e42 CID 2001:33 (15 September) Braden et al.

M A J O R A R T I C L E
Simultaneous Infection with Multiple Strains
of Mycobacterium tuberculosis
Christopher R. Braden,
1,a
Glenn P. Morlock,
1
Charles L. Woodley,
1
Kammy R. Johnson,
2,3,b
A. Craig Colombel,
3
M. Donald Cave,
4
Zhenhua Yang,
4,d
Sarah E. Valway,
1
Ida M. Onorato,
1,c
and Jack T. Crawford
1
1
Division of Tuberculosis Elimination, National Center for Human Immunodeciency Virus, Sexually Transmitted Disease (STD), and Tuberculosis
(TB) Prevention, and Division of Acquired Immune Deciency Syndrome, STD, and TB Laboratory Research, National Center for Infectious
Diseases, Centers for Disease Control and Prevention (CDC); and
2
Epidemic Intelligence Service, Division of Applied Public Health Training, CDC,
Atlanta;
3
Washington State Department of Health, Olympia and Seattle; and
4
Central Arkansas Veterans Health Care System, Little Rock
Drug-susceptible and drug-resistant isolates of Mycobacterium tuberculosis were recovered from 2 patients, 1
with isoniazid-resistant tuberculosis (patient 1) and another with multidrug-resistant tuberculosis (patient 2).
An investigation included patient interviews, record reviews, and genotyping of isolates. Both patients worked
in a medical-waste processing plant. Transmission from waste was responsible for at least the multidrug-
resistant infection. We found no evidence that specimens were switched or that cross-contamination of cultures
occurred. For patient 1, susceptible and isoniazid-resistant isolates, collected 15 days apart, had 21 and 19
restriction fragments containing IS6110, 18 of which were common to both. For patient 2, a single isolate
contained both drug-susceptible and multidrug-resistant colonies, demonstrating 10 and 11 different restriction
fragments, respectively. These observations indicate that simultaneous infections with multiple strains of M.
tuberculosis occur in immunocompetent hosts and may be responsible for conicting drug-susceptibilityresults,
though the circumstances of infections in these cases may have been unusual.
The prevailing model for tuberculosis pathogenesis is
based on exposure of a susceptible person to an in-
fectious quantum, an undened infectious dose of ba-
cilli dispersed into the air by an infectious source and
inhaled into alveoli of the new host [1]. Once infected,
the host is considered to have relative immunity to
additional M. tuberculosis infections; thus, if either pri-
Received 7 November 2000; revised 22 February 2001; electronically published
6 August 2001.
Financial support: Centers for Diseases Control and Prevention and National
Tuberculosis Genotyping and Surveillance Network (cooperative agreement).
Current afliations:
a
Division of Bacterial and Mycotic Diseases (C.R.B.),
b
Division
of Environmental Hazards and Health Effects (K.R.J.), and
c
Division of HIV/AIDS
Prevention (I.M.O.), CDC, Atlanta; and
d
Epidemiology Department, School of Public
Health, University of Michigan at Ann Arbor, Michigan (Z.Y.).
Reprints or correspondence: Dr. Christopher R. Braden, Centers for Disease
Control and Prevention, Mailstop A-38, 1600 Clifton Rd., Atlanta, GA 30333
(crb5@cdc.gov). Alternate corresponding author: Dr. Jack T. Crawford, Centers for
Disease Control and Prevention, Mailstop F-08, 1600 Clifton Rd., Atlanta, GA 30333
(jtc4@cdc.gov).
Clinical Infectious Diseases 2001; 33:e427
2001 by the Infectious Diseases Society of America. All rights reserved.
1058-4838/2001/3306-00E1$03.00
mary or reactivation tuberculosis disease develops, the
original infecting single strain of M. tuberculosis is con-
sidered responsible for tuberculosis disease at all ana-
tomic sites and any potential relapse of disease after
treatment. The degree of immunity to a second M.
tuberculosis infection is not known, however, and si-
multaneous infection by multiple strains or reinfection
by a second M. tuberculosis strain may be responsible
for a portion of tuberculosis cases.
To what extent simultaneous infections or reinfection
with M. tuberculosis is responsible for primary, reacti-
vation, or relapse tuberculosis has been the subject of
controversy. One leading tuberculosis epidemiologist
has argued convincingly that disease due to reinfection
is a relatively rare event [2], and another argued that
in areas or times with a very high risk of infection,
reinfection plays a predominant part in the pathogen-
esis of tuberculosis in adults [3]. A recent study from
an area of South Africa where tuberculosis is endemic
revealed that 12 of 16 patients had exogenous reinfec-

b
y

g
u
e
s
t

o
n

A
p
r
i
l

1
4
,

2
0
1
2
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

M. tuberculosis Multistrain Infection CID 2001:33 (15 September) e43
Figure 1. Time line of onset of cough, treatment-start dates, and
specimen collections for patients 1 and 2.
tion that was responsible for relapse of tuberculosis after cu-
rative treatment [4]. More recently, this topic has taken on
greater relevance in the era of HIV infection because of the
potential loss of immune protection against reinfection [5].
Infection with multiple M. tuberculosis strains has clinical
and programmatic implications. Current recommendations for
treatment of latent tuberculosis infection [6] and use of BCG
vaccine are predicated on a solitary strain theory, whereby ben-
ecial effects of a course of therapy for latent tuberculosis in-
fection or vaccination last for life [7].
In this report, we describe 2 HIV-negative patients with tu-
berculosis caused by multiple M. tuberculosis strains. One pa-
tient had 2 related subpopulations of an M. tuberculosis strain
in 2 clinical specimens collected 15 days apart: 1 isoniazid-
susceptible, the other isoniazid-resistant. The second patient
had 2 distinct strains of M. tuberculosis in 1 clinical specimen:
1 multidrug-resistant (MDR) and the other fully susceptible.
Both patients worked in a medical-waste treatment facility at
the time of the diagnoses. In addition to other types of medical
and laboratory waste, this facility received and processedculture
material that had not been decontaminated by autoclave or
other microbicidal processes. Results of epidemiological, en-
vironmental, and laboratory investigations strongly suggest that
MDR tuberculosis in the second patient was due to exposures
to infectious aerosols in the workplace [8]. The sources of the
other M. tuberculosis infections in these 2 patients could not
be determined.
METHODS
To determine potential sources of tuberculosis infection, the
patients were interviewed and their medical records were re-
viewed. In order to investigate whether mislabeling or labo-
ratory cross-contamination could be responsible for the un-
expected mycobacteriology results, all hospitals, clinics, and
clinical laboratories involved in the care of the patients and in
mycobacterial testing of specimens were identied from a re-
view of medical records and interviews with patients and hos-
pital and laboratory staff members. Hospital infection-control
records were reviewed to identify other tuberculosis patients in
the facilities at the same time as the 2 case-patients, and clinical
laboratory mycobacteriology procedures and records were re-
viewed to identify potential sources of contamination in clinical
laboratories. Potential sources of contamination included any
acid-fast bacilli (AFB) culturepositive specimen or isolate re-
ceived in the laboratory within 2 weeks before or after the date
of receipt of the specimen in question.
Multiple isolates from each case and any potential source
isolate for laboratory contamination underwent DNA nger-
print analysis by IS6110 restriction fragment length polymor-
phism (RFLP), with use of standard methods [9]. Drug sus-
ceptibility testing was performed with both the BACTEC
(Becton Dickinson) and agar proportion methods. Gene mu-
tations responsible for antituberculosis drug resistance were
identied by automated sequencing of PCR products.
RESULTS
Patient descriptions. Patient 1 was a 52-year-old white non-
Hispanic woman who was born in West Virginia, moved to
Washington State at age 14, and lived in the same county for
20 years. She had no history of chronic illness, foreign travel,
substance abuse, incarceration, stays in long-term-carefacilities,
employment as a health care worker, or BCG vaccination. She
had no known exposure to M. tuberculosis and had never un-
dergone a tuberculin skin test. In 1992, she began working at
a medical-waste treatment facility. She developed a cough in
December 1996 and presented with fatigue and shortness of
breath in March 1997 (gure 1). Chest radiographs revealed
right apical lung densities, and by late April 1997, right upper
and left lower lobe inltrates with cavities were present. Three
sputum samples collected over 5 days were AFB smearpositive
and yielded M. tuberculosis susceptible to all drugs tested (table
1).
The next sputum specimen, collected 15 days later, yielded
M. tuberculosis that was resistant to isoniazid. All of the sub-
sequent 6 specimens collected over the ensuing 32 days were
culture-positive, with isolates resistant to isoniazid. Sputum
specimens thereafter were culture-negative. The patient was
treated with isoniazid, rifampin, pyrazinamide, and ethambu-
tol, beginning on the fourth day of initial sputum collection
and 16 days before collection of the sputum specimen yielding
the isoniazid-resistant isolate. Isoniazid was withdrawnafter sus-
ceptibility test results documenting isoniazid resistance were con-
rmed. She completed 1 year of therapy without complications.
Patient 2 was a previously healthy 33-year-old white non-
Hispanic man who was born in Washington State and had lived

b
y

g
u
e
s
t

o
n

A
p
r
i
l

1
4
,

2
0
1
2
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

e44 CID 2001:33 (15 September) Braden et al.
Table 1. Percentage of resistance, to antibiotics, as determined
by the agar proportions method, of M. tuberculosis isolates from
sputum samples.
Drug (mg/mL)
Percentage of isolate resistance,
per sample-collection day
a
Patient 1 Patient 2
1 5 20 29 1 2 41 52
INH (0.2) 0 0 25 50 100 100 0 100
INH (1.0) 0 0 0 0 100 100 0 100
RIF (1.0) 0 0 0 0 100 100 0 100
EMB (10.0) 0 0 0 0 0 0 0 0
STREP (2.0) 0 0 0 0 50 50 0 50
PZA (25) 0 0 0 0 0 0 0 0
NOTE. INH, isoniazid; RIF, rifampin; EMB, ethambutol; STREP, strepto-
mycin; PZA, pyrazinamide.
a
First specimen collection day and days thereafter; not all subsequent spec-
imens are included.
in the same county since 1990. He denied foreign travel but
had been in a residential drug treatment unit for several weeks
in 1988 and 1989. He had not received BCG vaccination or
undergone tuberculin skin tests. He began working at the same
medical-waste treatment facility in April 1995 and was exposed
to patient 1 during her infectious period. He had no other
known tuberculosis exposures. He had a cough that began in
July 1997 (gure 1). A chest radiograph in late August revealed
bilateral apical densities, and therapy with isoniazid, rifampin,
ethambutol, and pyrazinamide was begun. Two sputum spec-
imens were AFB smearnegative but yielded M. tuberculosis
resistant to isoniazid, rifampin, and streptomycin (table 1).
Because of these results, therapy was changed to administration
of ethambutol, levooxacin, para-aminosalicylic acid, and
cycloserine.
The next culture-positive sputumspecimen, collected 41 days
after the rst specimen, yielded an isolate susceptible to all
drugs tested. However, specimens collected at 52 and 56 days
yielded isolates resistant to isoniazid, rifampin, and strepto-
mycin. He completed 2 years of therapy and remained clinically
well.
Investigation of mislabeling or laboratory cross-contami-
nation of specimens. Between these 2 patients, initial care
was received at 1 clinic, 2 hospitals, and 4 clinical laboratories.
Records and procedures at all facilities were reviewed. Myco-
bacteriological testing included uorescence acid-fast stains and
cultures on Lowenstein-Jensen medium, on 7H10 or 7H11 agar,
and in the BACTEC TB-460 system. Identication was accom-
plished by DNA probe and biochemical testing, and suscepti-
bility testing was done with the BACTEC and 7H11 agar
plateproportion methods.
For patient 1, the identication of an isoniazid-resistant iso-
late within 15 days of the collection of a specimen yielding a
fully susceptible isolate, while the patient was receiving 4 an-
tituberculosis drugs, prompted an investigation of both spec-
imens to identify mislabeling or laboratory cross-contamina-
tion. Hospital and clinic records failed to identify another
tuberculosis patient receiving care at the same time whose spec-
imen may have been mislabeled as being from patient 1. Lab-
oratory records revealed no potential source of cross-contam-
ination in laboratories. Subsequent retesting of the isolates from
patient 1 conrmed that the rst 2 isolates were isoniazid-
susceptible and that multiple subsequent isolates were isonia-
zid-resistant.
For patient 2, the rst MDR isolate was suspected of being
a contaminant because the source of infection was thought to
be patient 1 and because the resistance pattern identied is rare
in the area. He received care at 1 clinic, and his specimens were
tested at 2 laboratories. There was no other patient with tu-
berculosis with whom a specimen could have been switched.
A possible source of laboratory cross-contamination was iden-
tied: an M. avium complex isolate resistant to isoniazid, ri-
fampin, and streptomycin, among other drugs. However, DNA
probes specic for M. avium complex (Gen-Probe) did not
hybridize to nucleic acid of AFB in the original culture bottle
or in drug-susceptibility culture bottles for the specimen from
patient 2. A high degree of hybridization was observed for DNA
probes for M. tuberculosis complex in all these bottles. We con-
cluded that contamination with the M. avium complex isolate
was not responsible for the observed resistance pattern. No
other potential sources of contamination were identied in
laboratories.
Three subsequent isolates from patient 2 with the same re-
sistance pattern conrmed the initial nding. Another inves-
tigation was performed as a result of identifying the third isolate
from patient 2, which was collected 41 days after the rst and
was susceptible to all drugs. Another fully susceptible isolate
from the state public health laboratory (where susceptibility
testing was performed) was identied as a potential source of
contamination; however, this isolate had a DNA ngerprint
pattern distinct from that of the susceptible isolate from patient
2.
DNA ngerprint analysis. DNA ngerprint patterns of
isolates from patient 1 and patient 2 were distinct from one
another (gures 2 and 3). DNA ngerprint patterns of se-
quential isolates from patient 1 revealed a signicant change
in pattern coincident with identication of isoniazid resistance.
The 2 fully susceptible isolates had identical DNA ngerprint
patterns with 21 hybridizing fragments; 3 subsequent isoniazid-
resistant isolates had identical patterns with 19 hybridizing frag-
ments, 18 of which were common to the patterns of the sus-
ceptible isolates (gure 2).
DNA ngerprint analysis of 4 sequential isolates frompatient
2 yielded 3 distinct patterns (gure 3). The rst isolate had a

b
y

g
u
e
s
t

o
n

A
p
r
i
l

1
4
,

2
0
1
2
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

M. tuberculosis Multistrain Infection CID 2001:33 (15 September) e45
Figure 2. DNA ngerprint analyses of isolates from patient 1. S:
molecular weight standard; lane 1: isoniazid-resistant isolate with 19
fragments; lane 2: isoniazid-susceptible isolate with 21 fragments. Arrows
indicate the difference of fragments in lane 2 compared to lane 1.
Figure 3. DNA ngerprint analyses of isolates from patient 2. S:
molecular weight standard; lanes 19: ngerprints of individual colonies
from second isolate, showing 2 separate strains (lanes 15, 10 fragments,
drug-susceptible; lanes 69, 11 fragments, multidrug-resistant); lane 10:
ngerprint of mixed DNA from individual colonies from second isolate
with 21-band pattern; lane 11: ngerprint of whole second isolate with
same 21-band pattern; lane 12: ngerprint of rst isolate with 11-band
pattern of multidrug-resistant strain.
DNA ngerprint pattern with 11 hybridizing fragments, the
second isolate had a pattern with 21 hybridizing fragments, and
the third isolate, which was fully susceptible, had a ngerprint
pattern with 10 fragments. Subsequent MDR isolates had the
11-fragment pattern identical to that of the rst isolate. The
10- and 11-fragment patterns shared no fragment sizes in com-
mon, but taken together, they accounted for all fragments in
the 21-fragment pattern (gure 3).
The second isolate with a 21-fragment pattern was cultured
on solid media, and 10 individual colonies were picked from
the plate and grown separately for DNA ngerprint analysis.
One of these colony cultures was lost to contamination. The
remaining 9 colony cultures revealed 4 with the 11-fragment
pattern and 5 with the 10-fragment pattern (gure 3). Com-
bining DNA from colonies with these 2 patterns yielded a n-
gerprint pattern identical to the 21-fragment pattern seen for
this isolate as a whole.
Resistance mutations. To determine if transposition of
insertion sequence IS6110 was causally associated with devel-
opment of isoniazid resistance in the isolate from patient 1, we
identied the mutation responsible for isoniazid resistance in
her third isolate (table 1). An inhA mutation CrT was iden-
tied in the presumed promoter region. For patient 2, the MDR
isolate had the common katG Ser315rThr mutation, the rarer
rpoB mutation Asp516rVal, and a silent (no amino acid
change) pncA mutation Ala38rAla.
DISCUSSION
We have identied 2 patients simultaneously infected with M.
tuberculosis isolates demonstrating distinct characteristics. For
patient 1, the isolates identied had different susceptibilities to
isoniazid and IS6110 RFLP patterns that differed by the ad-
dition or movement of 3 hybridizing fragments. Whether these
differences constitute different strains is a matter of judgement.
Since these 2 isolates were from the same patient and their
DNA ngerprint patterns exactly matched for 18 of at most 21
fragments, we considered them as subpopulations with the
same clonal origin. We do not believe that isoniazid resistance
developed in patient 1 from selective pressure of drug therapy.
Therapy with isoniazid, rifampin, ethambutol, and pyrazin-
amide was started 16 days prior to collection of the rst iso-
niazid-resistant isolate, and the patient reportedly took all drugs

b
y

g
u
e
s
t

o
n

A
p
r
i
l

1
4
,

2
0
1
2
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

e46 CID 2001:33 (15 September) Braden et al.
prescribed and did well throughout her clinical course. How-
ever, she was not receiving directly observed therapy.
We do not believe that the unexpected susceptibility test
results were due to a switch in specimens or cross-contami-
nation of cultures in laboratories; no other samples to switch
or contaminate were identied, and multiple samples from this
patient eventually conrmed the ndings. The mutation likely
responsible for isoniazid resistance was a point mutation, not
the transposition of IS6110 into an isoniazid-resistance gene.
Whether patient 1 was originally exposed and infected with
these 2 subpopulations or they diverged during development
of infection and disease within this patient is impossible to
determine.
For patient 2, we identied 2 distinct strains of M. tuberculosis
within the same specimen. Laboratory cross-contamination of
the culture was not identied as an explanation, and separate
sputum samples each yielded the different strains. Thus, in
some separate sputum samples, only one or the other of the 2
strains was detected, whereas in one sample, the 2 strains were
detected in approximately equal proportions. An hypothesis to
account for this observation is that the 2 strains inhabited
separate anatomic sites in the lung. These sites then contributed
to sputum samples separately or in combination over time. As
for patient 1, it is impossible to determine if patient 2 was
exposed and infected simultaneously or in succession with the
2 different strains.
Previous reports have documented infections with multiple
M. tuberculosis strains. Using phage typing, Bates et al. [10]
identied 3 patients in whom tubercle bacilli with different
phage types were isolated from different anatomic sites. In a
study of 233 isolates from Eskimo patients, 33 had mixed
phage types, 9 of which were determined to be major changes
involving as many as 7 phages and differences in drug suscep-
tibility [11]. Raleigh et al. [12] studied pretreatment and relapse
M. tuberculosis isolates from 26 patients; in 9 patients they
found a major change in phage type. In yet another study
involving an outbreak of tuberculosis in a shelter for homeless
persons in Boston, 25 patients shared the same strain, identied
by phage typing and resistance to isoniazid and streptomycin.
Seven of the 25 patients shared this strain, despite documen-
tation of tuberculosis infection or disease prior to their ex-
posure to the source case in the outbreak, a nding suggesting
they were reinfected by the outbreak strain [13].
The advent of DNA ngerprint analysis of M. tuberculosis
has vastly improved the ability to distinguish strains, and several
reports have identied different strains associated with devel-
opment of drug resistance or relapse of disease in both HIV-
infected and HIV-noninfected patients [5, 1417]. Still, the
proportion of second-episode or relapse tuberculosis cases due
to reinfection is unknown, and therefore, the relative protection
offered by natural infection remains a mystery.
Given that reinfection may occur, the failings of immune
protection from BCG vaccination are understandable and have
important implications for the current efforts to develop new
and better vaccines. However, the determination of relative
immune protection by natural infection vis-a`-vis reinfection is
theoretically complicated by simultaneous infection with mul-
tiple strains of M. tuberculosis, as implicated in the cases pre-
sented above. Multiple strains may infect simultaneously or
over a short period of time, escaping immune protection in a
immune-na ve host. Multiple episodes of tuberculosis or disease
affecting multiple sites in the body may then occur, owing to
some potential selective advantage of one strain over another,
such as antituberculosis drug resistance or a general propensity
to reactivate. Thus, multistrain infections may be responsible
for observations attributed to reinfection.
Infection with multiple M. tuberculosis strains may be very
difcult to detect. For instance, 1 or multiple strains may be
sequestered at a different anatomic location at the time of test-
ing and not be present in the sample provided, or the strains
may be such a small proportion of the entire bacillary popu-
lation of the sample that their identity is impossible to
determine.
In addition, the ability of laboratory testing to discriminate
mixed strains is very limited. The only routine clinical labo-
ratory test that has ability to discriminate among some M.
tuberculosis strains is drug-susceptibility testing, given that
strains involved have differing drug-susceptibility proles to
rst-line antituberculosis drugs. One may suspect multiple
strains if differing drug-susceptibility proles are obtained for
clinical specimens from 2 different anatomic sites or from one
site over time. Drug-susceptibility testing cannot differentiate
mixed strains within a single sample unless individual colonies
from the isolate are tested.
DNA ngerprint analysis provides a much more discrimi-
nating tool, but the DNA ngerprint pattern for a sample con-
taining a mix of multiple strains would most likely represent
the predominant strain or a composite pattern for multiple
strains, as was seen with the 21-band pattern for patient 2. To
distinguish among multiple strains in a single sample, nger-
print analysis of individual colonies from the isolate would be
required, as shown in gure 2.
Both patients in this report worked in a medical-waste pro-
cessing facility where there was a potential for unprotected
exposure to aerosols of infectious wastes, including nondecon-
taminated M. tuberculosis cultures [8]. Though it is not possible
to determine denitively the source of all their infections, the
possibility exists that they were infected with multiple strains
via this unusual exposure. Therefore, our observations con-
cerning these 2 patients may be due to the unique epidemio-
logical circumstances of their workplace and may not represent

b
y

g
u
e
s
t

o
n

A
p
r
i
l

1
4
,

2
0
1
2
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

M. tuberculosis Multistrain Infection CID 2001:33 (15 September) e47
risk for multiple-strain infections among tuberculosis patients
in general.
However, these observations and past experience with phage
typing and DNA ngerprinting would indicate that multiple-
strain infections may be responsible for conicting drug-sus-
ceptibility results. Especially in areas where the incidence of
tuberculosis is high, exposures to multiple strains may occur.
When conicting drug-susceptibility test results are observed,
accuracy of tests should be assessed rst, along with the po-
tential exchange of patients samples and laboratory cross-con-
tamination of cultures, as is described in this investigation. If
the explanation is not found with these preliminary investi-
gations, then additional testing by means of DNAngerprinting
may be sought to conrm infection with multiple strains.
Acknowledgments
We are indebted to Dona Osmond and the State of Wash-
ington Public Health Laboratory, for thoughtful analysis of clin-
ical laboratory results and the laboratorys role as a central
repository for M. tuberculosis isolates in this study.
References
1. Stead WW. Unied concept of tuberculosis in man. Bull Int Union
Tuberc 1974; 49:3446.
2. Stead WW. Pathogenesis of a rst episode of chronic pulmonary tu-
berculosis in man: recrudescence of residuals of the primary infection
or exogenous reinfection? Am Rev Respir Dis 1967; 95:72945.
3. Styblo K. Epidemiology of Tuberculosis. In: Selected papers, vol. 24.
The Hague, The Netherlands: Royal Netherlands Tuberculosis Asso-
ciation, 1991.
4. van Rie A, Warren R, Richardson M, et al. Exogenous reinfection as
a cause of recurrent tuberculosis after curative treatment. N Engl J
Med 1999; 341:11749.
5. Small PM, Shafer RW, Hopewell PC, et al. Exogenous reinfection with
multidrug-resistant Mycobacterium tuberculosis in patients with ad-
vanced HIV infection. N Engl J Med 1993; 328:113744.
6. American Thoracic Society. Targeted tuberculin testing and treatment
of latent tuberculosis infection. Am J Respir Crit Care Med 2000; 161:
S22147.
7. Dowdle WR, Hopkins DR, eds. The eradication of infectious diseases.
Chichester, United Kingdom: John Wiley & Sons, 1998.
8. Johnson KR, Braden CR, Cairns KL, et al. Transmission of Mycobac-
terium tuberculosis from medical waste. JAMA 2000; 284:16838.
9. Van Embden JDA, Cave MD, Crawford JT, et al. Strain identication
of Mycobacterium tuberculosis by DNA ngerprinting: recommenda-
tions for a standard methodology. J Clin Microbiol 1993; 31:4069.
10. Bates JH, Stead WW, Rado TA. Phage type of tubercle bacilli isolated
from patients with two or more sites of organ involvement. Am Rev
Respir Dis 1976; 114:3538.
11. Mankiewicz E, Liivak M. Phage types of Mycobacterium tuberculosis in
cultures isolated from Eskimo patients. Am Rev Respir Dis 1975; 111:
30712.
12. Raleigh JW, Wichelhausen RH, Rado TA, et al. Evidence for infection
by two strains of Mycobacterium tuberculosis in pulmonary tuberculosis:
report of 9 cases. Am Rev Respir Dis 1975; 112:497503.
13. Nardell E, McInnis B, Thomas B, at al. Exogenous reinfection with
tuberculosis in a shelter for the homeless. N Engl J Med 1986; 315:
15705.
14. Shafer RW, Singh SP, Larkin C, at al. Exogenous reinfection with mul-
tidrug-resistant Mycobacterium tuberculosis in an immunocompetent
patient. Tuber Lung Dis 1995; 76:5757.
15. Godfrey-Faussett P, Githui W, Batchelor B, et al. Recurrence of HIV-
related tuberculosis in an endemic area may be due to relapse or
reinfection. Tuber Lung Dis 1994; 75:199202.
16. Horn DL, Hewlett D, Hass WH, et al. Superinfection with rifampin-
isoniazid-streptomycin-ethambutol (RISE)resistant tuberculosis in
three patients with AIDS: conrmation by polymerase chain reaction
ngerprinting. Ann Intern Med 1994; 121:1156.
17. Theisen A, Reichel C, Rusch-Gerdes S, et al. Mixed-strain infection
with a drug-sensitive and multidrug-resistant strain of Mycobacterium
tuberculosis. Lancet 1995; 345:15123.

b
y

g
u
e
s
t

o
n

A
p
r
i
l

1
4
,

2
0
1
2
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m