This is, to our knowledge, the first report on MDR-TB treatment outcomes from a cohort of HIV-infected patients in India 
. Despite India's large burden of MDR-TB there are discouragingly few reports of successful treatment programs from the sub-continent 
. A recent study from the Tuberculosis Research Center in Chennai, reported good MDR-TB treatment outcomes in a cohort of 38 patients, however, HIV-infected patients were excluded from the study 
. Our study of a Mumbai cohort of HIV-infected patients only, shows encouraging results in terms of survival, cure rates, culture conversion and immunological recovery, especially considering high rates of fluoroquinolone resistance and previous second-line treatment in this cohort. MDR-TB treatment programs with largely HIV-negative patients have reported cure rates and death rates of 61–77% and 5–19% respectively 
. The early results of a treatment program in Lesotho 
, however, showed that MDR-TB treatment outcomes in HIV-infected cohorts were likely to be significantly worse: overall, 29% of patients died. There was a trend towards poorer outcomes in HIV-infected individuals, with significant mortality in the first few weeks after DR-TB treatment initiation; these results were consistent with previous studies showing increased mortality in HIV-positive compared to HIV-negative patients with drug-susceptible TB 
. We observed lower mortality rates in our cohort of co-infected patients; however, the high defaulter rate recorded may mask unreported mortality. The high early mortality seen in the cohorts of HIV-infected patients argues strongly for early initiation of both ART and second-line TB drugs.
In Mumbai, MDR-TB treatment was until recently (July 2010) only available through an unregulated, private health sector 
. However, the private sector does not generally consider the risk of resistance amplification or the ongoing challenges of adherence to a very long and costly treatment regimen fraught with a high pill burden and significant adverse events. Private physicians rarely (if ever) use DOT as treatment strategy. The particular setting of Mumbai, a mega-metropolis where the marginalised may have inadequate nutrition, rudimentary sanitation and poor home-ventilation is particularly challenging. Difficult transportation and an extremely mobile population, further complicates the management and the follow-up of these patients. Thirty-seven percent of the patients in our cohort had already received second-line TB treatment before they were enrolled in the MSF clinic. According to their medical records and referral notes (when available), they were often on erratic regimes, under-dosed and had their treatment interrupted as they could not afford the high cost of second line anti-TB drugs for long periods of time 
The Mumbai MSF clinic has been trying to address these MDR-TB challenges since the program started in 2007. The outcomes recorded so far in the program demonstrate the feasibility of treating HIV/MDR-TB co-infected patients in such a setting. Given the high rate of resistance to fluoroquinolones in India 
and erratic previous treatment for many of the patients the outcomes are reasonable. We suggest that the following factors contributed to these results: first, all patients received DOT thanks to a network of individual DOT providers, allowing each patient to easily access care within walking distance from his/her home. By moving treatment into the community, it was possible to overcome problems of patient access to clinics. Second, the monitoring plan for early identification of adverse events, that included systematic clinical and laboratory assessment, and the availability of ancillary drugs helped to promptly diagnose and appropriately manage adverse drug effects. Third, therapy followed internationally recommended guidelines with maximally effective drug combinations, based on DST, and at the highest recommended doses 
; this likely reduced the risk of treatment failure and amplification of resistance. Fourth, psychosocial support was given to all patients with monthly home visits and specific support tools were developed. A multidisciplinary team (doctors, nurses, counsellors, psychologists) was essential to manage adherence challenges and psychological issues related to the disease and its treatment, stigma and isolation. Fifth, TB and HIV care were fully integrated: patients received treatment for both conditions in the same clinic as part of a “one-stop service”. This increased convenience for patients, saved time and travel expenses, and facilitated detection of drug interactions and adverse events. Moreover, the integrated approach has broadened expertise in the clinic team on the management of co-infected patients. Sixth, ambulatory treatment and follow-up may have reduced default and the risk of re-infection during hospitalization 
A large investment has been made in the clinic in TB infection control measures that prevent transmission of TB. A comprehensive infection control plan was designed and implemented, including environmental and administrative measures and personal protective equipment for the clinic staff. The clinic defined specific MDR-TB days, carefully organized the spacing between MDR-TB consultations, improved ventilation in consultations rooms, corridors and waiting areas and trained the staff in prevention of TB transmission. An infection control committee was established and it is now planned to expand infection control measures to the community and household level through innovative, small-scale interventions.
However, several challenges remain in this program. The high pre-treatment and early mortality rates highlight the need for intensive case finding strategies, rapid diagnosis and initiation of DR-TB treatment to reduce mortality and improve outcomes. Late referral to the clinic and advanced disease may partly explain these high rates. Studies from HIV programs have shown than even if excellent outcomes can be achieved and maintained among patients receiving antiretroviral therapy, program outcomes such as high pre-treatment mortality and pre-treatment loss-to-follow-up rates remain largely unreported 
. Similarly, failure to initiate therapy and default from treatment rates were high in this cohort, despite the provision of counselling and other psychosocial support. Our data suggests that default from treatment may be lower in more recent cohorts. While the MSF clinic reports very low loss-to-follow-up rates among non-MDR-TB patients on ART (<3%) it is disappointing that the overall rate was 26% for the co-infected patients on treatment. While there are known challenges related to the management of co-infection, several measures may have to be urgently taken to analyze and address this important issue. In our clinic, for example, we identified important gaps in patient support tools and interventions, including lack of adequate emotional and psychosocial support as well as limited access to mental health services. We have also noticed that patients referred to our clinic from the private sector have already been on various treatment regimens for long time periods and often experience treatment fatigue. Support from families is also reported to wane over time.
In this Mumbai cohort, several patients were started empirically on second-line drugs before DST results were available, based on WHO recommendations 
. The RNTCP in India is particularly concerned with empirical treatment and strongly discourages clinicians to initiate treatment without laboratory confirmation. Considering the challenges with the unregulated private sector and the lack of trained and experienced clinicians in the management of MDR-TB in the Indian public sector, these concerns may be well justified. Ideally rather than relying on empiric treatment, efforts should be directed at introducing more rapid diagnosis that will offer both bacteriological confirmation and result in more rapid treatment initiation. The recently approved Xpert MTB/RIF (Xpert) test, has been demonstrated to be feasible in different high burden settings and has the potential to dramatically improve case detection for drug-resistant TB 
. Moreover, a recent study suggested that Xpert MTB/RIF shows good potential in the diagnosis of extrapulmonary TB and this is relevant for HIV-infected individuals 
In Mumbai, a community-based model of MDR-TB treatment was found to be feasible, as has been shown in other countries 
. DOT providers at the community level were trained to provide injections, observe pill-taking, monitor for adverse effects and support and refer patients. Adverse events occurred relatively often in this cohort, similar to a cohort from Lesotho 
. We recorded relatively high rates of hypothyroidism and psychosis. However, with aggressive management of side effects, alterations to the TB regimen were rarely necessary.
There are some limitations to our analysis. Overall, the size of the cohort is relatively small and therefore does not allow definitive conclusions to be drawn regarding factors that may contribute to patient outcomes, particularly the high pre-treatment mortality and default from treatment. Further research aims to investigate these areas in more detail.
However, to date, there is a dearth of documented program experiences from resource-constrained settings treating co-infected patients; the global cohort of HIV/MDR-TB co-infected patients being treated is still discouragingly small.
Despite these limitations, and despite high fluoroquinolone resistance and extensive prior second-line treatment, this report provides encouraging evidence of the feasibility of offering ambulatory MDR-TB treatment in HIV-infected populations, even in a challenging, urban, resource-constrained setting. This community-based treatment model may have been successful in reducing mortality of HIV/MDR-TB co-infected patients. Rapid scale-up of both antiretroviral therapy and second line treatment for MDR-TB is needed to ensure survival of co-infected patients, and control the emerging epidemic of drug resistant TB in Mumbai.