Overview of TB-related studies at IAS

5 October 2009. Related: Conference reports, TB coinfection, IAS 5th Cape Town 2009.

Nathan Geffen, TAC

Some useful TB research has been published since the last issue of HTB. With approximately 120 TB-related abstracts at IAS2009, we summarise some of the most important.

Epidemiology

The most interesting TB research presented at the IAS conference dealt with the epidemiology of TB/HIV co-infection.

A presentation by Keren Middelkoop and colleagues analysed the association between the introduction of HAART in a community and active TB rates. [1]

They collected TB notification data, prior to HAART (1998 to 2004) and after HAART was introduced (2004 to 2008), in Masiphumelele township in Cape Town. HIV prevalence was estimated to be 23% in the township. Pre-ART adult TB notifications increased by an average of 212 cases per 100,000 people per year (p =0.005 for trend). Post-HAART, adult cases decreased by 116 per 100,000 people per year (p=0.16, NS for trend).

TB rates in HIV-negative people did not change substantially over the period and averaged 697 cases per 100,000 per year. TB rates in HIV-positive people increased by an average of 826 cases per 100,000 per year over the same period (p value for trend 0.08), but after the introduction of HAART, declined by 600 cases per 100,000 per year (p=0.16, NS for trend). For HIV-positive people not on HAART, the average decline in the post-HAART era was 421 cases per 100,000. For people on HAART, it declined by 1,394 cases per 100,000 and this trend was significant (p=0.05). The authors concluded that wide-scale HAART implementation and the community’s well-functioning TB programme were associated with modest TB declines.

Masiphumelelo’s population is only 20 to 30,000 which might explain a lack of power to detect statistically significant trends. But another study in Masiphumelelo also conducted by Middelkoop and colleagues makes the argument for the effect of ART on a declining TB rate more compelling. [2]

In 2008 they repeated a randomly sampled cross-sectional survey that was first performed in 2005. About 30% of HIV-positive adults were estimated to be on HAART in 2008. Two sputum samples were obtained from each participant. Participants also filled in questionnaires on TB history. Those who were not being treated for TB and had two TB-positive sputum or culture tests were considered to have undiagnosed TB. The survey measured a significant decline in the TB rate in HIV-positive people since 2005 (all testing was anonymous), including a significant decline in undiagnosed TB in this group. No such decline was seen in HIV-negative people. The study also found that after adjusting for age, sex and HIV status, the overall 2008 TB prevalence was significantly lower than the overall 2005 prevalence (p=0.02). See Table 1.

Shortly after the conference, members of this research group published an article in AIDS that examined the association between current CD4 count and risk of TB. [3]

They analysed TB incidence in 1480 patients receiving ART for up to 4.5 years. Updated CD4 cell counts were measured 4-monthly. During 2785 person-years of observation, 203 cases of TB were diagnosed, giving an overall incidence of 7.3 cases/100 person-years. TB incidence by CD4 count strata is described in Table 2 (p<0.001).

Table 1: Results of surveys in 2005 and 2008 in Cape Town township

Table 2: TB incidence by CD4 count strata (from Lawn et al.)

The authors found that updated CD4 cell counts were the only patient characteristic independently associated with long-term TB risk. They concluded that updated CD4 cell counts were the dominant predictor of TB risk during ART in this low-resource setting. They also found that among those with baseline CD4 cell counts less than 200 cells/mm³, the excess adjusted risk of TB during the first four months of ART was consistent with unmasking of disease missed at baseline screening. They noted that TB incidence at CD4 cell counts of 200-500 cells/mm³ remained high and concluded that TB prevention would be improved by ART policies that minimised the time patients spend with CD4 cell counts below 500 cells/mm³.

These studies provide evidence that widespread ART implementation can reverse the growth in active TB rates in Southern Africa.

A study by Van Rie and colleagues analysed risk factors for TB in patients who had been on HAART for greater than six months in a prospective cohort from the Themba Lethu clinic in Johannesburg. [4]

This study provided quantitative data on TB incidence and its relationship to patients failing treatment. Of 5934 adults, 217 (4%) developed TB after six months. Median time to TB was 418 days (IQR: 276-672, incidence of 2.3 cases per 100py). The incidence was four times lower than in the first six months of ART. Significant risk factors associated with active TB were BMI <18.5 compared to BMI =25 (HR: 6.52, 95%CI: 3.60-11.80) history of TB treatment (HR: 1.50, 95%CI: 1.09-2.50), current viral load >10,000 copies/mL (HR: 2.44, 95%CI: 1.57-3.79) and CD4 count = 50 cells/mm³ (HR=0.84, 0.45, and 0.34, for CD4 51-100, 101-200, and 201-350 respectively).

In a study from Khayelitsha, Pepper and colleagues examined patients in their cohort who had clinical deterioration. [5]

They enrolled 298 people with who had initiated TB treatment from June to August 2008 and then followed them for six months. In this group, 209 (71%) were HIV-positive, with a median CD4 count of 129 cells/mm³ (IQR: 62-277). At TB diagnosis, 35 (17%) HIV-positive patients were receiving HAART. This rose to 112 (54%) on HAART six months later.

Within 6 months, 117 patients (39%) experienced 208 episodes of clinical deterioration. Of these, patients, 71% were HIV-positive. There was an escalating risk of clinical deterioration in HIV-positive patients as CD4 counts decreased (CD4>350: RR:1.4; 95%CI=0.7-2.9; CD4 200-350: RR:2.0, 95%CI: 1.1-3.6; CD4< 200: RR=3.0, 95%CI=1.9-4.8).

AIDS-defining illnesses (n=30), TB-IRIS (n=22) and MDR-TB (n=10) were important causes for clinical deterioration. The number of deaths was 17, of whom 15 were HIV-positive with a CD4 count <200 cells/mm³ at TB diagnosis. The authors also noted that health-care use was significantly higher in HIV-positive patients with low CD4 counts. They pointed out that starting HAART initiation at higher CD4 counts is likely to reduce this clinical burden.

De Bruyn and colleagues presented data from Soweto that showed an association between ART and neutrophil count. [6]

They explain that neutrophil granules contain antimicrobial peptides that kill M.tuberculosis. In HIV-negative people exposed to TB, neutrophil count is inversely associated with risk of latent TB infection and positively associated with ability to contain mycobacterial growth in vitro. However, neutrophil functional defects occur in HIV-positive patients. The authors therefore examined the association of incident TB with neutrophil count.

They followed a prospective cohort of almost 2700 HIV-positive adults for over 5500 person-years. Median age was 32. Women comprised 79% of the cohort. Median CD4 count was 282 cells/mm³. The median neutrophil count was 2.46 copies/nL. TB incidence was 5.2/100py (95%CI: 4.6-5.8). ART was associated with a reduced risk of TB (HR: 0.26; p<0.001), as was increasing CD4 count. Increasing neutrophil count was also associated with increased risk of TB (HR: 1.18; p=0.02). For patients on ART, there was a trend showing that risk of TB was reduced by 75% per nL increase in neutrophil count (HR: 0.25, p=0.08).

The authors conclude that the association between neutrophil count and risk of tuberculosis in HIV-positive adults varies according to whether HAART is administered. Their results suggest that HAART critically influences the relationship between neutrophils and the risk of TB.

COMMENT

The potential role of earlier ART in reducing the epidemiological impact on TB is encouraging.

This supports the recommendation to start HAART at CD4 counts over 350 cells/mm³ in patients coinfected with TB.

Drug resistance

Research on drug-resistant TB continued to be a concern.

A study by Max O’Donnell and colleagues who have produced excellent data on the drug-resistant TB epidemic, found that health care workers in Kwazulu-Natal had a much higher incidence of TB than the general population. [7]

Based on data from King George V Hospital, MDR-TB incidence was 58.9 per 100,000 people for the province’s health workers and 10.7/100,000 for the province’s general population (OR: 5.53; 95%CI; 4.70-6.50). XDR-TB incidence was 4.0/100,000 among health workers and 1.0/100,000 in the general population (OR: 3.89 95%CI: 2.02-7.11). There were 235 cases of health workers with drug-resistant TB and 3,391 cases for non health-workers at the hospital. About half of drug-resistant patients were HIV-positive and this did not differ between health workers and the general population. The high incidence amongst health workers is clearly related to occupational exposure. The researchers therefore conclude that screening and controlling occupational exposure among health workers is critical to limit nosocomial spread of drug-resistant TB.

In another Khayelitsha study, Helen Cox and colleagues reported on the prevalence of drug-resistant TB. [8]

This team conducted a representative cross-sectional survey of clients attending two clinics suspected of having pulmonary TB between May and November 2008. Of 1,850 TB suspects surveyed, 536 (30%) were culture-positive. HIV status was known for 427 (80%) cases with 261 HIV-positive (61%). Rifampicin resistance was found in 4% of new cases and 10% of previously treated cases (p=0.003), and in 8.0% of HIV-positive and 4.8% of HIV-negative cases (p=0.18). They estimated rifampicin resistance in Khayelitsha to be 50 to 72/100,000 people per year. They concluded that there is extremely high prevalence of drug-resistant TB in the township. Moreover, that it is high amongst HIV-negative people too.

PACTG 1041 was a double-blind placebo controlled trial to test isoniazid preventative therapy (IPT) in perinatally HIV-exposed infants. The primary endpoint was TB disease, infection or death. The DSMB stopped the trial because of futility. Partial results were reported at ICAAC 2008 (covered in HTB Nov/Dec 2008) and CROI 2009. At IAS2009 a poster by Anneke Hesseling and colleagues reported the results of an analysis of the 22 culture-confirmed cases of TB in the trial. [9]

Of these, 18 were sent for drug-susceptibility testing. Five were drug-resistant (one to INH, four MDR-TB). The authors conclude that the high rate of drug-resistant TB in the trial is consistent with the growth of the adult drug-resistant TB epidemic and has potential consequences for the programmatic implementation of IPT.

COMMENT

The growing evidence of a rising drug-resistant TB epidemic that might undermine the benefits of IPT, particularly in infants, is concerning. So too are the high rates of drug-resistance in HIV-negative people beyond nosocomial infection in Khayelitsha. The South African government is still not demonstrating adequate commitment to co-ordinating and prioritising infection control measures and contact tracing. For example, it would be useful to have a project to revamp clinic waiting rooms (along similar lines to the waiting room at the Ubuntu clinic in Khayelitsha, which has heaters and a roof, but no walls allowing a continuous flow of air). A public information campaign to keep windows open in public places (e.g. buses and taxis) is also important.

Izoniazid preventative therapy (IPT)

One concern about community-wide IPT is that it will result in higher isoniazid resistance. A study by Halsema and colleagues provides promising data that will help allay, albeit not completely, this fear. [10]

The Thibela TB study is a large cluster-randomised trial to study IPT strategies in South African gold mines. Individual mine shafts are randomised to receive either standard TB control (IPT to miners with silicosis or HIV infection) or the intervention (IPT to everyone in the mine, from miners to executives).

In this case-controlled study, drug susceptibility data from TB cases among people who received IPT in the Thibela TB intervention clusters were compared to two groups: (1) TB cases in the control clusters and (2) a subset of patients from a laboratory substudy confined to the first TB episodes in the control clusters. The comparison cases were restricted to the same calendar period as the intervention cases. The Thibela TB intervention began in July 2006 and the study included all TB cases in the intervention clusters up to mid-February 2009.

The intervention group included all participants receiving IPT who attended at least one follow-up visit and were subsequently treated for TB, unless they did not have positive TB cultures. Of the 126 individuals who met the inclusion criteria in the intervention arm all but one were male. The median age was 43 years. Of 103 patients with known HIV status, 89 (86.4%) were HIV-positive. Median CD4 cell count was 196 cells/mm³ (n=51). The median time from starting IPT to TB treatment was 316 days (IQR: 174-491). For 75% people (n=94) this was their first TB episode and 25.4% (n=32) were retreated. TB was pulmonary for 87 (69.0%), extra-pulmonary for 22 (17.5%) and disseminated for 17 (13.5%).

Amongst the intervention cases, 96 outcomes had been documented at the time of the analysis: 39 (41%) were cured, a further 23 (24%) completed treatment, 8 (8.3%) died, there was one treatment interruption, one treatment failure, 11 (11.5%) transferred out or were lost-to-follow-up and for 13 (13.5%) cases the outcomes were unknown.

The authors concluded that TB disease after IPT may be largely due to re-infection in this high HIV prevalence group.

Data on drug susceptibility was presented for 58 people in the intervention group, 182 in the control clusters and 32 in the laboratory substudy. Table 3 presents the results of their analysis.

Table 3: Drug-resistance in three groups from Thibela TB study

There were no significant differences between the groups in any of these outcomes. However, the authors noted the very wide confidence intervals (depicted graphically on their poster). In their discussion, they explained that if IPT is more effective at treating isoniazid susceptible latent TB, then an increased proportion, but not absolute numbers, of isoniazid resistant TB cases would be expected among those who have taken IPT. They concluded however, that the proportion of TB episodes in the intervention groups did not differ from the controls, and that these data do not support concerns about IPT induced resistance.

TB treatment

The standard model for the management of TB patients that is promoted by the WHO involves directly observed treatment (DOT). In this model, patients come daily to their health facilities to take their pills under the supervision of a health worker.

Atkins et al. described an alternative model piloted from April 2007 to March 2008 in five health facilities in Cape Town. [11]

In this intervention, adult TB patients received adherence counselling. They also selected a treatment buddy. Lay health workers supported patients’ self-supervised treatment.

Using information stored in a routine electronic TB register, TB data from these five clinics was compared against another five clinics that use DOT. Across the five intervention clinics 75% of new patients were treated using the new model. In these clinics, treatment success was 72.4% (95%CI: 67.4-77.4) and in the control clinics it was 75.9% (95%CI: 70.8 to 80.9), a non-significant difference.

In the previous issue of HTB South, we reported on the results of the CAPRISA trial that compared immediate versus deferred HAART in patients coinfected with TB. It showed significant reduced mortality in the immediate arm. We commented on the importance of integrating TB and HIV services to make this intervention easier to implement. Further evidence of the benefits of integrating TB and HIV treatment comes from a Malawian study.

Malawian TB case notifications have risen dramatically over the last two decades. Chan and colleagues described what happened when the Zomba Central Hospital ART Clinic, which opened in 2004, began integrating TB services. [12]

Zomba district has a population of 670,000 (80% rural). HIV prevalence amongst 15 to 49 year-olds is estimated to be 16.5%. Routine national TB programme data shows that 69% of TB patients are HIV-positive.

Integration of services began in September 2007 through monthly HIV/TB integration days. By April 2008, services were fully integrated with a daily TB/HIV Integration clinic where all patients registered for TB were also tested for HIV and referred for HIV care in the same physical area. Ministry of Health TB patient records and HAART records from September 2007 to December 2008 were reviewed to assess uptake of HAART. Following integration, HAART uptake increased dramatically from 4% to 33% in HIV monoinfected patients and from 25% to 50% in patients with HIV/TB coinfection. See Table 4.

The CARINEMO-ANRS 12146 Trial is a randomised, open-label non-inferiority study comparing 48 weeks virological suppression and safety of nevirapine (400mg daily without leading dose) vs efavirenz (600mg daily) co-administered with rifampicin. The other ARVs in the study regimen were d4T and 3TC. HAART was started four to six weeks after TB treatment. Bhatt and colleagues presented preliminary safety data, covering the period November 2007 to December 2008. [13]

By the end of this period, 236 patients with CD4 counts <250 cells/mm³ and who were co-infected with active TB had been randomised. Of these, 11 (4.7%) discontinued the study (6 due to death, 3 withdrew consent, 1 lost-to-follow-up and 1 other).

Follow-up included weekly clinical assessments for the first eight weeks of HAART and monthly assessments thereafter. Patients also had aminotransferase (ALT) measurements every two weeks during the first eight weeks followed by monthly measurements. 204/236 patients (86.4%) presented at least one adverse event. There were 26 serious adverse events, of which six resulted in death. None of the deaths were drug-related.

Skin-related adverse events were reported in 47 patients (19.9%), but none were severe. Also, 11 (4.7%) patients had an ALT increase (> grade 3). Five patients (2.1%) interrupted treatment due to hepatitis. However, there were no cases of severe rash. The researchers concluded that this plus the relatively low number of cases of severe hepatitis and treatment interruptions due to adverse events were reassuring but needed to be confirmed. Final results are expected at the end of 2010. A study by Kamateeka and colleagues of children taking rifampicin and nevirapine is also reviewed in this issue of HTB.

A study from a Uganda found that efavirenz is associated with a greater decline in TB incidence than nevirapine. Hermans and colleagues reported data from their large cohort of ART patients (n=7,648). [14]

Between May 2002 and January 2009 they identified TB events in patients who had been on HAART for two years or less.

At baseline, median CD4 was 111 cells/mm³ (IQR: 38-179) in the cohort and 85 cells/mm³ (IQR: 30-149) in patients with TB coinfection. For the whole cohort, 30% were in WHO stage I or II, 40% in stage III and 30% in stage IV (the TB patients had similar proportions).

In the first two years of HAART (almost 13,600 PYFU), there were 360 (4.7%) new TB events (2.65 per 100PY; 95%CI: 2.39 ¬ 2.94). Incidence rates declined with time on HAART. For 0-3, 3-6, 6-12 and 12-24 months they were 9.91 (95%CI: 8.51-11.55), 5.14 (95%CI: 4.11-6.44), 2.16 (95%CI: 1.66-2.82) and 0.82 (95%CI: 0.64-1.05), respectively.

Table 4: Numbers of patients (%) in HAART uptake following integration of TB and HIV clinics at Zomba Central Hospital , Malawi

In a multivariate analysis, baseline CD4 count <50 cells/mm³ (HR 1.58; 95%CI: 1.10-2.27; p=0.01) and male sex (HR 1.43; 95%CI: 1.15-1.77; p=0.001) were significantly associated with increased risk for TB.

A key finding of the study is that 100 patients out of 2842 receiving AZT/3TC/efavirenz versus 227 out of 3974 using d4T/3TC/nevirapine developed TB (832 used other regimens). Compared to the d4T/3TC/nevirapine regimen, the HR for the AZT/3TC/efavirenz was 0.7 (95%CI: 0.53-0.89; p=0.003).

This difference could not be explained by differences in baseline CD4, calendar year starting HAART or immune restoration status after 14 months of HAART. In a multivariate analysis, the HR was 0.67 (95%CI: 0.53-0.86; p=0.002). The researchers further point out that this association occurred despite clinician bias towards prescribing efavirenz to patients with any TB symptoms to avoid subsequent switching due to interactions between nevirapine and rifampicin. This has not been previously described.

Is therapeutic drug monitoring needed in people with MDR-TB taking ofloxacin? This was a question that arose in a small proof of technology study reported by Mugabo and colleagues at Brooklyn Chest Hospital in Cape Town. [15]

Previously, PK values for ofloxacin have been obtained primarily using high performance liquid chromatography (HPLC) from cohorts in rich country. This study tested liquid chromatography coupled with mass spectrometry and found it to be simple, specific, accurate, sensitive and reproducible.

The inclusion criteria for their study included adult patients (18-65 yrs old) on ofloxacin therapy for at least two weeks who had TB that was resistant to isoniazid and rifampicin but sensitive to second line anti-TB drugs (i.e. strict definition of MDR-TB). Pregnant or breastfeeding women, patients intolerant of ofloxacin or patients on any drugs, other than ARVs, known to interact with ofloxacin PK were excluded.

They researchers found that the PK values of their eight patients with MDR-TB on ofloxacin differed from previous studies, with reduced AUC and Cmax, and prolonged T1/2 and Tmax.

Five patients were HIV-positive (one was female and four male). The woman and two men were on HAART (d4T, 3TC and efavirenz). All eight patients received kanamycin, ethambutol, ethionamide and pyrazinamide. None were on capreomycin, aminosalicylic acid and terizidone.

Obviously this is a very small study, but the results are concerning because they suggest MDR-TB patients are receiving sub-optimal doses of ofloxacin. The authors therefore recommend ofloxacin plasma monitoring in order to maintain therapeutic plasma levels. Larger studies of patients with MDR-TB taking ofloxacin are also needed to ensure that optimal dosages and timing are determined, taking into account the effects of HIV, liver and kidney dysfunction.

Bhaijee and colleagues reported on a drug-induced life-threatening condition related to the commonly prescribed anticoagulant warfarin. [16]

The incidence of warfarin induced skin necrosis is low (estimated 0.01-0.1%), and by 2000, only 300 cases had been reported. Most of these were in patients receiving treatment for venous thromboembolism. This study was a retrospective review of six cases that occurred in GF Jooste Hospital in Cape Town from April 2005 to July 2008. This is a high concentration at one facility for such a rare condition. All patients were HIV-positive women (aged 27 to 42) with venous thrombosis and with active TB coinfecton. Four died, likely from systemic sepsis when resistant bacteria infected their wounds and one of the survivors underwent bilateral mastectomies and extensive skin grafting at a specialist centre. Median time from skin necrosis to death was 43 days (range 23-45).

No common pattern was detected: three were on HAART, two had TB-IRIS, two had previous TB. While five had low nadir CD4 counts (range 10-56), one of these (on HAART) has a CD4 count of 396 cells/mm³ at the time of the necrosis. The site of skin necrosis included breasts, buttocks, and thighs.

The authors made four recommendations: (a) active prevention and appropriate management of venous thromboses, (b) parallel heparin therapy for at least the first four days of warfarinisation in patients with venous thrombosis (which they suggest may limit the occurrence of skin necrosis), (c) effective infection control measures, and (d) expedited referral to specialist centres for surgical review for patients who develop this warfarin induced skin necrosis.

Wilkinson and colleagues prospectively analysed their cohort to find immunological differences in drug-sensitive and drug-resistant patients with TB IRIS. [17]

They compared 12 rifampicin-resistant cases (nine had MDR-TB) to 27 case controls. They found no significant differences in the median duration of IRIS, days of HAART to development of IRIS, baseline CD4 count or days of TB treatment prior to HAART between drug-resistant and drug-sensitive groups. They also found no difference between the IFN-gamma spot forming cells/ million PBMCs in response to several M. Tuberculosis antigens (ESAT-6, Acr1, Acr2, 38kDa, PPD and heat killed H37Rv). C reactive protein was elevated in both groups, but without significant difference from each other. The authors concluded that both drug-sensitive and drug-resistant TB-IRIS, are clinically and immunologically indistinguishable, and that the occurrence of TB-IRIS is an opportunity to screen for previously undetected drug resistance.

Comment

While news on TB treatment is hardly breathtaking, some of the studies described above are important and merit further comment.

The model of care described by Atkins, based on the HAART model, demonstrated that there are workable, more affordable and more convenient alternatives to DOT that give patients greater autonomy. It deserves further study, and ideally a randomised trial.

The increased uptake of HAART following TB/HIV integration at Zomba Central Hospital offers further evidence of the importance of integration. Although the data from this study can be used for advocacy, one caution should be noted: increased uptake could also have been linked to the general improvements in the facility over time.

The Hermans data is important. One limitation of the study is that a complex and potentially error-prone method was used to merge two separate databases containing patient data to determine the number of TB events. Nevertheless, this data offers evidence that efavirenz and AZT reduced the risk of TB compared to d4T and nevirapine. Their findings are worth testing in clinical trials and perhaps the CARINEMO-ANRS 12146 trial will provide more insight, at least regarding nevirapine versus efavirenz. Furthermore, if d4T-including regimens offer less protection against TB, it is another reason to limit their use in southern African countries.

Diagnostics data at IAS were more disappointing. Data on several methods were presented, including, but not limited to, acid-fast stain, urine lipoarabinomannan and the quantiFERON-TB Gold In-Tube assay. In the last of these, one study found an association between indeterminate results and increased risk of disease progression, but these patients also had lower median current and nadir CD4 counts, which are both, probably, better predictors. [18]

However, no studies at IAS showed algorithms with a combination of high speed, sensitivity and specificity.

The problem is global. One Cambodian study analysed sensitivity and specificity of smear and culture of urine, stool and lymph node aspirate as well as blood culture. It found they added little additional value. The authors aptly concluded, “In HIV settings, there is an urgent need for simple methods for mycobacterial cultures to detect earlier smear-negative tuberculosis.” [19]

References:

Unless othewise stated, all references are to the Programme and Abstracts of 5th IAS Conference on HIV Pathogenesis, Treatment and Prevention, 19-22 July 2009, Cape Town.

1. Middelkoop K et al. Can antiretoviral therapy contain a previously escalating TB epidemic in a high HIV prevalence community? 5th IAS 2009, Cape Town. Poster abstract CDB041.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=2932
2. Middelkoop K et al. Widespread ART is associated with decline in TB prevalence. 5th IAS 2009, Cape Town. Oral abstract WELBB105.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=3844
3. Lawn et al. Short-term and long-term risk of tuberculosis associated with CD4 cell recovery during antiretroviral therapy in South Africa. AIDS. 2009 Aug 24;23(13):1717-25.
   http://www.ncbi.nlm.nih.gov/pubmed/19461502
4. Van Rie et al. Risk factors for incident tuberculosis six months or more after ART initiation – experience from Johannesburg, South Africa. 5th IAS 2009, Cape Town. Poster abstract TUPEB135.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=1933
5. Pepper DJ et al. Clinical deterioration during TB chemotherapy in HIV-1 infected patients. 5th IAS 2009, Cape Town. Poster abstract LBPED07.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=3674
6. De Bruyn et al. Neutrophil count and risk of tuberculosis in HIV-infected adults. 5th IAS 2009, Cape Town. Poster abstract CDB038.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=1539
7. O’Donnell et al. High incidence of multidrug resistant and extensively drug resistant tuberculosis among South African health care workers. 5th IAS 2009, Cape Town. Poster abstract TUPEB149.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=3139
8. Cox H et al. Prevalence of drug resistant tuberculosis and association with HIV in Khayelitsha, South Africa. 5th IAS 2009, Cape Town. Poster abstract MOPEB019.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=2902
9. Hesseling A et al. High prevalence of drug resistance amongst HIV-exposed and infected children with culture confirmed tuberculosis enrolled on a tuberculosis prevention trial. 5th IAS 2009, Cape Town. Poster abstract MOPEB019.
   http://www.ias2009.org/pag/Abstracts.aspx?AID=1610
10. Van Halsema CL et al. Good tuberculosis treatment outcomes and no evidence of increased drug resistance in individuals previously exposed to isoniazid preventive therapy in a population with high HIV prevalence. 5th IAS 2009, Cape Town. Poster abstract MOPEB021.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=1966
11. Atkins S et al. An evaluation of a new tuberculosis treatment support programme: Implications for integrating TB and HIV/AIDS treatment programmes. 5th IAS 2009, Cape Town. Poster abstract WEPED231.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=2944
12. Improved uptake of antiretroviral therapy (ART) following integration of TB and HIV services in a district in southern Malawi. 5th IAS 2009, Cape Town. Poster abstract CDD062.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=2166
13. Bhatt NB et al. Preliminary safety results of co-administration of nevirapine (NVP) or efavirenz (EFV), and rifampicin (RMP) in HIV-tuberculosis (TB) co-infected patients in Maputo (Mozambique): CARINEMO-ANRS 12146 Trial. 5th IAS 2009, Cape Town. Poster abstract MOPEB032.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=589
14. Hermans SM et al. The use of efavirenz is associated with a decreased incidence of tuberculosis after antiretroviral therapy initiation in an urban HIV clinic in sub-Saharan Africa. 5th IAS 2009, Cape Town. Poster abstract TUPEB136.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=2109
15. Mugabo P et al. Determination of plasma concentrations using LC/MS and pharmacokinetics of ofloxacin in patients with multi-drug resistant tuberculosis and in patients co-infected with multi-drug resistant tuberculosis and HIV/AIDS. 5th IAS 2009, Cape Town. Abstract CDB101.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=1916
16. Bhaijee F et al. Warfarin induced skin necrosis in HIV-1 tuberculosis patients with venous thrombosis: a case series. 5th IAS 2009, Cape Town. Poster abstract CDB115.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=1358
17. Wilkinson KA. Immunological analysis of the overlap between Tuberculosis immune reconstitution inflammatory syndrome (TB-IRIS) and antitubercular drug resistance. 5th IAS 2009, Cape Town. Poster abstract TUPEB130.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=1448
18. Aichelburg MC et al. Indeterminate results of the QuantiFERON-TB Gold In-Tube assay indicate an increased risk for disease progression in HIV-1-infected individuals. 5th IAS 2009, Cape Town. Poster abstract TUPEB131.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=1546
19. Pe R et al. Diagnostic evaluation of smear microscopy of different samples as screening tool for tuberculosis in HIV patients at Sihanouk Hospital Center of Hope, Phnom Penh, Cambodia. 5th IAS 2009, Cape Town. Poster abstract CDA102.
    http://www.ias2009.org/pag/Abstracts.aspx?AID=1714