Embed Size (px)
Citation preview
JOURNAL OF CLINICAL MICROBIOLOGY, May 2011, p. 1943–1950 Vol. 49, No. 50095-1137/11/$12.00 doi:10.1128/JCM.02299-10Copyright © 2011, American Society for Microbiology. All Rights Reserved.
Investigating Transmission of Mycobacterium bovis in the UnitedKingdom in 2005 to 2008�
Sema Mandal,1† Louise Bradshaw,1 Laura F. Anderson,1 Tim Brown,2 Jason T. Evans,3Francis Drobniewski,2 Grace Smith,3 John G. Magee,4 Anne Barrett,4
Oliver Blatchford,5 Ian F. Laurenson,6 Amie-Louise Seagar,6Michael Ruddy,7 P. Lewis White,7 Richard Myers,8
Peter Hawkey,3,9 and Ibrahim Abubakar1*Tuberculosis Section, Health Protection Agency Centre for Infections, London, United Kingdom1; National Mycobacterium
Reference Laboratory, Health Protection Agency Centre for Infections, London, United Kingdom2; Regional Centre forMycobacteriology, Health Protection Agency, Birmingham, United Kingdom3; Regional Centre for Mycobacteriology,
Health Protection Agency, Newcastle, United Kingdom4; Health Protection Scotland, Glasgow, United Kingdom5;Scottish Mycobacteria Reference Laboratory, Edinburgh, United Kingdom6; Wales Centre for Mycobacteria,
Cardiff, United Kingdom7; Bioinformatics Unit, Health Protection Agency Centre forInfections, London, United Kingdom8; and University of
Birmingham, Birmingham, United Kingdom9
Received 14 November 2010/Returned for modification 1 February 2011/Accepted 7 March 2011
Due to an increase in bovine tuberculosis in cattle in the United Kingdom, we investigated the characteristicsof Mycobacterium bovis infection in humans and assessed whether extensive transmission of M. bovis betweenhumans has occurred. A cross-sectional study linking demographic, clinical, and DNA fingerprinting (using15-locus mycobacterial interspersed repetitive-unit–variable-number tandem-repeat [MIRU-VNTR] typing)data on cases reported between 2005 and 2008 was undertaken. A total of 129 cases of M. bovis infection inhumans were reported over the period, with a decrease in annual incidence from 0.065 to 0.047 cases per100,000 persons. Most patients were born pre-1960, before widespread pasteurization was introduced (73%),were of white ethnicity (83%), and were born in the United Kingdom (76%). A total of 102 patients (79%) hadMIRU-VNTR typing data. A total of 31 of 69 complete MIRU-VNTR profiles formed eight distinct clusters. Theoverall clustering proportion determined using the n � 1 method was 33%. The largest cluster, comprising 12cases, was indistinguishable from a previously reported West Midlands outbreak strain cluster and includedthose cases. This cluster was heterogeneous, having characteristics supporting recent zoonotic and human-to-human transmission as well as reactivation of latent disease. Seven other, smaller clusters identified haddemographics supporting recrudescence rather than recent infection. A total of 33 patients had incompleteMIRU-VNTR profiles, of which 11 may have yielded 2 to 6 further small clusters if typed to completion. Theincidence of M. bovis in humans in the United Kingdom remains low, and the epidemiology is predominantlythat of reactivated disease.
The incidence of bovine tuberculosis (TB) increased duringthe 1990s (4), particularly affecting southwest England andWales (5). Despite concerns about the risk to human health, noincrease in Mycobacterium bovis infection in humans was ob-served between 1993 and 2003 (14). A recent report of zoono-sis in the United Kingdom, however, concluded that 18 of 35strains isolated from humans had genotypes associated withcattle in the United Kingdom and that this was suggestive ofdomestic acquisition of disease (13).
The main mode of transmission of M. bovis to humans isbelieved to be from cattle via consumption of unpasteurizeddairy products (9). Less frequently, aerosol inhalation follow-ing close contact with infectious cattle, e.g., on farms or in
abattoirs (9), and from other animals, including householdpets, has been described (24). Person-to-person transmissionof M. bovis, although infrequent (9), has been reported world-wide (7, 21, 23). In the United Kingdom, there have now beentwo such clusters, one with intrafamilial spread in Gloucester(26) and a larger outbreak consisting of six persons in the WestMidlands (8). Nonetheless, M. bovis infection in humans isstill comparatively rare and represented only 0.5% (24/5,014) of culture-confirmed tuberculosis cases in the UnitedKingdom in 2007 (11). Although M. bovis infection is notgenerally considered a human public health threat in theUnited Kingdom, recent evidence of possible shared straintypes between human and cattle M. bovis isolates (13), clus-ters involving human-to-human transmission (8, 26), thepossible transmission of M. bovis from a pet to its owner (17,24), and an observed increase in M. bovis infection in non-bovine species (15) warrant further molecular and epidemi-ological analysis of M. bovis infection in humans in theUnited Kingdom.
Mycobacterial interspersed repetitive-unit–variable-number
* Corresponding author. Mailing address: Tuberculosis Section, Re-spiratory Diseases Department, Health Protection Agency Centre forInfections, 61 Colindale Avenue, London NW9 5EQ, England, UnitedKingdom. Phone: 44(0)20 8327 6427. Fax: 44(0)20 8327 6112. E-mail:[email protected].
† Present address: 1600 Clifton Rd., MS C09, Atlanta, GA 30333.� Published ahead of print on 23 March 2011.
1943
tandem-repeat (MIRU-VNTR) typing can be used to success-fully distinguish between M. bovis isolates, and when appropri-ate loci are selected, this approach may have a discriminatorypower comparable to that of other genotyping methods (1, 12,22, 25). This method can therefore be used to confirm conven-tional epidemiological links, trace transmission routes, andidentify sources of infection. We describe the characteristics ofM. bovis infection in humans reported in the United Kingdomfrom 2005 to 2008, inclusive, and investigate whether extensivehuman-to-human transmission of M. bovis has occurred.
MATERIALS AND METHODS
Characteristics of M. bovis infection in humans in the United Kingdom. Thestudy population comprised humans infected with M. bovis as reported throughthe United Kingdom Mycobacterial Surveillance Network (MycobNet) between2005 and 2008. A case was defined as an instance of an individual who wasculture positive for M. bovis. Epidemiological data on sex, age at diagnosis,ethnicity, country of birth, and region of residence, clinical data on smear pos-itivity and site of disease, and exposure data on contact with a human tubercu-losis patient, contact with cattle and wild animals, occupational exposure toinfected animals, travel abroad, and consumption of unpasteurized dairy prod-ucts were obtained from the Health Protection Agency (HPA) enhanced M. bovissurveillance database. Where epidemiological data on cases were incomplete, theHealth Protection Unit (HPU) was contacted to complete missing fields. De-scriptive statistics on demographic and clinical characteristics of all cases werecalculated.
Investigation of transmission of M. bovis in humans in the United Kingdom.To specifically investigate epidemiological evidence of transmission in the UnitedKingdom, patients born abroad and patients born before widespread pasteuri-zation, i.e., before 1960, were excluded from further analysis. The remainingpatients (subset A) from the original study population were more likely to haveprimary disease acquired in the United Kingdom. Figure 1 illustrates the flow ofdata for epidemiological and molecular analyses of the subsets of interest.
M. bovis cultures were identified using the GenoType MTBC test (Hain Life-science, Nehren, Germany). This PCR/reverse-hybridization assay differentiatesmembers of the Mycobacterium tuberculosis complex on the basis of specificsingle-nucleotide polymorphisms within gyrB and the presence or absence ofRD1 (20). Strain-typing data on all cases were analyzed for evidence of cluster-ing. The HPA (United Kingdom) employs MIRU-VNTR markers, typing 15 loci,including 10 MIRUs (MIRUs 2, 10, 16, 20, 23, 24, 26, 27, 39, and 40) and fiveexact tandem repeats (ETR; A to E) (10). A cluster was defined as two or moreisolates that were indistinguishable by MIRU-VNTR typing. Strain typingdata from Northern Ireland and for 15 isolates from English regions were notavailable.
MIRU-VNTR results were entered into BioNumerics version 5 (AppliedMaths, St-Martin-Latem, Belgium). The DNA fingerprinting results were ana-lyzed using the categorical coefficient which assumes no ordered relationshipbetween MIRU-VNTR loci. The proportion of clustered cases was calculatedusing the n � 1 method (18). We compared unique (subset B) and clustered(subset C) cases with complete MIRU-VNTR profiles in a minimum data set ofdemographic and clinical characteristics. Clustered isolates (subsets C1 and C2)were considered for further risk factor investigation. The data were analyzedusing Microsoft Excel 2003 and SAS 9.2. The chi-square or Fisher exact test wasused to compare proportions.
FIG. 1. Diagram of case flow through epidemiological and molecular analyses.
1944 MANDAL ET AL. J. CLIN. MICROBIOL.
RESULTS
Characteristics of all cases of M. bovis infection in humansin the United Kingdom. A total of 129 M. bovis cases wereidentified between 1 January 2005 and 31 December 2008 inthe United Kingdom (Table 1). The number of cases decreasedbetween 2005 and 2008 (39 cases in 2005, 33 cases in 2006, 28
cases in 2007, and 29 cases in 2008), with an average of 32.3cases per year. The corresponding annual incidence rate forthe United Kingdom also decreased between 2005 and 2008,from 0.065 to 0.047 cases per 100,000 persons. By comparison,the numbers of M. tuberculosis isolates were 5,488 in 2005,5,548 in 2006, 5,303 in 2007, and 5,468 in 2008. The number ofM. bovis cases in 2005, however, was influenced by the recog-nized cluster of human-to-human transmission in the WestMidlands, described further in this study.
There was no age difference by gender, but 72 (56%) of the129 patients were over 65 years old. The majority of patientswere resident in the West Midlands (18 patients), London (15patients), the South West (14 patients), the South East (11patients), and Scotland (13 patients) at the time of diagnosis.The high number in the West Midlands was expected since acluster has been described to occur there (8). Of the 91 pa-tients where information on country of birth was available, 69(76%) were born in the United Kingdom. Patients born in theUnited Kingdom predominated in every region except Lon-don. Of the 22 patients not born in the United Kingdom, 4were born in European countries, 16 originated from develop-ing countries with high TB endemicity and without establishedM. bovis eradication programs, and country of birth was un-known for 2.
Ethnicity was known for 94 patients, and of these, 78 (83%)were white. White ethnicity predominated in all regions exceptLondon and was most common among patients born in theUnited Kingdom (63 patients; 97%) and patients over 65 yearsold (55 patients; 98%). The over-65 age group was predomi-nant in all regions except London, Northern Ireland, and theEast and West Midlands. Among the 104 cases where thedisease site could be identified, the most common clinical siteof M. bovis disease was pulmonary (42%), followed by pleural(13%) and lymph node (13%) sites.
Transmission of M. bovis in humans in the United Kingdomand associated factors. (i) Patients born in the United King-dom after widespread pasteurization. There were nine patientsborn in the United Kingdom after 1960 (when pasteurizationbecame widespread), which may represent recent transmission.The “epidemiology” branch of the flow chart in Fig. 1 illus-trates how these nine cases (subset A) were derived from ouroriginal study population of 129 cases. This subset was morelikely to have primary infection acquired recently from animalsor humans. The patients in subset A originated from 5 of the12 regions in the United Kingdom (West Midlands, East Mid-lands, South East, South West, and Northern Ireland) andwere all adults aged 19 to 49 years. Five had known travelabroad, but none of the patients had known contact with wildanimals. Four of the nine patients in subset A were from theWest Midlands and had strain typing identical to that of theWest Midlands outbreak strain cluster (8). There were insuf-ficient data to describe risk factors for these four cases. Of thefive patients in subset A that did not have the West Midlandsoutbreak strain, two had no MIRU-VNTR data and the re-maining three patients had unique strains. Only one had ahistory of immunosuppression, and the remaining four re-ported occupational contact with both livestock and nonlive-stock animals, including one patient who reported contactwith both tuberculin test (TT)-positive cattle and a TB-
TABLE 1. Demographic and clinical characteristics of all cases ofM. bovis infection in humans from the United Kingdom and
cases with complete strain typing
Characteristic
No. with characteristic/no. for which information isavailable (%)a for:
All cases(n � 129)
Cases with complete MIRU-VNTR strain typing (n � 69)
Clustered(subset C)(n � 31)
Unique (subsetB) (n � 38)
�Male sex 66/128 (51.6) 15/31 (48.4) 20/38 (52.6)
Age (yr)15-44 35/129 (27.1) 9/31 (29.0) 14/38 (36.8)45-64 22/129 (17.1) 6/31 (19.4) 6/38 (15.8)65 and over 72/129 (55.8) 16/31 (51.6) 18/38 (47.4)
Born after 1960 35/129 (27.1) 9/31 (29.0) 14/38 (36.8)Born in the UK 69/91 (75.8) 22/22 (100)* 15/38 (53.6)*
Ethnic groupWhite 78/94 (83.0) 22/24 (91.7)** 19/28 (67.9)**Black Caribbean 3/94 (3.2) 2/24 (8.3) 1/28 (3.6)Black African 9/94 (9.6) 0 7/28 (25.0)Indian subcontinent 2/94 (2.1) 0 0Other 2/94 (2.1) 0 1/28 (3.6)
UK residenceEast Midlands 5/123 (4.1) 1/30 (3.3) 3/36 (8.3)East of England 9/123 (7.3) 4/30 (13.3) 3/36 (8.3)London 15/123 (12.2) 1/30 (3.3) 6/36(16.7)Northern Ireland 12/123 (9.8)North East 5/123 (4.1) 1/30 (3.3) 2/36 (5.6)North West 8/123 (6.5) 1/30 (3.3) 3/36 (8.3)Scotland 13/123 (10.6) 3/30 (10.0) 4/36 (11.1)South East 11/123 (8.9) 1/30 (3.3) 6/36 (16.7)South West 14/123 (11.4) 6/30 (20.0) 2/36 (5.6)Wales 8/123 (6.5) 3/30 (10.0) 3/36 (8.3)West Midlands 18/123 (14.6) 8/30 (26.7) 4/36(11.1)Yorkshire and
Humber5/123 (4.1) 1/30 (3.3) 0
Disease siteAbdominal 4/104 (3.8) 1/27 (3.7) 1/31 (3.2)Bone/joint 6/104 (5.8) 1/27 (3.7) 3/31 (9.7)Cutaneous 3/104 (2.9) 1/27 (3.7) 1/31 (3.2)Reproductive 3/104 (2.9) 1/27 (3.7) 1/31 (3.2)Lymph node 13/104 (12.5) 3/27 (11.1) 3/31 (9.7)Meningitis 3/104 (2.9) 2/27 (7.4) 1/31 (3.2)Miliary 3/104 (2.9) 0 1/31 (3.2)Pulmonary 44/104 (42.3) 15/27 (55.6) 12/31(38.7)Renal 5/104 (4.8) 1/27 (3.7) 1/31 (3.2)Pleural 13/104 (12.5) 0 7/31 (22.6)Other/mixed sites 7/104 (6.7) 2/27 (7.4) 0
No. of pulmonarycases with smearpositivity
29/33 (87.9) 9/9 (100) 9/11 (81.8)
a �, P � 0.05 (chi-square test); ��, P � 0.05 (Fisher’s exact test).
VOL. 49, 2011 M. BOVIS TRANSMISSION IN THE UNITED KINGDOM 1945
afflicted human and another patient who had cared for asmear-positive cat.
(ii) Evidence of clustering from strain typing. MIRU-VNTRanalysis of 102 completely and incompletely typed isolatesidentified 10 possible clusters (Fig. 2). Sixty-nine (68%) of 102cases had complete MIRU-VNTR profiles, of which 38 wereunique (subset B) and 31 were indistinguishable from at leastone other isolate (subset C). Subset C comprised eight distinctclusters, with cluster sizes ranging from 2 to 12 cases. Theoverall proportion of clustering among completely typed iso-lates (as determined using the n � 1 method) was 33% (23/69cases). Of the 33 cases with incomplete strain typing, 11 hadthe potential to cluster if typed to completion (subset D). Withfull strain types, these 11 cases may have yielded two to sixadditional pairs of clusters by clustering with other complete orincomplete unique strains or expanded existing clusters formedfrom complete strain types. The “molecular” branch of theflow chart in Fig. 1 demonstrates how the 11 completely typedclusters (subsets C1 and C2) and the potential clusters frompartial strain types (subset D) were derived from the originalstudy population. The dendrogram in Fig. 2 illustrates theclusters from completely typed MIRU-VNTRs. Since the out-puts of potential clustering were hypothetical and could not beconfirmed, the 11 partially typed cases were not included in theanalysis of characteristics of clustered cases.
The known characteristics of the 38 unique (subset B) pa-tients and the 31 clustered (subset C) patients with completeMIRU-VNTR typing results are shown in Table 1. Comparedto unique patients, a significantly higher proportion of clus-tered patients were white (92% versus 68%; P � 0.05) andwere born in the United Kingdom (100% versus 53.6%; P �0.05). Clustered patients were also more likely than uniquepatients to live outside London (97% versus 83%), have pul-monary disease alone (56% versus 39%), and be smear positivewith pulmonary disease (100% versus 82%).
The possible risk factors for M. bovis among these 31 clus-tered patients (subsets C1 and C2) are shown in Table 2. Ofthese 31 patients, 15 patients reported possible risk factorssuch as consumption of unpasteurized milk (11 patients), oc-cupational contact with livestock (5 patients), occupationalcontact with nonlivestock animals (2 patients), and close con-tact with a TB-afflicted human (2 patients). One patient re-ported contact with a TT-positive herd, and another had ahistory of TB. For seven patients, multiple risk factors werereported.
The largest cluster of 12 patients (subset C1) shared thesame MIRU-VNTR profile (ETR strain type, 75553; MIRUstrain type, 2222415322) as the strain identified in the WestMidlands outbreak caused by human-to-human transmission(8). Eight patients (five of whom were described in the originaloutbreak report) were from the West Midlands, but three werefrom the South West, and one patient was from Wales. Fivepatients were diagnosed after 2006. Seven patients were bornafter 1960 and also lived in the West Midlands. Only one ofthese seven patients had known risk factors, which were ahistory of consumption of raw dairy products and occupationalcontact with wild animals and livestock. Among the five pa-tients born pre-1960, three had a documented exposure tolivestock or consumption of unpasteurized milk, and four livedin the South West or Wales.
DISCUSSION
This 4-year review of human M. bovis disease in the UnitedKingdom suggests overall decreases in the number and inci-dence of cases, which is consistent with other studies (4, 14).Over half of all M. bovis patients were over 65 years old, andover three-quarters were born in the United Kingdom, white,and living outside London. It is not unexpected that the olderage group was affected, as it is well established that most M.bovis infection in the United Kingdom was acquired in thepre-pasteurization era and manifests now as reactivations (4, 9,14). The site of disease was mainly pulmonary. For M. tuber-culosis cases, this would usually imply infectivity and primaryinfection (16, 19); however, this is not necessarily the case forM. bovis disease. There was limited evidence of human-to-human transmission. The geographical distribution of M. boviscases in more-rural areas is consistent with zoonotic transmis-sion. It is noteworthy that there was an excess of cases in theWest Midlands and the bordering region of South West Eng-land, demonstrating geographical overlap with the distributionof bovine tuberculosis. However, the higher incidence in theWest Midlands is, at least in part, due to a recent outbreak (8).The majority of patients in London were born abroad, of ayounger (15- to 44-year-old) age group, and not white. Thisprobably reflects acquisition of disease abroad. The age profileof patients in the West Midlands also indicates a youngergroup, but here, the reported cluster of human-to-humantransmission could account for the spill over into younger agegroups (8). The cohort of patients known to be born in theUnited Kingdom and after 1960 consisted of only nine people,including four individuals with molecular links with the WestMidlands outbreak strain cluster associated with human-to-human transmission. Four other patients not linked to a clusterwere from rural regions and reported occupational or farmcontact with animals, and for two patients, the animals wereknown to have infectious TB. For these four patients, recentanimal-to-human transmission may have occurred. We wereunable to investigate epidemiological evidence of transmissionof M. bovis between humans and companion animals (24).Further work in this area is recommended, including the in-vestigation of appropriate MIRU-VNTR loci that would besufficiently discriminating to allow the assessment of zoonotictransmission. While the 24-locus MIRU-VNTR typing schemehas been proposed as the standard for M. tuberculosis typing(27), further research on the role of VNTR genotyping for M.bovis is needed. Several studies have confirmed the superiorityof the 24-locus scheme over the 15-locus VNTR typing schemefor M. tuberculosis (2, 3), limiting the discriminatory power ofthis study. However, data from the previously published clusterin the West Midlands suggest that the 15-locus VNTR typingused in this study provides a high degree of discrimination (8).Others have suggested alternatives (1, 12, 22, 25). Neverthe-less, the small number of patients (five) among those born inthe United Kingdom, where consumption of nonpasteurizedmilk was unlikely, who were not part of the group infected bythe West Midlands outbreak strain suggests that transmissionfrom pets is not a major public health issue.
Analysis of clusters. The strain type analysis revealed evi-dence of clustering beyond the well-recognized West Midlandsoutbreak strain cluster, with 31 (subset C) of 69 (45%) com-
1946 MANDAL ET AL. J. CLIN. MICROBIOL.
FIG. 2. MIRU-VNTR results for 102 M. bovis isolates with strain typing data for England, Scotland, and Wales for 2005 to 2008. Dendrogrambased on MIRU-VNTR results, constructed using the categorical coefficient and displayed using the unweighted-pair group method using averagelinkages (UPGMA). Dashes indicate nonamplification at a particular locus. The West Midlands outbreak strain cluster (subset C1) and sevendiscrete clusters of complete MIRU-VNTRs (subset C2) are highlighted. Eleven partial MIRU-VNTRs with the potential to cluster are alsoindicated (�); with a full strain type, these may yield a further 2 to 6 clusters or augment existing clusters.
1947
TA
BL
E2.
Cha
ract
eris
tics
ofcl
uste
red
case
s(s
ubse
tsC
1an
dC
2)id
entifi
edfr
omco
mpl
ete
MIR
U-V
NT
Rst
rain
type
s(n
�31
)a
Clu
ster
Yr
ofdi
agno
sis
Eth
nici
tySe
xA
ge(y
r)D
isea
sesi
te(s
)Sm
ear
stat
us(e
s)N
o.bo
rnpr
e-19
60
No.
born
inth
eU
KD
ocum
ente
dri
skfa
ctor
(s)
for
case
sw
ithin
clus
ter
UK
resi
denc
eM
IRU
-VN
TR
stra
inty
pe
C1b
(n�
12)
2005
(4)
Bla
ckC
arib
bean
(2)
M(7
)20
–29
(3)
Men
ingi
tis(1
)Po
sitiv
e(5
),ne
gativ
e(1
)5
9O
ccup
atio
nalc
onta
ctw
ithno
nliv
esto
ckan
imal
s(2
)So
uth
Wes
t(3
)75
5532
2224
1532
2
2006
(3)
Whi
te(9
)U
nkno
wn
(1)
F(5
)30
–39
(3)
Pulm
onar
y(9
)U
nkno
wn
(6)
Con
sum
ptio
nof
unpa
steu
rize
dm
ilk(3
),w
ork
cont
act
with
catt
le(3
)
Wal
es(1
)
2007
(3)
40–4
9(1
)Pu
lmon
ary,
mili
ary
(1)
Con
tact
with
TB
-affl
icte
dhu
man
(1),
cont
act
with
TT
-pos
itive
catt
le(1
),co
ntac
tw
ithw
ildan
imal
s(1
)
Wes
tM
idla
nds
(8)
2008
(2)
50–5
9(1
)U
nkno
wn
(1)
Liv
edon
afa
rm(1
)60
–69
(3)
Unk
now
n(7
)80
–89
(1)
C2a
(n�
4)20
05(2
)W
hite
(3)
M(2
)60
–69
(1)
Pulm
onar
y(2
)Po
sitiv
e(2
),no
tdo
ne(1
)4
4Pr
evio
ushi
stor
yof
TB
(1)
Eas
tof
Eng
land
(1)
3354
3223
2425
322
2006
(2)
Unk
now
n(1
)F
(2)
70–7
9(2
)Pu
lmon
ary,
bone
/join
t(1
)U
nkno
wn
(1)
Wor
ked
inm
ilkpr
oduc
tion
(1)
Nor
thE
ast
(1)
80–8
9(1
)U
nkno
wn
(1)
Con
sum
ptio
nof
unpa
steu
rize
dm
ilk(3
),liv
edon
afa
rm(1
)Sc
otla
nd(1
)
Con
tact
with
TB
-affl
icte
dhu
man
(1)
Sout
hW
est
(1)
Unk
now
n(1
)
C2b
(n�
2)20
05(1
)W
hite
(2)
F(2
)80
–89
(1)
Lym
phno
de(1
)Po
sitiv
e(1
)2
2C
onsu
mpt
ion
ofun
past
euri
zed
milk
(1)
Eas
tof
Eng
land
(1)
3554
3222
2226
322
2008
(1)
90–9
9(1
)Pu
lmon
ary
(1)
Not
done
(1)
Unk
now
n(1
)W
ales
(1)
C2c
(n�
4)20
05(2
)W
hite
(1)
M(3
)20
–29
(1)
Abd
omin
al(1
)N
egat
ive
(1)
31
Con
sum
ptio
nof
unpa
steu
rize
dm
ilk(1
)L
ondo
n(1
)65
5432
2224
1532
2
2006
(1)
Unk
now
n(3
)F
(1)
50–5
9(1
)C
utan
eous
(1)
Unk
now
n(3
)U
nkno
wn
(3)
Wor
ked
onfa
rmw
ithno
dire
ctan
imal
cont
act
(1)
Scot
land
(1)
2008
(1)
60–6
9(2
)R
enal
(1)
Unk
now
n(3
)So
uth
Wes
t(2
)U
nkno
wn
(1)
C2d
(n�
3)20
06(2
)W
hite
(2)
F(3
)20
–29
(1)
Lym
phno
de(1
)N
egat
ive
(1)
22
Con
sum
ptio
nof
unpa
steu
rize
dm
ilk(1
),liv
edab
road
(1)
Eas
tM
idla
nds
(1)
7554
3222
2415
322
2008
(1)
Unk
now
n(1
)60
–69
(1)
Men
ingi
tis(1
)U
nkno
wn
(2)
Unk
now
n(1
)W
orke
don
afa
rmw
ithliv
esto
ck(1
),un
know
n(1
)E
ast
ofE
ngla
nd(1
)80
–89
(1)
Unk
now
n(1
)U
nkno
wn
(1)
C2e
(n�
2)20
05(1
)W
hite
(1)
M(1
)60
–69
(1)
Pulm
onar
y(2
)Po
sitiv
e(2
)2
1U
nkno
wn
(2)
Scot
land
(1)
7554
3222
2425
322
2006
(1)
Unk
now
n(1
)F
(1)
70–7
9(1
)U
nkno
wn
(1)
Sout
hE
ast
(1)
C2f
(n�
2)20
05(2
)W
hite
(2)
M(1
)80
–89
(2)
Bon
e/jo
int
(1)
Neg
ativ
e(1
)2
2U
nkno
wn
(2)
Eas
tof
Eng
land
(1)
7554
3222
2426
322
F(1
)Pu
lmon
ary
(1)
Unk
now
n(1
)Y
orks
hire
&H
umbe
r(1
)
C2g
(n�
2)20
05(1
)W
hite
(2)
M(1
)60
–69
(1)
Lym
phno
de(1
)Po
sitiv
e(1
)2
2C
onsu
mpt
ion
ofun
past
euri
zed
milk
(2)
Nor
thW
est
(1)
7454
3222
2426
322
2008
(1)
F(1
)90
–99
(1)
Rep
rodu
ctiv
e(1
)U
nkno
wn
(1)
Wal
es(1
)
aN
umbe
rsin
pare
nthe
ses
repr
esen
tth
enu
mbe
rof
case
s.T
T,t
uber
culin
test
;M,m
ale;
F,f
emal
e.b
Incl
udes
five
case
sre
port
edto
occu
rin
the
orig
inal
Wes
tM
idla
nds
outb
reak
stra
incl
uste
r.
1948 MANDAL ET AL. J. CLIN. MICROBIOL.
plete strains clustered into eight groups of indistinguishablestrains (Table 2). This gave a clustering proportion of 33% (asdetermined using the n � 1 method). All clusters were small,comprising two to four cases, with the exception of the WestMidlands outbreak strain cluster. However, this may be a con-servative estimate of clusters and cluster size, as 33 patientshad incomplete DNA fingerprinting due to nonamplification atparticular MIRU-VNTR loci. Eleven of these incompletelytyped cases (subset D) exhibited the potential to cluster ifisolates were retyped to completion, and might yield furtherclusters and/or increase the sizes of existing clusters, as illus-trated in Fig. 1 and 2. Equally, the number of clusters may bereduced by increasing the number of MIRU-VNTR markersused for analysis or by using further analytical techniques, ashas been demonstrated for discriminative analysis of M. tuber-culosis (27, 28).
Pulmonary disease was more common among clusteredcases than among unique cases. We have hypothesized that thismay reflect recent transmission among clustered cases due tothe greater likelihood of transmission when patients have pul-monary disease (19). Although it is widely accepted that evi-dence of clustering in patients with M. tuberculosis suggestsrecent human-to-human spread, the clusters identified heremay not be homogenous in that respect and actually fall intotwo subsets, the “West Midlands outbreak strain” cluster (sub-set C1) and “other” clusters (subset C2). Each has uniquefeatures, and differing hypotheses on the predominant mode oftransmission can be inferred. They are therefore consideredseparately below.
“Other” clusters (subset C2). Although cases in the “other”clusters could be attributable to recent animal-to-human orperson-to-person transmission, the predominance of older age,white ethnicity, and rural residence in these small clusters,along with the absence of any obvious epidemiological links,provides an alternative hypothesis: the clusters could also beexplained by reactivation of latent infection acquired via mul-tiple yet independent episodes of zoonotic transmission of anendemic strain.
West Midlands outbreak strain cluster (subset C1). Thestrain typing analysis revealed further occurrence of the WestMidlands outbreak strain type originally reported as six casesbetween 2004 and 2006 by Evans et al. (8). We found 12identical isolates, of which 7 were newly identified through ouranalysis.
The geographical and age profiles of this cluster changedwith the addition of the new cases. Whereas Evans et al. re-ported that their six patients were resident in the West Mid-lands region and were all born after 1960, the seven patientsnewly identified in our study included four who were bornpre-1960 and lived in regions contiguous with the West Mid-lands (Wales and the South West). Three of these four newpatients resident outside the West Midlands also had docu-mented occupational exposure to animals and/or had con-sumed unpasteurized dairy products. Thus, these new casesmay have arisen following recrudescence of latent infection,possibly acquired through multiple but independent zoonotictransmission routes of a strain endemic to those regions, ashypothesized for subset C2. In contrast, three of the sevennewly identified patients, of whom two were born after 1960,lived in the West Midlands. One of these had exposure to
tuberculin-positive cattle. The spoligotype pattern (SB0263) ofthe West Midlands human outbreak strain is the second mostcommon spoligotype pattern for isolates from cattle in GreatBritain, the most prevalent pattern for cattle in the West Mid-lands, and almost uniquely detected in Britain (8). This sup-ports the assertion that indigenous British cattle were the pri-mary source of infection. Human-to-human transmission likelycontributed to the subsequent outbreak. Indeed, the originalinvestigation of six clustered cases described a plausible mech-anism of spread through social networks with associated pre-disposing factors for tuberculosis (8).
Limitations of study. The main limitation of this study wasthe incompleteness of the data set for some variables, includingrisk factors for transmission. In addition, underascertainmentof cases due to difficulties in obtaining specimens and achiev-ing culture confirmation may have occurred. The small numberof cases is also susceptible to bias and random variation. Thismakes any assessment of trends over time difficult. The originalWest Midlands outbreak strain cases were subjected to a moreextensive epidemiological investigation locally than the othercases, for which surveillance data were the only source ofinformation. The n � 1 method of calculating the number ofclustered cases is based on the assumptions that one case percluster was due to reactivation and that this “index” infectiouscase generated other cases in the cluster either by direct in-fection or by indirect infection via a secondary case (18). Thus,this method yields a more conservative estimate of the propor-tion of clustered cases, likely due to primary disease and hencerecent transmission, than the n method (18). Conversely, theproportion of clustered events may be artificially elevated bythe n � 1 method if clustering is a function of multiple in-stances of zoonotic transmission of an endemic strain ratherthan human-to-human spread. Additionally, consensus onwhich MIRU-VNTR loci are the most discriminative in estab-lishing animal to animal and human-to-human transmissionhas yet to be reached. Data on global endemic animal strainsand ecotypes were not available for comparison with humanstrain data, and so zoonotic transmission hypotheses could notbe investigated further.
Implications and future research. This study demonstratesthat our knowledge of M. bovis transmission in the United King-dom has been improved by better surveillance and wider use ofmore-discriminating molecular techniques. The number and in-cidence of patients diagnosed with M. bovis infection in theUnited Kingdom have decreased overall between 2005 and 2008.The predominant characteristics of patients (elderly, born in theUnited Kingdom, and white) suggest reactivation of latent infec-tion. Despite a number of small clusters, there is little evidence ofwidespread recent human-to-human transmission of M. bovis.Currently, there is no evidence that the number of M. bovis in-fections from food animals or pets to humans has increased. Tomore thoroughly investigate risk factors, data should be consis-tently collected in order for sufficiently powered analytic epide-miological studies to be performed.
M. bovis infection in humans is extremely rare compared toM. tuberculosis infection in humans and does not pose a sig-nificant public health problem (4). Nevertheless, it would seemsensible to maintain vigilance since public demand for unpas-teurized products persists and the incidence of TB in cattlecontinues to rise (5). Reestablishment of eradication programs
VOL. 49, 2011 M. BOVIS TRANSMISSION IN THE UNITED KINGDOM 1949
in the United Kingdom and increased spending on vaccineresearch for cattle are therefore welcomed (5). Since our studyalso suggested that occupational contact with infected animalsmay be a factor in transmission of M. bovis, improved educa-tion and protection of exposed workers should be encouraged.M. bovis is included in the Chief Medical Officer’s 2004 ActionPlan for Stopping TB in England, and one indicator of successis a reduction in the number of M. bovis TB cases over 3 yearsamong persons born in the United Kingdom who were under35 years old (6). Our data indicate no such discernible trend, withone or two cases annually between 2005 and 2008. While a moremeaningful reworking of this indicator is perhaps warranted, theplan does recommend standardized protocols for molecular typ-ing of M. bovis together with centralized databases combiningepidemiological and DNA fingerprinting data (6).
Although the risk of human-to-human transmission is low, itremains prudent for enhanced surveillance of M. bovis to con-tinue to ensure prompt detection of transmission, such as that ofthe West Midlands outbreak strain. Further collaborative re-search between the Veterinary Laboratories Agency, HPA, Pub-lic Health Wales, and Health Protection Scotland is required forcomparative analysis of human and animal M. bovis strains, in-cluding that of evolutionary relatedness and epidemiological link-age of human and animal cases in the United Kingdom.
ACKNOWLEDGMENTS
We are grateful to Rhian Williams and Michael Perry (Wales Centrefor Mycobacteria, Cardiff, United Kingdom), Alan Rayner and Chris-tine Doig (Scottish Mycobacteria Reference Laboratory), Sarah Gar-diner (Regional Centre for Mycobacteriology, Birmingham, UnitedKingdom), and all laboratory staff involved in MIRU-VNTR typing atthe aforementioned mycobacteriology reference laboratories. Wethank the clinical TB teams, the HPU staff, and all who contribute tothe enhanced TB and enhanced M. bovis surveillance systems.
Sema Mandal is a public health doctor trained in the United King-dom and currently working in the United States. This work was un-dertaken as part of the United Kingdom public health specialist train-ing program at the national Centre for Infections, Health ProtectionAgency, Colindale, London, United Kingdom.
S.M. collected and analyzed data and drafted the manuscript, tables,and figures. I.A. initiated and supervised the study, contributed tomanuscript revisions, and will act as guarantor for the paper. L.B.compiled the data set, collected and cleaned data, contributed to dataanalysis and revisions of figures and tables, and commented on drafts.L.F.A. reviewed, commented on, and contributed to manuscript revi-sions. T.B., J.T.E., F.D., J.G.M., I.F.L., A.-L.S., M.R., and P.H. dis-cussed the proposed paper, implemented MIRU-VNTR protocols intheir respective laboratories, supervised analysis and reporting ofDNA fingerprinting data, and reviewed, commented on, and edited themanuscript. G.S., A.B., A.-L.S., and P.L.W. carried out analysis andreporting of DNA fingerprinting in their respective laboratories. O.B.contributed case data and reviewed drafts. R.M. created the dendro-gram. All investigators approved the final manuscript.
M.R. and P.L.W. receive funding from the Welsh Assembly forprospective typing not related to this study. All researchers are inde-pendent of any funders. There are no competing interests to declare.All authors have completed the Unified Competing Interest form athttp://www.icmje.org/coi_disclosure.pdf (available on request from thecorresponding author) and declare that they have had no relationshipswith companies that might have an interest in the submitted work inthe previous 3 years, that their spouses, partners, or children have nofinancial relationships that may be relevant to the submitted work, andthat they have no nonfinancial interests that may be relevant to thesubmitted work.
REFERENCES
1. Allix, C., et al. 2006. Evaluation of the epidemiological relevance of variable-number tandem-repeat genotyping of Mycobacterium bovis and comparisonof the method with IS6110 restriction fragment length polymorphism anal-ysis and spoligotyping. J. Clin. Microbiol. 44:1951–1962.
2. Christianson, S., et al. 2010. Evaluation of 24 locus MIRU-VNTR genotyping ofMycobacterium tuberculosis isolates in Canada. Tuberculosis (Edinb.) 90:31–38.
3. Comas, I., S. Homolka, S. Niemann, and S. Gagneux. 2009. Genotyping ofgenetically monomorphic bacteria: DNA sequencing in Mycobacterium tubercu-losis highlights the limitations of current methodologies. PLoS One 4:e7815.
4. de la Rua-Domenech, R. 2006. Human Mycobacterium bovis infection in theUnited Kingdom: incidence, risks, control measures and review of the zoo-notic aspects of bovine tuberculosis. Tuberculosis (Edinb.) 86:77–109.
5. Department for Food Environment and Rural Affairs. 2009. Animal healthUK—2008. The report of the Chief Veterinary Officer. Department forFood, Environment and Rural Affairs. HMSO, London, United Kingdom.
6. Department of Health. 2004. Stopping tuberculosis in England. Action planfrom the Chief Medical Officer. Department of Health, London, United King-dom.
7. Dupon, M., and J. M. Ragnaud. 1992. Tuberculosis in patients infected withhuman immunodeficiency virus 1. A retrospective multicentre study of 123cases in France. The Groupe des Infectiologues du Sud de la France. Q. J.Med. 85:719–730.
8. Evans, J. T., et al. 2007. Cluster of human tuberculosis caused by Mycobac-terium bovis: evidence for person-to-person transmission in the UK. Lancet369:1270–1276.
9. Grange, J. M. 2001. Mycobacterium bovis infection in human beings. Tuber-culosis (Edinb.) 81:71–77.
10. Hawkey, P. M., et al. 2003. Mycobacterial interspersed repetitive unit typingof Mycobacterium tuberculosis compared to IS6110-based restriction frag-ment length polymorphism analysis for investigation of apparently clusteredcases of tuberculosis. J. Clin. Microbiol. 41:3514–3520.
11. Health Protection Agency. 2009. Tuberculosis in the UK: annual report ontuberculosis surveillance in the UK: 2008. HPA, London, United Kingdom.
12. Hilty, M., et al. 2005. Evaluation of the discriminatory power of variablenumber tandem repeat (VNTR) typing of Mycobacterium bovis strains. Vet.Microbiol. 109:217–222.
13. HPA/VLA. 2008. Health Protection Agency and Veterinary LaboratoriesAgency joint collaboration report 2006–2007. HPA/VLA, London, United King-dom.
14. Jalava, K., et al. 2007. No increase in human cases of Mycobacterium bovisdisease despite resurgence of infections in cattle in the United Kingdom.Epidemiol. Infect. 135:40–45.
15. King, D. 2007. Bovine tuberculosis in cattle and badgers. DEFRA, London,United Kingdom.
16. Kruijshaar, M. E., and I. Abubakar. 2009. Increase in extrapulmonary tu-berculosis in England and Wales 1999–2006. Thorax 64:1090–1095.
17. Meikle J. 2000. Vets link cats with risk of catching TB. In The Guardian, 1 April2000. http://www.guardian.co.uk/uk/2000/apr/01/jamesmeikle.
18. Murray, M., and D. Alland. 2002. Methodological problems in the molecularepidemiology of tuberculosis. Am. J. Epidemiol. 155:565–571.
19. Musellim, B., S. Erturan, E. Sonmez Duman, and G. Ongen. 2005. Com-parison of extra-pulmonary and pulmonary tuberculosis cases: factors influ-encing the site of reactivation. Int. J. Tuberc. Lung Dis. 9:1220–1223.
20. Richter, E., M. Weizenegger, S. Rusch-Gerdes, and S. Niemann. 2003. Eval-uation of genotype MTBC assay for differentiation of clinical Mycobacteriumtuberculosis complex isolates. J. Clin. Microbiol. 41:2672–2675.
21. Rivero, A., et al. 2001. High rate of tuberculosis reinfection during a noso-comial outbreak of multidrug-resistant tuberculosis caused by Mycobacte-rium bovis strain B. Clin. Infect. Dis. 32:159–161.
22. Roring, S., A. N. Scott, R. Glyn Hewinson, S. D. Neill, and R. A. Skuce. 2004.Evaluation of variable number tandem repeat (VNTR) loci in molecular typingof Mycobacterium bovis isolates from Ireland. Vet. Microbiol. 101:65–73.
23. Samper, S., et al. 1997. Transmission between HIV-infected patients ofmultidrug-resistant tuberculosis caused by Mycobacterium bovis. AIDS 11:1237–1242.
24. Shrikrishna, D., R. de la Rua-Domenech, N. H. Smith, A. Colloff, and I. Coutts.2009. Human and canine pulmonary Mycobacterium bovis infection in the samehousehold: re-emergence of an old zoonotic threat? Thorax 64:89–91.
25. Skuce, R. A., et al. 2005. Discrimination of isolates of Mycobacterium bovis inNorthern Ireland on the basis of variable numbers of tandem repeats(VNTRs). Vet. Rec. 157:501–504.
26. Smith, R. M., et al. 2004. Mycobacterium bovis infection, United Kingdom.Emerg. Infect. Dis. 10:539–541.
27. Supply, P., et al. 2006. Proposal for standardization of optimized mycobac-terial interspersed repetitive unit-variable-number tandem repeat typing ofMycobacterium tuberculosis. J. Clin. Microbiol. 44:4498–4510.
28. Velji, P., V. Nikolayevskyy, T. Brown, and F. Drobniewski. 2009. Discrimi-natory ability of hypervariable variable number tandem repeat loci in pop-ulation-based analysis of Mycobacterium tuberculosis strains, London, UK.Emerg. Infect. Dis. 15:1609–1616.
1950 MANDAL ET AL. J. CLIN. MICROBIOL.
