HIV, TB and ART: the CD4 enigma
Brian Williams
Treatment as Prevention in Africa 1 May 2014, Gaborone, Botswana
Williams (2012) http://arxiv.org/pdf/1211.2798v2.pdf
HIV drives TB Nunn et al. Nature Reviews of Immunology. 2005; 5: 819-26
Delay ~ 5 years
IRR ~ 15
Kisumu, Kenya
9.4
1.11.11.0
5.9
2.2
0
2
4
6
8
10
1991-1994 1995-1997 1998-1999
Ann
ual i
ncid
ence
(%)
. HIV- HIV+
TB incidence among gold miners in SACorbett et al. J Infect Dis. 2003;188: 1156-63.
HIV- HIV+
1.02.2
1.1
5.9
1.1
9.4
1991-1994 1995-1997 1998-1999TB in South African Gold Miners
10
8
6
4
2
0
Inci
denc
e (%
p.a
.)9.4
1.11.11.0
5.9
2.2
0
2
4
6
8
10
1991-1994 1995-1997 1998-1999
Ann
ual i
ncid
ence
(%)
. HIV- HIV+
TB incidence among gold miners in SACorbett et al. J Infect Dis. 2003;188: 1156-63.
HIV- HIV+
1.02.2
1.1
5.9
1.1
9.4
1991-1994 1995-1997 1998-1999TB in South African Gold Miners
10
8
6
4
2
0
Inci
denc
e (%
p.a
.) IRR ~ 10
IRR ~ 2 No change
But not in HIV-negative people
Gold Miners in South Africa
Corbett et al. Journal of Infectious Diseases 2003; 188: 1156-63
0
100
200
300
400
500
600
700
1980 1985 1990 1995 2000 2005 2010
TB
inci
denc
e/10
0k/y
rTB
inci
denc
e/10
0k/y
r
0
100
200
300
400
500
600
700
1980 1985 1990 1995 2000 2005 2010
TB
inci
denc
e/10
0k/y
rTB
inci
denc
e/10
0k/y
r
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1980 1985 1990 1995 2000 2005 2010
HIV
pre
vale
nce
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
1980 1985 1990 1995 2000 2005 2010
HIV
pre
vale
nce
0
100
200
300
400
500
1980 1985 1990 1995 2000 2005 2010
TB in
cide
nce/
100k
/yr
IRR ~ 34 IRR ~ 7.4
11x 2.7x
Zimbabwe Botswana
Williams et al. Proc Nat Acad Sc USA. 2010; 107: 17853-4.
Tube
rcul
osis
H
IV
Impact depends on the setting
The only two things that are known to be predictive of TB are CD4 cell counts in
HIV-positive people and BMI.
Lonnroth et al. Int J Epidemiol. 2010; 39: 149-55.
If CD4 falls by 100 cells/mL TB increases by 36% (19%−42%)
CD4 cells/mL
Ann
ual i
ncid
ence
of T
B
Antonucci et al. JAMA. 1995; 274: 143-8; Badri et al. Lancet. 2002; 359: 2059-64.
I µ e - 0.0036 C
1500
1000
500
0
Ethio
pia
Guine
aBi
ssau
Nige
riaS.
Afric
aTa
nzan
ia
Ugan
da
Zamb
ia
Ugan
da
CD4/µL
Williams et al. J Inf. Dis. 2006; 194: 1450-8; Bussman et al. Clin. and Diag. Lab. Immunology 2004
1,200
600
1,276
233
Mean (95%) Range (95%)
CD4 highly variable in HIV-negative people
Botsw
ana
0
200
400
600
800
1000
0 1 2 3 4 5 6 7
Time spent on ART/years
Incr
ease
in C
D4
cell
coun
t But not when you recover under ART
Gras et al. J Acquir Immune Defic Syndr. 2007; 45: 183-92.
So putting all of this together: What can modelling tell us?
Williams et al. PNAS 2005; 102: 9619-24.
Fit the trend in HIV prevalence
HIV in Botswana
Putting lots of people on ART
HIV in Botswana
Leaving fewer people not on ART
HIV in Botswana
Reduces infectiousness and hence incidence
HIV in Botswana
But has a much greater impact on mortality
HIV in Botswana
TB in Botswana
HIV-negative
On ART Off ART
Total
What do we know?
• HIV drives TB with a delay of about 5 years and an IRR of ~ 5 to 40. • TB in HIV+ people does not increase TB in HIV- people • ART reduces the risk of TB by about 60% at all CD4 cell counts.
What do we think we understand?
• The short disease duration in HIV-positive people limits the impact on TB in HIV-negative people.
• CD4 may be the key link between HIV and TB. • CD4 recovery is independent of when we start ART; the increase
due to ART is independent of starting CD4. • ART restores some, but not all, CD4 function reducing AIDS-
related TB by about 60%
What do we still not understand? • If CD4 is so variable why is it so important? • Why is HIV-survival independent of the initial CD4 cell count? • If CD4 cell counts determine the risk of TB in HIV-positive people is
it also true for HIV-negative people? • Since CD4 is so variable on the way down; why is it so predictable
on the way up? • Why is the incidence rate ratio for TB with and without HIV in
Zimbabwe five times that in Botswana?
What do we (tentatively) conclude?
• ART will reduce HIV-TB by up to 60% in the short term. • To eliminate HIV-related TB we must eliminate HIV.
The classic tuberculous granuloma, found in active disease and latent infection
Barry et al. Nat Rev Microbiol. 2009; 7: 845-55.
During the acute phase of HIV: CD4 cell counts drop by 25% (9%--41%)1
TB incidence rises 3.8 (1.6--15.2) times.2
1. Williams et al. J Infect Dis. 2006; 194: 1450-8; Williams et al. Proc Nat Acad Sc USA. 2010; 107: 17853-4; 2. Antonucci et al. JAMA. 1995; 274: 143-8; Badri et al. Lancet. 2002; 359: 2059-64.
Explains the early impact on gold mines
CD4 recovery is independent of when you start… In
crea
se in
CD
4 ce
ll co
unt
0
100
200
300
400
0 1 2 3 4 5 6 7
Rate of increase = 224 (192--261)/mL/yr
Asymptote = 371 (355 389)/mL
Time spent on ART/years
CD4 cells/mL
Ann
ual T
B in
cide
nce
…but (TB) recovery is never complete
HIV-
Gupta et al. PLoS One 2012; 7: e34156.
4x
0
100
200
300
400
500
600
700
1980 1985 1990 1995 2000 2005 2010
TB
inci
denc
e/10
0k/y
rTB
inci
denc
e/10
0k/y
r
0
100
200
300
400
500
1980 1985 1990 1995 2000 2005 2010
TB in
cide
nce/
100k
/yr
IRR ~ 34 IRR ~ 7.4
Zimbabwe Botswana
Williams et al. Proc Nat Acad Sc USA. 2010; 107: 17853-4.
Tube
rcul
osis
Impact depends on the setting
0
200
400
600
800
0 2 4 6 8 10 120
200
400
600
800
0 2 4 6 8 10 12
CD
4/m
L
CD
4
CD
4/m
L
High CD4: Low TB Fast decline: big increase
Low CD4: High TB Slow decline: small increase
TB in HIV positive people does not affect HIV-negative people
Prevalence = Incidence x Duration
HIV-positive people with TB progress about 15 times faster
than HIV negative people.
Corbett et al. Am J Respir Crit Care Med. 2004; 170: 673-9; Wood et al. Am J Respir Crit Care Med. 2007; 175: 87-93; Williams et al. Am J Respir Crit Care Med. 2007; 175: 6-8.
Time to death (years)
CD
4 ce
ll co
unts
ART drives CD4 back up
Falling epidemic
Rising epidemic
Probability density
Modelling CD4 progression
Williams JID 2006; CASCADE Lancet. 2000; Gras JID 2007
1
10
100
1000
15 20 25 30 35BMI (kg/m2)
TB in
cide
nce/
100k
/yea
r
Navy recruits, (USA, 2); Male smokers (Finland); Mass radiography (Norway); Pensioners (Hong Kong); NHANES (USA).
Lonnroth et al. Int J Epidemiol. 2010; 39: 149-55.
p = 10-51 13.8% (13.4%14.2%)/BMI unit
0
50
100
150
200
250
-50 0 50 100 150 200 250Change in CD4 cell counts/µl/yr
Rel
ativ
e fre
quen
cy .
Williams et al. arXiv 2009 http://arxiv.org/abs/0908.1556 Rodriguez et al. JAMA 2006; 296: 1498-506.
Predicted distribution of the rate of change of CD4 cell counts over time
“First do no harm”
Data: Hargrove AIDS 2010; Model: Williams JID 2006.
0.001
0.010
0.100
1.0000 500 1000 1500 2000
CD4+ cell count M
orta
lity
at o
ne y
ear
HIV-positive
HIV+ and HIV- post-partum women in Harare (1997-2000)
2.5x 3.9x 100x
HIV-negative
0.00
0.05
0.10
0.15
0.20
0 500 1000 1500
0.00
0.05
0.10
0.15
0.20
0 500 1000 1500 2000
Predicted distribution of the distribution of CD4 cell counts in HIV-positive people given the distribution
in HIV-negative people in South Africa Williams et al. J Infect Dis. 2006; 194: 1450-8.
ART treatment according to various guidelines
Per
cent
nee
ding
trea
tmen
t Young men, Orange Farm, 2002
Williams et al. arXiv 2012; http://arxiv.org/abs/1208.3434
IAS/DHHS WHO
Countries 20 5
40
3
3
1
Year Organiz. CD4/mL Viral load/mL Countries 1996 IAS 500 30k 1997 IAS 500 5k 2000 IAS 350 5k 2002 IAS 200 50k 2002 WHO 200 2003 WHO 200 2007 WHO 200 2009 WHO 350 2010 IAS 500 100k 2012 Practice Immediate 1
500 3 350-500 3 350 40 200-350 5 250 5 200 15