Report from the XVI International AIDS Conference

The 2006 XVI International AIDS Conference took place in Toronto, Canada, from August 13 to 18. The largest IAS conference to date, it brought an estimated 26,000 registrants to the city’s convention center and hotels. The meeting had a characteristically broad agenda of clinical, basic, and social science.

The large attendance posed several logistical challenges for conference organizers, most notably at registration (some participants reported waiting more than 2 hours to receive conference materials) and at plenary sessions with famous keynote speakers such as Bill Gates and Bill Clinton. Despite these irritations, the meeting was highly worthwhile, with a significant number of new studies presented that are very relevant to HIV clinical care and prevention.

The physician-editors of AIDS Clinical Care have summarized some of the most important studies below. For additional conference coverage and web casts, go to http://www.kaisernetwork.org/aids2006/. — Paul E. Sax, MD

ACTG 5142 COMPARES CLASS-SPARING REGIMENS IN TREATMENT-NAIVE PATIENTS

Riddler and colleagues added to the search for the “best” antiretroviral combination for drug-naive patients with ACTG 5142, a large, open-label trial of three different class-sparing treatment strategies for antiretroviral-naive patients [Abstract THLB0204].

A total of 753 patients with relatively advanced disease (median baseline CD4 count, 182 cells/mm³) were randomized to receive lopinavir/ritonavir with two NRTIs (L/R+2NRTI), efavirenz with two NRTIs (E+2NRTI), or lopinavir/ritonavir with efavirenz (L/R/E). The L/R+2NRTI group had a significantly shorter time to virologic failure than the other two groups. At 96 weeks, 77% of the L/R+2NRTI group had viral loads <50 copies/mL, in contrast to 83% of the L/R/E group and 89% of the E+2NRTI group. However, both lopinavir groups had significantly higher increases in CD4-cell count than the E+2NRTI group; in addition, subjects randomized to E+2NRTs were more likely to develop 2-class resistance on virologic failure. Treatment-limiting toxicity rates were similar in all groups. Other information from the study, such as results of sequential DEXA scans, was not included in this presentation.

Combined with other findings (see ACC Sep 1 2006), these results support a standard efavirenz+2NRTI regimen as particularly efficacious for viral suppression. How long-term clinical efficacy and side effects such as lipoatrophy will affect the cost/benefit balance of these initial-treatment strategies awaits further analysis. — Abigail Zuger, MD

“KLEAN” RESULTS WITH FOSAMPRENAVIR/RITONAVIR

For several years, lopinavir/ritonavir has been the preferred PI in U.S. Department of Health and Human Services (DHHS) treatment guidelines. At a late-breaking session, J. Eron reported results of a randomized trial comparing lopinavir/r (400 mg/100 mg twice daily) with fosamprenavir/ritonavir (700 mg/100 mg twice daily), each with abacavir/3TC as the NRTI backbone, in 878 treatment-naive subjects [Abstract THLB0205]. The data from this trial, known as KLEAN (Kaletra versus Lexiva with Epivir and Abacavir in ART-Naïve patients), are also reported in the Lancet (2006; 368:476).

At baseline, each group had a median viral load of 5.1 log₁₀ copies/mL, with a viral load >100,000 copies/mL in about 54%. The two groups also had similar median CD4 counts (about 190 cells/mm³), with a count <50 cells/mm³ in about 17%.

At 48 weeks, the fosamprenavir/r and lopinavir/r groups had similar percentages of patients with a viral load <400 copies/mL (73% and 71%, respectively) and a viral load <50 copies/mL (66% and 65%). Response rates were consistent across baseline viral-load and CD4-cell–count strata. Rates of resistance on virologic rebound were low in both groups, with no development of phenotypic resistance to PIs. Notably, the two groups had comparable rates of treatment-related adverse events (including gastrointestinal symptoms) and of grade 3/4 laboratory abnormalities, with nearly identical increases in fasting total-, LDL-, and HDL-cholesterol levels, as well as triglyceride levels.

In this well-designed study, treatment-response and adverse-event rates were remarkably similar with twice-daily ritonavir-boosted fosamprenavir and boosted lopinavir. The study also adds substantially to the evidence on once-daily abacavir/3TC as an NRTI backbone. Given the findings, ritonavir-boosted fosamprenavir likely will join lopinavir/r as a “preferred” initial PI in the DHHS guidelines, as it has already in the new IAS treatment guidelines (JAMA 2006; 296:827). This study’s one unavoidable limitation is that enrollment began in May 2004, so subjects received the now-discontinued soft-gel formulation of lopinavir/r, which has a higher pill burden and a somewhat greater adverse-effect profile than the newer tablet formulation. — Paul E. Sax, MD

WHICH ANTIRETROVIRAL REGIMEN FIRST? DATA FROM CPCRA

Long-term clinical outcomes are the most relevant measure of therapeutic success. On behalf of the Community Programs for Clinical Research on AIDS (CPCRA) team, R.D. MacArthur reported data from a randomized trial comparing the long-term consequences of three different antiretroviral regimens as initial therapy: (1) NRTI + PI, (2) NRTI + NNRTI, and (3) NRTI + NNRTI + PI [Abstract TUAB0102]. Within each regimen, the individual drugs were chosen by clinicians and patients. The primary endpoint for the NNRTI versus PI comparison was a composite of AIDS events, death, and CD4-count decline to <200 cells/mm³; the primary endpoint for the 3-class versus 2-class comparison was CD4-cell count change after 32 months.

This diverse study population of 1397 antiretroviral-naive patients (21% women, 54% black) had a surprisingly low median CD4 count of 163 cells/mm³ at baseline. During a median follow-up of 5 years, 302 participants had an AIDS event or died, 188 died, and 388 developed the composite clinical/CD4-count endpoint. In the 2-drug comparison, the NNRTI-based and PI-based groups had similar hazard ratios (HRs) for the composite endpoint and for AIDS or death, but the NNRTI-based regimen was associated with a significantly lower HR for virologic failure (HR, 0.66), defined as >1000 copies/mL at 4 months.

The two 2-class groups did not differ significantly from the 3-class group in mean CD4-count increases by 32 months (+227 cells/mm³ and +234 cells/mm³, respectively) or in the HR for AIDS or death. An advantage of 3-class therapy in virologic failure did not quite reach significance (HR, 0.87; 95% CI, 0.75–1.00); results were consistent by baseline CD4 count (<200 vs. ≥200 cells/mm³). However, the risk for toxicity-related discontinuation was significantly greater with 3-class than with 2-class therapy (HR, 1.58).

NNRTI-based and PI-based strategies as initial therapy did not differ significantly for a composite outcome that included CD4-cell count decline, AIDS events, and death at a median follow-up of 5 years, although the NNRTI-based strategy yielded superior virologic suppression. Immunologic and clinical outcomes were not significantly better with a 3-class strategy than with a 2-class strategy, even in patients with low baseline CD4-cell counts. Further, the 3-class strategy produced more drug toxicity. Given that 60% of participants beginning a PI-based regimen were started on nelfinavir, an agent now rarely used for initial PI-based therapy, the study's findings may not be fully generalizable to current practice. — Gerald H. Friedland, MD

EFAVIRENZ FOR HIGH VIRAL LOAD AND LOW CD4 — A SECONDARY ANALYSIS OF ACTG 5095

Although efavirenz plus two NRTIs is widely used as first-line therapy, concern persists about the virologic potency of such regimens in patients with advanced HIV disease, who typically have high viral loads and low CD4-cell counts. ACTG 5095 was designed to evaluate three PI-sparing regimens for initial treatment of HIV: AZT/3TC/abacavir, efavirenz + AZT/3TC, and efavirenz + AZT/3TC/abacavir.

In this secondary analysis of ACTG 5095 data, researchers examined the effects of baseline viral load and CD4-cell count on virologic and immunologic outcomes in the two efavirenz arms and assessed whether the addition of a third NRTI affected responses. The two efavirenz arms included 765 participants, who were followed for a median of 144 weeks. Participants were stratified by viral load and CD4-cell count at baseline. Seventeen percent had viral loads between 100,000 and 300,000 copies/mL, and 24% had values >300,000 copies/mL; 20% had CD4 counts <50 cells/mm³.

In an intent-to-treat analysis, risk for virologic failure was similar among the viral-load and CD4-cell subgroups, except that it was slightly higher in participants with the highest CD4 counts (>500 cells/mm³). This difference was explained by the withdrawal of patients who decided after enrollment that they did not want to be treated, and it is supported by results from the as-treated analysis, which showed no difference among the groups. In a Cox proportional-hazards analysis that used the highest viral load (>300,000 copies/mL) and lowest CD4-count (<50 cells/mm³) groups as the reference, no statistically significant differences were seen. That is, the risk for virologic failure did not differ across the range of baseline viral loads and CD4-cell counts.

No difference in the risk for virologic failure was seen between the 3- and 4-drug regimens, regardless of baseline viral load or CD4-cell count. So adding abacavir to AZT/3TC + efavirenz did not affect the risk for virologic failure, and, once again, four drugs failed to be better than three. This study clearly shows that an efavirenz-containing regimen with two NRTIs is effective across the full range of initial viral loads and CD4-cell counts, at least for up to 3 years. — Judith Feinberg, MD

LOPINAVIR/RITONAVIR MONOTHERAPY

Three years ago, a promising pilot study aroused interest in lopinavir/ritonavir monotherapy (ACC Nov 1 2003). Half a dozen presentations this year examined this treatment option from different angles.

In a late-breaker session, Delfraissy reported on an industry-sponsored trial of lopinavir/ritonavir monotherapy as initial treatment for antiretroviral-naive subjects [Abstract THLB0202]. At week 48, viral loads <50 copies/mL were seen in significantly fewer monotherapy recipients than patients on lopinavir/ritonavir combined with AZT/3TC (84% vs. 98%); more episodes of low-level viremia (50–400 copies/mL) were seen in the monotherapy group.

Two late-breaker presentations reported industry-sponsored studies of lopinavir/ritonavir monotherapy maintenance after successful viral suppression with multidrug regimens. Cameron and colleagues randomized antiretroviral-naive patients to lopinavir/ritonavir or efavirenz, both together with AZT/3TC; after successful viral suppression, lopinavir/ritonavir recipients were maintained on monotherapy [Abstract THLB0201]. At 96 weeks, 50% of the monotherapy group and 61% of the three-drug group had sustained viral suppression (viral load <50 copies/mL), a nonsignificant difference. Most of the virologic breakthroughs that were counted as failures with lopinavir/ritonavir monotherapy were low-level and transient.

Arribas and colleagues reported data from a study in which patients with long-term virologic suppression were randomized to continue a standard three-drug lopinavir/ritonavir-based regimen or to receive lopinavir/ritonavir alone [Abstract THLB0203]. After 48 weeks, similar percentages of both groups had viral loads <50 copies/mL. Of six monotherapy patients with virologic failure, two had PI resistance and four had successful virologic resuppression with nucleosides.

In another industry-supported trial, Nunes and colleagues found that of 60 patients with successful virologic suppression on three-drug therapy, patients subsequently randomized to lopinavir/ritonavir monotherapy and those on a three-drug lopinavir/ritonavir-based regimen had similar rates of continued suppression at 48 weeks (83% vs. 87%) [Abstract TUAB0103].

In addition, two posters described nonrandomized clinical experience with lopinavir/ritonavir monotherapy [Abstracts THPE0132 and THPE0134]. Fifty percent of patients in one series and 66% in the other had undetectable viral loads after about a year.

In sum, lopinavir/ritonavir monotherapy appears inferior to three-drug regimens for induction and is probably not quite as powerful for maintenance. In addition, given how rarely lopinavir/r therapy in combination is associated with PI resistance, there is a suggestion that lopinavir/r monotherapy may be associated with greater risk for this complication. On the other hand, all these reports showed that at least 50% of patients will maintain virologic suppression on monotherapy, making it a reasonable last-ditch option for selected patients who cannot tolerate more-traditional regimens. It remains to be determined whether adopting this monotherapy strategy will limit drug toxicity and reduce costs. — Abigail Zuger, MD

SMART AND DART: INTERMITTENT ANTIRETROVIRAL TREATMENT IS SUBOPTIMAL AND SHOULD BE AVOIDED

Data from two structured treatment interruption studies — both prematurely halted based on safety concerns — showed that episodic antiretroviral therapy is inferior to continuous therapy across a number of subgroups.

In the SMART (Strategies for Management of Antiretroviral Therapy) study, 5472 patients were randomized to either treatment interruption (stopping therapy at CD4 counts >350 cells/mm³ and restarting therapy at CD4 counts <250 cells/mm³) or continuous therapy. Although the proportion of patients who experienced disease progression or death at a CD4 count <350 cells/mm³ was similar in both groups, the proportion of patients who reached this combined endpoint at a CD4 count ≥350 cells/mm³ was significantly greater among those in the treatment-interruption group compared with the continuous-treatment group (P<0.05) [Abstract WEAB0203]. Overall, participants in the treatment-interruption arm were 2.5 times more likely to experience opportunistic illnesses or death. Race appeared to play a role in disease progression; the hazard rate was higher in blacks (3.6 vs. 2.0; P=0.02) than in patients of other races [Abstract WEAB0204]. Notably, the rate of adverse outcomes was higher in the treatment-interruption arm than in the continuous-therapy arm.

The decline in CD4-cell count following interruption of therapy in the SMART study was steepest during the first 2 months and was associated with high CD4-cell count at study entry, low CD4 nadir, baseline viral load <400 copies/mL, and prior AIDS (P<0.01 for each) [Abstract THPE0144]. Not only were patients in the treatment-interruption group more likely to experience disease progression, but they also showed significantly worsened physical functioning, general health perception, and energy scores [Abstract THPE0145].

The Development of Antiretroviral Therapy in Africa (DART) researchers enrolled 3314 antiretroviral-naive adults with CD4 counts <200 cells/mm³. After at least 2 years on therapy, some 800 patients who had achieved a CD4 count of at least 300 cells/mm³ were randomized to 12-week cycles on and off therapy or to continuous treatment. The rate of development of new or recurrent WHO stage 4 events or death was 3.2 per 100 person-years with continuous treatment versus 8.3 per 100 person-years with treatment interruption. The treatment-interruption strategy was associated with a 2.6-fold increased rate of disease progression (95% confidence interval, 1.4–5.1) [Abstract THLB0207].

Based on rates of clinical progression, adverse events, quality-of-life measures, and other criteria, patients on continuous therapy do better than those who interrupt treatment. Episodic therapy should be avoided until more can be learned about the reasons for the poor outcomes in these studies. — Salim S. Abdool Karim, MD

CCR5 INHIBITORS

Vicriviroc

R. Gulick presented data from the ACTG phase II study of Schering’s vicriviroc (VCV) [Abstract THLB0217]. The investigators randomized 118 treatment-experienced adults (92% men; 66% white; 33% T-20–experienced) with viral loads ≥5000 copies/mL and R5-only phenotype (assessed by Monogram assay) to receive VCV (at 5, 10, or 15 mg daily) or placebo for 14 days. Then, all participants added a background regimen, including a ritonavir-boosted PI, optimized according to genotypic and phenotypic resistance tests. At baseline, the median viral load was 36,380 copies/mL, and the median CD4 count was 146 cells/mm³.

Due to suboptimal virologic response and a nonsignificant trend toward increased R5-to-X4 coreceptor switching, all 5-mg VCV recipients had their doses increased to 15 mg/day. By 14 days, the mean viral load had decreased by about –1.5 log₁₀ copies/mL in all VCV groups, with a cumulative reduction to –1.7 to –2.0 log₁₀ by 24 weeks in the two highest-dose VCV groups; the placebo group had no decrease in viral load by 14 days and a negligible one (–0.3 log₁₀ copies/mL) by 24 weeks. A ≥90% decrease in viral load by 24 weeks was achieved by 61% to 73% of VCV recipients, depending on the dose, and 18% of placebo recipients; 26% to 40% of VCV recipients and 7% of placebo recipients achieved a viral load <50 copies/mL. Mean CD4 counts changed little with placebo, but increased by 75 to 135 cells/mm³ with VCV.

Although all subjects had R5-only virus at screening, 10% had evidence of dual/mixed (D/M) tropism upon enrollment; this subgroup’s decline in mean viral load was significantly smaller than that of the rest of the cohort (–0.77 vs. –1.83 log₁₀ copies/mL), according to an analysis that included only the 10- and 15-mg VCV groups. Changes in tropism occurred in 13 of 106 patients assessed: 1 placebo recipient and 7, 3, and 2 of the 5-, 10-, and 15-mg VCV recipients, respectively.

The four groups did not differ in their rates of grade 3/4 adverse events, but five VCV recipients and two placebo recipients developed malignancies. Subsequent events in the five VCV recipients, two of whom had previous diagnoses of lymphoma, included gastric adenocarcinoma, Hodgkin disease, and non-Hodgkin lymphoma. The two placebo recipients subsequently developed squamous-cell carcinoma.

VCV had a potent antiviral effect at 14 days that, with the addition of optimized background therapy, was sustained at 24 weeks, although subjects with D/M tropism had a poorer virologic response. The question of a possible link to increased risk for malignancy remains unanswered. The trial is ongoing.

Maraviroc

H. Mayer presented the 24-week results of an ongoing 48-week phase II study of Pfizer’s maraviroc in 190 patients with triple-class experience or resistance to two classes [Abstract THLB0215]. Nearly 90% of subjects had D/M CCR5/CCX4-tropic virus at baseline, assessed by Monogram assay. In this double-blind trial, subjects were randomized to receive 150-mg maraviroc (either once or twice daily) or placebo, plus optimized background therapy (OBT) comprising 3 to 6 agents. OBT included a PI in about 80% of subjects and T-20 in about half of subjects. Each treatment group had a median baseline CD4 count <50 cells/mm³ and a median baseline viral load of 5.0–5.1 log₁₀ copies/mL.

At 24 weeks, the mean decrease in viral load among patients with D/M-tropic virus at baseline did not differ greatly among the groups: –0.97 log₁₀ copies/mL with placebo, –0.91 log₁₀ copies/mL with once-daily maraviroc, and –1.20 log₁₀ copies/mL with twice-daily maraviroc; decreases were greater in subjects who also received T-20 (–0.89, –1.26, and –1.44 log₁₀ copies/mL, respectively). Across the three groups, similar percentages of patients achieved viral loads <400 copies/mL and <50 copies/mL. However, mean increases in CD4 count were greater with once-daily maraviroc (+60 cells/mm³) and twice-daily maraviroc (+62 cells/mm³) than with placebo (+36 cells/mm³), although none of these increases was associated with a change in CD4-cell percent.

Twelve subjects in each maraviroc arm had X4-only virus at the time of virologic failure, compared with two subjects in the placebo arm. Maraviroc was well-tolerated, with no cases of lymphoma or adenocarcinoma. Despite the selection of X4-only virus, maraviroc-treated patients actually had a greater CD4-cell–count increase than the placebo group — an unexpected result given the known association between the emergence of X4 virus and more rapid disease progression. — Judith Feinberg, MD

INTEGRASE INHIBITORS

Integration of virally derived DNA into a host-cell chromosome is an essential step in the life cycle of HIV-1. The enzyme that catalyzes this step, HIV integrase, is the last of the viral enzyme targets to be exploited by antiretroviral agents. Currently-tested drugs interfere with DNA strand transfer, the second of three steps in the integration process during which viral cDNA is inserted into target DNA in the host cell.

M. Markowitz presented 24-week results from the phase II trial of Merck’s novel HIV-1 integrase inhibitor MK-0518 in treatment-naive patients [Abstract THLB0214]. For an initial 10 days, four groups of 8 patients each received twice-daily doses of MK-0518 monotherapy (100, 200, 400, or 600 mg), yielding mean viral-load declines ranging from –1.7 to –2.2 log₁₀ copies/mL. Then, an additional 30 patients were enrolled in each dose arm; they received tenofovir/3TC plus the specified dose of MK-0518. A control group received tenofovir/3TC plus efavirenz. Most patients were male and nonwhite; mean baseline CD4 counts ranged from 271 cells/mm³ to 338 cells/mm³, and viral loads from 4.6 to 4.8 log₁₀ copies/mL.

At 24 weeks, response rates were similar across all the MK-0518 arms, and between MK-0518 and efavirenz. Interestingly, significantly more MK-0518 recipients than efavirenz recipients achieved viral loads <50 copies/mL at weeks 4 and 8, indicating faster viral suppression with the integrase inhibitor. CD4-cell count increases were similar in all the treatment arms. MK-0518 was very well-tolerated at all doses; nausea was the most common adverse event with MK-0518 (11% of recipients vs. 13% of efavirenz recipients). Grade 3/4 laboratory abnormalities were uncommon, although liver-enzyme elevations caused one recipient of 600-mg MK-0518 to withdraw from the study.

No clinical trial data for the Gilead Sciences integrase inhibitor GS-9137 were reported, but results from drug-interaction studies involving it and AZT, tenofovir, and FTC were presented in posters [Abstracts TUPE0080 and TUPE0088]. The three NRTIs were administered in standard doses with GS-9137 (dose range, 50–200 mg), boosted with once-daily ritonavir (100 mg). No clinically significant changes in NRTI or GS-9137 concentrations were noted.

At this point, integrase inhibitor clinical trial data are available primarily for MK-0518, and the early results look extremely favorable. Since MK-0518 is metabolized by glucuronidation and is not a substrate for CYP3A, the potential for problematic drug–drug interactions appears low. The speed of viral-load decline by 24 weeks was remarkable, and the adverse-event rate was very low, making this a most promising new drug. We eagerly await MK-0518 data on longer-term outcomes in treatment-naive patients and initial data in treatment-experienced patients, as well as data from clinical trials of GS-9137. Merck is developing an expanded access program for MK-0518 that is likely to be open in late 2006 or early 2007. — Charles B. Hicks, MD

PREEXPOSURE PROPHYLAXIS STUDIES MOVE FORWARD

After his keynote address at the 2002 Retrovirus conference, Bill Gates was asked for his thoughts about offering prophylactic antiretroviral therapy to uninfected high-risk individuals. “Wouldn’t it be simpler to use condoms?” he famously replied. However, the strategy of providing preexposure prophylaxis (PrEP) has moved forward in several clinical studies conducted at multiple international sites. One of those studies was presented at the conference by L. Peterson [Abstract THLB0103]; it was funded in part, ironically, by the Gates Foundation.

The study was a double-blind, randomized, placebo-controlled trial of Gilead’s tenofovir disoproxil fumarate for prevention of HIV infection in women ages 18 to 35 in Cameroon, Nigeria, and Ghana. All participants reported at least three coital acts per week and at least four sexual partners per month, and had normal baseline renal and hepatic function. Of 2040 women screened, 936 were enrolled from June 2004 to March 2006 and were randomized to receive tenofovir or placebo. All participants received free condoms and intensive behavioral counseling about HIV transmission.

During up to 12 months of follow-up, two tenofovir recipients and six placebo recipients seroconverted, a nonsignificant difference. In one of the tenofovir recipients who seroconverted, resistance was not detected; resistance data were not available for the other. The rate of serious adverse events was similar in the two groups, and no patients experienced creatinine elevations (creatinine >2.0 mg/dL). In the 56 patients who were HBsAg-positive at baseline, liver function remained stable after tenofovir cessation.

This study was not large enough to prove conclusively that tenofovir protects against HIV infection, but the results are encouraging regarding both the safety of this approach and its feasibility in developing countries. However, as with other PrEP studies, controversy surfaced at some of the sites regarding the guaranteed provision of antiretrovirals to participants who acquired HIV during the trial. The presenter did not mention this controversy.

Controversies notwithstanding, multiple similar studies are ongoing. In an excellent overview of HIV prevention trials, Ramjee cited four PrEP studies from Thailand, the U.S., Botswana, and Peru [Abstract TUPL02]. They are expected to enroll 4400 participants and to report results in 2008 or 2009. — Paul E. Sax, MD

EXTENSIVELY DRUG-RESISTANT TB IN HIV/TB–COINFECTED PATIENTS IN RURAL SOUTH AFRICA

The co-occurring epidemics of HIV and tuberculosis in sub-Saharan Africa have taken a large toll on the population and on the already fragile healthcare systems of that region. N.R. Gandhi, representing a U.S. and South African team, presented a late-breaking study [Abstract THLB0210] about an ominous development: drug-resistant TB among HIV/TB–coinfected people at a hospital in rural KwaZulu Natal, South Africa, a district where the prevalences of both HIV infection and TB are very high.

Patients with newly diagnosed HIV and TB in this district had an annual mortality rate of 40% prior to antiretroviral therapy. Recently introduced antiretrovirals have dramatically reduced mortality. However, in an ongoing study of TB and HIV treatment, 67% of deaths in coinfected patients were found to be due to multidrug-resistant (MDR) TB, defined as resistance to both isoniazid and rifampin [Abstract MOPE0181]. To learn more, the researchers obtained sputum cultures and conducted drug-susceptibility testing in patients with known or suspected TB from January 2005 to March 2006.

Sputum collected from 1540 patients revealed that 536 (35%) were positive for Mycobacterium tuberculosis. Of these, 221 (41%) had MDR TB, and 53 (24% of MDR isolates, 10% of all positive cultures) were resistant to all first- and second-line drugs tested (defined as extensively drug-resistant [XDR] TB). Molecular studies revealed that 90% of patients with XDR TB were infected with a genetically similar strain, unique to South Africa. The infection appeared to be concentrated in patients with HIV, as all 47 XDR patients with known HIV status were HIV-positive. Further, 52 of the 53 XDR TB patients died shortly after sputum collection. Notably, only 30% of patients with XDR TB had been previously treated for TB, and 56% had been hospitalized within the past 2 years. Finally, several hospital staff members acquired the infection and died.

This study documents markedly greater MDR TB prevalence than previously recognized and an unprecedented, untreatable, and lethal presence of XDR TB among HIV patients. Prior to this report, only 347 cases of XDR TB had been identified worldwide from 2000 through 2004. In the current study, there is strong evidence of recent nosocomial transmission of XDR TB given that (1) most patients had not been previously treated for TB and, therefore, had not acquired TB resistance from incomplete treatment; (2) a single genetic strain was identified as causing almost all XDR infections; and (3) hospital staff acquired and died from the infection. The convergence of the HIV/TB coinfection epidemic with MDR and XDR TB in resource-poor settings is a deadly threat to survival gains achieved by DOTS (Directly Observed Treatment, Short Course) for TB and by antiretroviral therapy. Further study is now needed to determine the full extent of the problem and strategies to reduce its consequences. — Gerald H. Friedland, MD

Dr. Friedland was the senior investigator for this study.