. of Patients (%) Mortality (%) Univariate Multivariate

Age (yr)
>65 67 (40.9) 26 (38.8) 1.61 (0.93–2.79)
65 97 (59.1) 25 (25.8)
Sex
Male 119 (72.6) 41 (34.5) 1.64 (0.83–3.28)
Female 45 (27.4) 10 (22.2)
Underlying comorbid condition
HIV/AIDS 23 (14.0) 8 (34.8) 1.98 (0.83–4.73) 2.49 (0.97–6.35)
Other comorbidities 64 (39.0) 29 (45.3) 2.97 (1.57–5.62) 1.06 (0.50–2.25)
No comorbidities 77 (47.0) 14 (18.2)
Smoking
Current smoker 59 (36.0) 19 (32.2) 1.23 (0.68–2.23) 1.43 (0.71–2.86)
Ex-smoker 13 (7.9) 7 (53.8) 2.38 (1.03–5.51) 2.18 (0.91–5.24)
Non-smoker 92 (56.1) 25 (27.2)
Pattern on chest image
Miliary 77 (47.0) 21 (27.3) 0.75 (0.42–1.32)
Not miliary 87 (53.0) 30 (34.5)
Hemoglobin (g/dL)
<10 72 (43.9) 30 (41.7) 2.09 (1.19–3.64) 1.21 (0.63–2.31)

10 92 (56.1) 21 (22.8)
Leukocyte (109/L)
>9000 48 (29.3) 15 (31.3) 0.82 (0.42–1.61)
5000–9000 68 (41.5) 17 (25.0) 0.59 (0.31–1.14)
<5000 48 (29.3) 19 (39.6)
Platelet (109/L)
<140 45 (27.6) 26 (57.8) 3.71 (2.14–6.43) 1.39 (0.69–2.89)

140 118 (72.4) 25 (21.2)
Albumin (g/dL)
<3.5 109 (74.7) 45 (41.3) 6.29 (1.96–20.26) 4.30 (1.28–14.46)

3.5 37 (25.3) 3 (8.1)
Total bilirubin (mg/dL)
>1.0 41 (25.5) 23 (56.1) 3.55 (2.03–6.20) 2.92 (1.40–6.07)
1.0 120 (74.5) 27 (22.5)
Creatinine (mg/dL)
>1.5 39 (24.4) 25 (64.1) 4.54 (2.61–7.91) 2.69 (1.34–5.40)
1.5 121 (75.6) 26 (21.5)
Sputum acid-fast smear
Positive 40 (24.4) 11 (27.5) 0.86 (0.44–1.67)
Negative 124 (75.6) 40 (32.3)
Resistance
MDR 5 (3.0) 2 (40.0) 1.95 (0.47–8.05)
Not MDR 16 (9.8) 5 (31.3) 1.11 (0.44–2.79)
No resistance 143 (87.2) 44 (30.8)
Clustering
Clustering, Beijing strain 30 (47.7) 8 (26.7) 0.71 (0.29–1.75)
Clustering, non-Beijing strain 20 (31.7) 10 (50.0) 0.32 (0.10–1.03)
Not clustering 13 (20.6) 5 (38.5)
Start anti-TB (d after initial visit)
>14 81 (49.4) 33 (40.7) 2.00 (1.13–3.55) 2.03 (1.07–3.89)
14 83 (50.6) 18 (21.7)

n 2007 Lippincott Williams & Wilkins 43

Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Wang et al Medicine  Volume 86, Number 1, January 2007

TABLE 4. Diagnostic Clues of Disseminated Tuberculosis

Disease Groups

All Patients (n = 164) HIV/AIDS Other Comorbidities No Comorbidities

No. (%) (n = 23) No. (%) (n = 64) No. (%) (n = 77) No. (%)
Miliary lung lesions 77 (47.0%) 16 (69.6%) 24 (37.5%) 37 (48.1%)
Ferritin
1000 mg/L 24 (14.6%) 7 (30.4%) 14 (21.9%) 3 (3.9%)
Infiltrative liver disease* 54 (32.9%) 7 (30.4%) 33 (51.6%) 14 (18.2%)
Adjusted calciumy >2.6 mmol/L 22 (13.4%) 0 12 (18.8%) 10 (13%)
Any 1 of the above clues 116 (70.7%) 19 (82.6%) 54 (84.4%) 43 (55.8%)
*Infiltrative liver disease was defined as ASP >440 U/L or GGT >106 U/L.
y
Adjusted calcium level was calculated by adding 0.2 mmol/L for every 1 g/dL decrease of serum albumin below 4 g/dL (ref. 25).

Of patients with culture-confirmed TB, 5.4% were Table 1), which partially contributed to its poorest outcome (in
disseminated. The percentages of patients with underlying univariate analysis). Factors leading to delayed diagnosis and
diseases were much lower than previously reported5, treatment in this group include the following: 1) Patients in
reflecting the high incidence of TB (62.4 per 100,000 this group had pleura, peritoneum, or pericardium as the most
population in 2003) and relatively low incidence of HIV/ common sites of extrapulmonary involvement (23.4%).
AIDS in Taiwan (1.3 per 10,000 population in 2004). The Because pleuro-pericardial effusions and ascites are common
clinical characteristics were very different among patients in such patients (with coexisting ESRD, malignancy, liver
with different underlying comorbid conditions. Factors asso- cirrhosis) even without concomitant TB, the diagnosis of TB
ciated with poor prognosis included albumin <3.5 g/dL, total in this group is difficult and often delayed. 2) Although the
bilirubin >1.0 mg/dL, creatinine >1.5 mg/dL, and delayed yields of pleural (55.6%) or peritoneal fluids (70.0%) culture
antituberculosis treatment. were high, the prolonged culture time offsets their usefulness.
Of the factors that were associated with prognosis, most 3) The high yields of pleural (61.5%) and peritoneal (85.7%)
(albumin, bilirubin, creatinine levels) reflect the severity of biopsy could be helpful in making a quick diagnosis, although
TB or underlying diseases. One (early treatment) directly the bleeding tendency in patients with ESRD or liver cirrhosis
reflects the importance of a timely, correct diagnosis. The
HIV/AIDS group had the highest rate of early treatment, while
the group with other underlying diseases had the lowest (see TABLE 6. Histopathologic Findings for 164 Patients With
Disseminated Tuberculosis
No of Patients % of Positive
TABLE 5. Results of Mycobacterial Culture for 164 Patients Specimen (Histology-Positive*/Studied) Specimens
With Disseminated Tuberculosis
Bone marrow 17/38 44.7
Culture No. of Patients % of Positive Lymph node 23/25 92.0
Specimen (Culture-Positive/Tested) Specimens Pleura 8/13 61.5
Blood* 15/39 38.5 Testis 5/5 100
Bone marrow 11/17 64.7 Soft tissue 15/18 83.3
Liver 3/5 60.0 Bone 11/14 78.6
Sputum 132/136 97.1 Joint 8/8 100
Pleural effusion 20/36 55.6 Prostate 3/4 75.0
Urine 24/34 70.6 Peritoneum 6/7 85.7
Ascites 21/30 70.0 Pericardium 0/1 0
Cerebrospinal fluid 15/28 53.6 Viscera 9/16 56.3
Lymph node 21/21 100 Liver 9/10 90.0
Synovial fluid 12/13 92.3 Lung 10/18 55.6
Tissue 27/44 61.4 Open biopsy 8/9 88.9
Stool 1/5 20.0 Transthoracic biopsy 2/3 66.7
Total 302/408 74.0 Transbronchial biopsy 1/10 10.0
Total 124/177 70.0
*Only 1 blood specimen was collected before the introduction of the
fluorometric BACTEC technique for mycobacterial culture, and it was *Histologic positivity was defined as the presence of caseating
culture positive for M. tuberculosis. granulomatous inflammation.

44 n 2007 Lippincott Williams & Wilkins

Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Medicine  Volume 86, Number 1, January 2007
often prohibits such procedures. 4) This group had the lowest
rate of miliary pattern (37.5%) on chest radiograph, which
lowered clinical suspicion of TB17. The group with HIV/
AIDS, on the contrary, had the highest rate of miliary TB
(nearly 70%), which helped early diagnosis and treatment.
5) Even when TB is highly suspected, the presence of renal
and/or hepatic insufficiency in these patients often prevents
the empirical use of antituberculosis chemotherapy.
Consistent with previous reports5, patients with dis-
seminated TB usually had several abnormal laboratory
findings, showing hypoalbuminemia, elevated ALP and
GGT, hyponatremia, and anemia (see Table 2). The laboratory
data reflect malnutrition, infiltrative liver disease, inappro-
priate secretion of antidiuretic hormone, and chronic illness,
respectively. Previous study of disseminated TB also showed
that 80% of patients had pathologic evidence of hepatic
involvement, and presented with elevated ALP and GGT with
only slight increases of hepatic transaminases and bilirubin23,
suggesting infiltrative liver disease such as TB, rather than
chronic active hepatitis, cirrhosis, or hepatic congestion13. We
also found that one-sixth of our patients had serum ferritin
>1,000 mg/L (see Table 4); a previous report also showed the
association of hyperferritinemia with miliary TB3. Though the
cause of hyperferritinemia was not certain, it most likely came
from chronic systemic infection3,27,32. Hypercalcemia was
detected in 13.4% of the patients with disseminated TB and
18.8% of those with underlying comorbid conditions other
than HIV/AIDS. The concomitant high serum phosphate and
low parathyroid hormone suggested a vitamin D-dependent
pathophysiology rather than malignancy or hyperparathyroid-
ism. Because none of our patients received vitamin D
supplementation, the vitamin D-dependent hypercalcemia
was most likely due to granulomatous disease6,10. The absence
of hypercalcemia in HIV/AIDS patients probably implied a
weakened host immunity and failure to mount a granuloma-
tous inflammation. We found that certain laboratory abnor-
malities often were associated with disseminated TB: elevated
ALP or GGT, serum ferritin >1,000 mg/L, and hypercalcemia.
Although these clinical clues are difficult to assess since the
specificity of all factors, except miliary lung lesions, is low,
they could raise the suspicion of disseminated TB.
Two previous reports15,29 comparing the clinical utility
of bone marrow biopsy and culture with blood cultures for
mycobacterial infection among HIV/AIDS patients showed
that the yields were similar and their results were highly
concordant (75%–83.7%), suggesting that the combined use
of both studies would provide the maximum diagnostic yield.
Because of the added cost, patient discomfort, and the risk of
needle stick to the operator during bone marrow studies,
routine blood culture was recommended during evaluations
for mycobacterial infections in HIV/AIDS patients15,29.
However, most of the disseminated mycobacterial diseases
in the 2 studies were due to Mycobacterium avium complex
(78.2%), rather than M. tuberculosis. In addition, the bone
marrow biopsy usually provided the first positive result in
the majority of these patients. In the first patient series that
we know of to assess the sensitivity of blood culture and
bone marrow culture for the diagnosis of disseminated TB,
n 2007 Lippincott Williams & Wilkins
Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Disseminated Tuberculosis
Crump et al5 found that mycobacterial blood culture was
as sensitive as bone marrow culture (58% vs. 54%), and
recommended performing mycobacterial blood culture in-
stead of bone marrow culture when disseminated TB was
suspected. However, histopathologic examination of bone
marrow was not assessed in that study. In the current series,
the sensitivity of bone marrow studies for disseminated TB
(mycobacterial culture: 64.7%, histopathologic examination:
44.7%, combined: 93.3%) was much higher than that of
mycobacterial blood culture (38.5%). The lower yield rate of
blood culture (38.5% vs. 58%) reflects the smaller proportion
of HIV/AIDS patients in the current study (14.0% vs. 46%). In
addition, bone marrow biopsy findings of caseating granulo-
matous inflammation characteristic of TB can be obtained
within days, allowing antituberculosis treatment to be started
early. The biopsy procedure is safer than liver, pleural, or
peritoneal biopsy for patients with renal or hepatic insuffi-
ciency and bleeding tendency. Thus, bone marrow biopsy can
be valuable in the early diagnosis of disseminated TB.
A major goal of molecular epidemiologic studies of
TB is to use molecular methods (DNA fingerprinting) to
distinguish between reactivation and recent transmission. The
fingerprints of each isolate are classified as clustered (that
is, sharing a fingerprint with another isolate in the study
sample) or unique. Most researchers interpret clusters to be
epidemiologically linked chains of recent transmitted disease,
and unique to be isolates that are cases of reactivation disease
resulting from ‘‘remote’’ M. tuberculosis infection19,21. Com-
patible with a previous study in the general TB population in
Taiwan which showed 24.8% clustered pattern among 113
resolved genotypes12, we found 22.7% of the resolved geno-
types had clustering patterns and accounted for about three-
fourths of the clinical isolates. To our knowledge, no studies
have suggested any certain genotype to be ‘‘dissemination-
prone.’’ Our findings imply that some genotypes are dominant,
and recent transmission may be important among patients with
disseminated TB in Taiwan. However, only 39% of our
isolates were genotyped, and we don’t have contact history to
support our speculation.
The current study was limited by its retrospective
nature; studies for disseminated TB were not routinely
performed. Thus the incidence might be underestimated.
Many of the laboratory studies listed in Table 2, as well as
the ‘‘diagnostic clues,’’ were not routinely performed in TB
patients with localized disease, especially in those with pure
pulmonary involvement. Therefore, the comparison between
disseminated disease and localized disease to assess the
diagnostic clues was difficult. Furthermore, in the absence of
postmortem examination, it is possible that only the most ill
patients were biopsied and cultured. Prospective studies are
necessary to clarify these issues.
In conclusion, 5.4% of the patients with culture-
confirmed TB had disseminated disease. Half of them had
underlying comorbid conditions. About three-fourths of
the M. tuberculosis isolates showed clustering patterns, sug-
gesting that recent transmission probably has a major con-
tribution to disseminated TB in Taiwan. Poor prognostic
factors include hypoalbuminemia, hyperbilirubinemia, renal
45
Wang et al
insufficiency, and delayed antituberculosis treatment. Clin-
ical findings suggestive of disseminated TB are present in
70.7% of all patients. Combined mycobacterial culture and
histopathologic examination of biopsy from bone marrow
were more sensitive and faster than mycobacterial blood
culture in diagnosing disseminated TB.
REFERENCES
1. Alvarez SZ. Hepatobiliary tuberculosis. J Gastroenterol Hepatol. 1998;
13:833–839.
2. Anonymous. Diagnostic Standards and Classification of Tuberculosis in
Adults and Children. This official statement of the American Thoracic
Society and the Centers for Disease Control and Prevention was adopted
by the ATS Board of Directors, July 1999. This statement was endorsed
by the Council of the Infectious Disease Society of America, September
1999. Am J Respir Crit Care Med. 2000;161:1376–1395.
3. Arranz Caso JA, Garcia Tena J, Llorens MM, Moreno R. High serum
ferritin concentration in an AIDS patient with miliary tuberculosis. Clin
Infect Dis. 1997;25:1263–1264.
4. Counsell SR, Tan JS, Dittus RS. Unsuspected pulmonary tuberculosis in
a community teaching hospital. Arch Intern Med. 1989;149:1274–1278.
5. Crump JA, Reller LB. Two decades of disseminated tuberculosis at a
university medical center: the expanding role of mycobacterial blood
culture. Clin Infect Dis. 2003;37:1037–1043.
6. Dusso AS, Kamimura S, Gallieni M, Zhong M, Negrea L, Shapiro S,
Slatopolsky E. gamma-Interferon-induced resistance to 1,25-(OH)2 D3
in human monocytes and macrophages: a mechanism for the hyper-
calcemia of various granulomatoses. J Clin Endocrinol Metab. 1997;82:
2222–2232.
7. Gelb AF, Leffler C, Brewin A, Mascatello V, Lyons HA. Miliary
tuberculosis. Am Rev Respir Dis. 1973;108:1327–1333.
8. Havlir DV, Barnes PF. Tuberculosis in patients with human immuno-
deficiency virus infection. N Engl J Med. 1999;340:367–373.
9. Heffner JE, Strange C, Sahn SA. The impact of respiratory failure on the
diagnosis of tuberculosis. Arch Intern Med. 1988;148:1103–1108.
10. Higgins RM, Cahn AP, Porter D, Richardson AJ, Mitchell RG, Hopkin
JM, Morris PJ. Mycobacterial infections after renal transplantation. Q J
Med. 1991;78:145–153.
11. Hussain SF, Irfan M, Abbasi M, Anwer SS, Davidson S, Haqqee R,
Khan JA, Islam M. Clinical characteristics of 110 miliary tuberculosis
patients from a low HIV prevalence country. Int J Tuberc Lung Dis.
2004;8:493–499.
12. Jou R, Chiang CY, Huang WL. Distribution of the Beijing family
genotypes of Mycobacterium tuberculosis in Taiwan. J Clin Microbiol.
2005;43:95–100.
13. Kamath PS. Clinical approach to the patient with abnormal liver test
results. Mayo Clin Proc. 1996;71:1089–1094. [quiz 1094–1095].
14. Kamerbeek J, Schouls L, Kolk A, van Agterveld M, van Soolingen D,
Kuijper S, Bunschoten A, Molhuizen H, Shaw R, Goyal M, van Embden
J. Simultaneous detection and strain differentiation of Mycobacterium
tuberculosis for diagnosis and epidemiology. J Clin Microbiol. 1997;
35:907–914.
15. Kilby JM, Marques MB, Jaye DL, Tabereaux PB, Reddy VB, Waites
46
Copyr ight © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Medicine  Volume 86, Number 1, January 2007
KB. The yield of bone marrow biopsy and culture compared with blood
culture in the evaluation of HIV-infected patients for mycobacterial and
fungal infections. Am J Med. 1998;104:123–128.
16. Kim JH, Langston AA, Gallis HA. Miliary tuberculosis: epidemiology,
clinical manifestations, diagnosis, and outcome. Rev Infect Dis. 1990;
12:583–590.
17. Long R, O’Connor R, Palayew M, Hershfield E, Manfreda J.
Disseminated tuberculosis with and without a miliary pattern on chest
radiograph: a clinical-pathologic-radiologic correlation. Int J Tuberc
Lung Dis. 1997;1:52–58.
18. Maartens G, Willcox PA, Benatar SR. Miliary tuberculosis: rapid
diagnosis, hematologic abnormalities, and outcome in 109 treated
adults. Am J Med. 1990;89:291–296.
19. McNabb SJ, Braden CR, Navin TR. DNA fingerprinting of Mycobac-
terium tuberculosis: lessons learned and implications for the future.
Emerg Infect Dis. 2002;8:1314–1319.
20. Munt PW. Miliary tuberculosis in the chemotherapy era: with a
clinical review in 69 American adults. Medicine (Baltimore). 1972;51:
139–155.
21. Murray M, Alland D. Methodological problems in the molecular
epidemiology of tuberculosis. Am J Epidemiol. 2002;155:565–571.
22. Sahn SA, Neff TA. Miliary tuberculosis. Am J Med. 1974;54:495–505.
23. Shafer RW, Kim DS, Weiss JP, Quale JM. Extrapulmonary tuberculosis
in patients with human immunodeficiency virus infection. Medicine
(Baltimore). 1991;70:384–397.
24. Sharma SK, Mohan A, Sharma A, Mitra DK. Miliary tuberculosis: new
insights into an old disease. Lancet Infect Dis. 2005;5:415–430.
25. Singer GG. Fluid and electrolyte management. In: Ahya SN, Flood K,
Paranjothi S, eds. The Washington Manual of Medical Therapeutics.
30th ed. Philadelphia: Lippincott Williams Wilkins; 2001:64.
26. Slavin RE, Walsh TJ, Pollack AD. Late generalized tuberculosis: a clinical
pathologic analysis and comparison of 100 cases in the preantibiotic and
antibiotic eras. Medicine (Baltimore). 1980;59:352–366.
27. Sunder-Plassmann G, Patruta SI, Horl WH. Pathobiology of the role of
iron in infection. Am J Kidney Dis. 1999;34:S25–S29.
28. Talavera W, Miranda R, Lessnau KKL, Klapholz A. Extrapulmonary
tuberculosis. In: Friedman LN, ed. Tuberculosis: Current Concepts and
Treatment. 2nd ed. Boca Raton, Fla.: CRC Press; 2001:518.
29. Talbot EA, Reller LB, Frothingham R. Bone marrow cultures for the
diagnosis of mycobacterial and fungal infections in patients infected
with the human immunodeficiency virus. Int J Tuberc Lung Dis. 1999;
3:908–912.
30. Wang JY, Hsueh PR, Lee LN, Liaw YS, Shau WY, Yang PC, Luh KT.
Mycobacterium tuberculosis inducing disseminated intravascular coag-
ulation. Thromb Haemost. 2005;93:729–734.
31. Wang JY, Lee LN, Chou CS, Huang CY, Wang SK, Lai HC, Hsueh PR,
Luh KT. Performance Assessment of a Nested-PCR Assay (the RAPID
BAP-MTB) and the BD ProbeTec ET System for Detection of
Mycobacterium tuberculosis in Clinical Specimens. J Clin Microbiol.
2004;42:4599–4603.
32. Weiss G. Pathogenesis and treatment of anaemia of chronic disease.
Blood Rev. 2002;16:87–96.
33. World Health Organization. WHO Report 2005: Global Tuberculosis
Control. Geneva: World Health Organization; 2005.
34. Yew WW, Lee J. Pathogenesis of cervical tuberculous lymphadenitis:
pathways to anatomic localization. Tuber Lung Dis. 1995;76:275–276.
n 2007 Lippincott Williams & Wilkins