Management of HIV-associated tuberculosis in resource-limited settings: a state-of-the-art review

Screening for TB in those living with HIV infection

In high burden settings, much prevalent TB disease remains ‘below the radar’ in those living with HIV. Postmortem studies conducted in hospitals across sub-Saharan Africa over the past 20 years have repeatedly shown that between 30% and 50% of HIV-infected adult inpatients who die have postmortem evidence of TB, much of which was neither clinically suspected nor diagnosed before death [14–17]. These studies have highlighted the abject failure of the diagnostic process and the low sensitivity of diagnostic tools available [18]. In the absence of more sensitive means of diagnosis, management algorithms for suspected sputum smear-negative disease were developed [19–21] and studies of empirical TB treatment for certain high risk patient groups with advanced immunodeficiency are being conducted [22].

However, in recent years there have been significant advances in screening and diagnosis. Traditional symptom screening for pulmonary TB based on chronic cough has low sensitivity for HIV-associated TB [23, 24]. A new WHO symptom screening tool for HIV-associated TB (one or more of the following symptoms: cough, fever, weight loss or night sweats, each of any magnitude or duration) has much higher sensitivity and is recommended for routine screening of those in HIV care at each visit [25]. However, in view of its low specificity, further research is needed to define which of the large number of patients with a positive screen should be prioritized for subsequent microbiological testing of clinical samples.

New diagnostic tools have also increased our capacity for microbiological diagnosis. This includes the Xpert MTB/RIF assay, which was endorsed by WHO in 2010. A single test is able to detect all sputum smear-positive disease, approximately 70% of smear-negative pulmonary disease and provides rapid simultaneous screening for RIF resistance [26]. In addition, this assay can be used to test a wide range of extrapulmonary sample types [26, 27]. The Xpert MTB/RIF assay has been incorporated into the national guidelines of many high burden countries. In South Africa, which alone accounts for approximately 30% of the global burden of HIV-associated TB, sputum smear microscopy has now been replaced by Xpert MTB/RIF as the initial diagnostic test for TB [26].

Determine TB-LAM is a low-cost, point-of-care lateral-flow (‘strip test’) assay that diagnoses TB through detection in urine of lipoarabinomannan (LAM): a lipopolysaccharide component of the M. tuberculosis cell wall [28]. It has high specificity whereas sensitivity is very strongly CD4 count dependent, at best detecting approximately two-thirds of cases in those with CD4 counts <50 cells/μl [28–31]. This assay therefore allows rapid (<30 minutes) bedside diagnosis among those who have the highest mortality risk [32]. The growing evidence base on this assay will be reviewed by WHO in 2014. Its role is likely to be as an add-on test within the diagnostic algorithm to permit point-of-care diagnosis and immediate TB treatment among patients with advanced immunodeficiency (CD4 counts <200 cells/μl) following admission to hospital or enrolling in ART clinics [28, 31].

Screening for HIV in those with TB or possible TB

A major step forward in improving HIV testing rates in patients with TB was the switch from voluntary counseling and testing (VCT) to provider-initiated testing and counseling (PITC) in 2007 [33]. With PITC, all patients undergo routine testing unless they specifically opt out. Testing has increased globally from 3.1% in 2004 to 40% of notified TB cases in 2011, but falls well short of the goal of universal testing [1]. Testing rates have reached 69% in Africa, >50% in the Americas and 32% in South-East Asia. In African countries, the proportion of TB patients testing positive is 46% overall (range, 8% to 77%) and exceeds 50% in ten counties in the south and east of the continent [1]. A further significant policy change has been to expand PITC to include all patients being investigated for TB regardless of whether or not TB is diagnosed [10, 12]. This change resulted from the observed high HIV prevalence and mortality among those presenting for investigation of possible TB even when this diagnosis was subsequently excluded [34]. It is critical, however, that improved testing rates are accompanied by improvement in the delivery of appropriate management.

Pharmacokinetic interactions with first-line ART

Although RIF induces the expression of cytochrome P450 2B6 (CYP2B6), which comprises the main metabolic pathway for EFV, studies have failed to demonstrate significantly reduced concentrations of EFV with concomitant RIF-based TB treatment [50–53]. This is consistent with the observed virological responses which are excellent in patients receiving RIF-based TB treatment treated with standard 600 mg daily doses of EFV [54–57] and were better than those in TB patients randomized to NVP-based ART in the recent CARINEMO trial [56]. Similarly, lowering the dose of EFV to 400 mg daily in the ENCORE1 trial did not compromise outcomes in non-TB patients [58]. Thus, although the US Federal Drug Administration (FDA) [59] recommends that the dose of EFV during RIF treatment is increased in adults weighing more than 50 kg, this is not supported by studies in TB patients [53] and is not recommended by the WHO for resource-limited settings.

Conversely, however, among patients with a slow CYP2B6 metabolizer genotype, EFV concentrations are increased during TB treatment, possibly due to inhibition by INH of accessory pathways metabolizing EFV [60, 61]. This genotype is relatively common in Africa, South-East Asia, and the Caribbean [50–52, 62, 63]. Whether EFV-induced central nervous system (CNS) adverse effects are more frequent during TB treatment or isoniazid preventive therapy in patients with this genotype needs to be evaluated.

NVP is a reasonably safe, acceptable alternative for TB patients unable to tolerate EFV. Through induction of the expression of CYP2B6, RIF treatment reduces NVP concentrations by an average of approximately 40% and NVP-based ART remains inferior to EFV-based regimens in TB patients [56]. During the 14-day lead-in phase of NVP dosing, plasma drug concentrations are very low in patients receiving RIF, potentially predisposing to the development of viral resistance mutations and contributing to an increased risk of virological failure [54]. The CARENIMO trial recently found that NVP was well tolerated when introduced at full doses (200 mg twice a day) in patients with CD4 cell counts <250 cells/mm³ receiving RIF [56]. The use of a dose escalation lead-in phase to avoid toxicity in patients receiving RIF is therefore not recommended.

Triple nucleoside/tide regimens are less effective than NNRTI-based or PI-based regimens particularly in patients with baseline viral loads >100,000 copies/ml [64]. However, small uncontrolled studies suggest they may provide an acceptable regimen for TB patients who have not failed an ART regimen [65, 66] even though the concentrations of abacavir and zidovudine may be reduced by concomitant RIF. This therefore provides an alternative option for those in whom EFV and NVP are contraindicated and integrase inhibitors unavailable.

Pharmacokinetic interactions with second-line ART

With increasing numbers of patients switching to protease inhibitor (PI)-based second-line ART regimens, defining safe and effective approaches to concurrent TB treatment is an urgent challenge. The pharmacokinetic interactions between rifamycins and PIs are extensive. RIF reduces concentrations of ritonavir-boosted PIs by 75% to 90% [67]. Conversely, through potent inhibition of CYP3A4 and p-glycoprotein, high-dose ritonavir offsets the effect of RIF-mediated induction such that ‘superboosting’ of lopinavir or saquinavir (lopinavir/ritonavir 400 mg/400 mg or saquinavir/ritonavir 400 mg/400 mg, twice daily) preserves plasma concentrations of the PI [68–70]. Adequate plasma concentrations of lopinavir are also achieved in adults by doubling the dose of lopinavir/ritonavir in the tablet formulation (to 800/200 mg twice daily); this is the simplest approach, especially in settings where the separate ritonavir is not available [71]. Although these approaches are associated with high rates of hepatotoxicity in studies of healthy volunteers, these seem to be much safer in HIV infected patients [71–76]. Nevertheless, hepatotoxicity, gastrointestinal side effects and poor tolerability are problematic and treatment discontinuation rates of up to nearly 50% have been reported [74, 75].

Rifabutin is an alternative rifamycin to RIF, but data on its use in TB patients receiving ritonavir-boosted PIs are limited. Studies of healthy volunteers show that ritonavir-boosted PIs increase the concentrations of rifabutin approximately fourfold and the concentrations of the active metabolite to an even greater extent. Thus, the dose of rifabutin needs to be reduced. Thrice weekly 150 mg doses of rifabutin in combination with standard doses of lopinavir/ritonavir may be reasonably tolerated [77, 78]. However, contrary to expectations based on pharmacokinetic data from healthy volunteers, small studies in coinfected patients have found that rifabutin 150 mg used thrice weekly in combination with lopinavir/ritonavir resulted in low rifabutin concentrations [79–82]. Such levels would be conducive to acquisition of rifamycin resistance in patients with severe immunosuppression [79, 83] as has been observed with twice weekly doses [84]. Thus, recent US national guidelines recommend a daily 150 mg dose of rifabutin for patients on ritonavir-boosted PIs [85].

There is extremely limited information about the safety or efficacy using rifabutin with PIs and this may vary between populations due to differential increases in rifabutin concentrations. Severe neutropenia and uveitis occur relatively frequently in patients with increased exposures [81, 86] and hepatitis, gastrointestinal symptoms, rashes and anemia are also important safety concerns [87, 88]. While rifabutin is becoming more widely available and affordable, it is not an ideal solution for high burden settings where limited patient monitoring is available and fixed dose drug formulations are preferred. Thus, there is an urgent need for research to define the optimum approaches for the cotreatment of patients with TB who have failed first-line ART, including the use of newer agents.

Pharmacokinetic interactions with newer ART drugs

Ritonavir-boosted darunavir has a favorable safety and tolerability compared to lopinavir/ritonavir and promising efficacy, especially in treatment of ART-experienced patients. A pharmacokinetic study in healthy volunteers suggests that it could be used in standard doses with rifabutin 150 mg thrice weekly, but the drug-drug interactions with RIF have not been studied. Integrase inhibitors have potent antiviral activity and are well tolerated, but any future role in ART programs in low-resource settings is at present undefined. However, initial data on use with TB treatment show promise. Pharmacokinetic studies suggest that doubling the dose of raltegravir to 800 mg twice daily compensates for the effect of RIF on overall exposure [89, 90] and this approach seems to be well tolerated and effective in patients with HIV-associated TB [91]. However, preliminary results of the REFLATE TB study suggest that such dose adjustment may not even be necessary as virological responses were similar in ART-naive TB patients receiving RIF who were randomized to receive 400 mg or 800 mg of raltegravir twice daily or EFV daily [92]. Similar to raltegravir, a pharmacokinetic study of dolutegravir in healthy volunteers suggests that the effect of RIF on antiretroviral therapy can be overcome by increasing the daily 50 mg dose of dolutegravir to 50 mg twice daily and that dose adjustment may not be necessary with rifabutin [93].

Timing of ART initiation during TB treatment

WHO guidelines reflect these findings, recommending that TB treatment should be started first and followed by ART as soon as possible within the first 8 weeks of treatment but within the first 2 weeks for those with profound immunosuppression (CD4 count <50 cells/μl) [11]. However, CD4 count measurements may either be unavailable or be inaccurate in some settings. In addition, within different CD4 count categories, there is great diversity in severity of disease and mortality risk. Thus, where feasible, decisions on timing for individual patients might also be further informed by taking into account clinical criteria such as body mass index, Karnofsky score, severity of anemia and extent of TB. Moreover, national guidelines might best be appropriately tailored for operational simplicity. One possible option, for example, might be to start ART in all patients after 2 weeks of TB treatment, accepting lower risk of mortality but higher risk of TB IRIS.

Patients with HIV-associated TB meningitis represent an important exception. A randomized trial from Viet Nam found no survival benefit from early ART in patients with TB meningitis [97], reflecting the awful prognosis (mortality approximately 60%) of these patients with advanced disease and the dire consequences of TB-IRIS within the confined space of the CNS [99]. Further studies are required in different geographical settings to better define appropriate management of these patients.

Adverse drug reactions and management

In patients without coinfection, DILI (variably defined as, for example, an elevation of alanine aminotransferase to >3 or >5 times the upper limit of the normal range) occurs in 5% to 33% of those receiving TB treatment [100] and in 5% to 11% of those receiving currently recommended ART regimens [101, 102]. HIV infection itself has been identified as a risk factor for DILI in patients receiving TB treatment in some [103, 104] but not all studies [105–108]. Of the currently used ART drugs, NVP is associated with highest risk of DILI; however, EFV and PIs are also recognized causes.

Concurrent TB treatment in patients receiving NNRTI-based ART has been associated with an increased risk of DILI in some [109–111] but not all [54] studies. In one of these, the absolute risk of severe hepatotoxicity in patients receiving EFV-based ART was low, but the risk associated with concurrent TB treatment exceeded that associated with positive hepatitis B surface antigen status [109]. Importantly, a randomized trial of NVP-based versus EFV-based ART in patients receiving TB treatment reported more treatment discontinuations related to DILI in the NVP arm (4 vs 0%) [56].

Development of DILI significantly complicates management of HIV-associated TB. Elevation of alanine transaminase (ALT) concentrations >3 to 5 times the upper limit of normal especially when accompanied by symptoms or jaundice requires that all potentially hepatotoxic medication is interrupted until derangements of liver function tests resolve. Thereafter, rechallenge of first-line TB medication should be considered followed by ART, although rechallenge is generally not undertaken if there was liver failure. Rechallenge strategies have not been studied in randomized trials in HIV-infected patients. However, in the largest randomized trial of TB without HIV coinfection, approximately 90% of patients were rechallenged with their first-line TB drugs without recurrence [112]. Risk of recurrence was not related to whether the four first-line TB drugs were reintroduced sequentially or concurrently. Further studies are needed to define the optimum rechallenge strategy in coinfected patients in whom both TB treatment and ART require reintroduction. Until further evidence emerges, the American Thoracic Society recommends that RIF can be reintroduced in coinfected patients once the ALT is less than two times the upper limit of normal followed by reintroduction of INH with monitoring of liver function [100]. However, they also suggest that pyrazinamide is not reintroduced.

While some cohort studies have suggested low morbidity and mortality in HIV-infected patients with DILI [109], mortality is substantial among those requiring hospital admission. In a South African study, mortality was 35% among patients admitted to hospital with DILI during TB treatment, ART or concurrent therapy [113]. Reasons for these deaths were sepsis and liver failure, although interruption of required TB treatment and ART are likely to have played a role.

TB treatment is associated with a spectrum of cutaneous adverse reactions including morbiliform rashes, Steven Johnson syndrome and toxic epidermal necrolysis, fixed drug eruption, lichenoid drug eruptions and acute generalized exanthematous pustulosis [114]. Cotrimoxazole, NVP, and to a lesser extent EFV, can also cause many of the same clinical presentations [102, 115, 116]. HIV coinfection was associated with a fivefold increased risk of rash or drug fever in one study [117] but small, non-significant increases in risk in others [105, 108]. If a clinically significant rash develops, all potentially responsible drugs need to be interrupted and then a carefully monitored rechallenge of first-line TB drugs can be considered once the rash has resolved. In a cohort of mainly HIV-infected patients rechallenged following cutaneous reactions to TB drugs, 50% developed reintroduction reactions but only a small minority were severe [118].

Renal dysfunction may be caused via different mechanisms in patients receiving tenofovir, RIF or aminoglycosides (used for MDR-TB). Tenofovir and aminoglycosides may both cause tubular cell toxicity at the level of the proximal renal tubules, whereas RIF infrequently causes a tubulointerstitial nephritis mediated by immune hypersensitivity. Case reports describe renal failure in patients receiving a combination of tenofovir and aminoglycosides, although cohort studies have not confirmed an increased risk [119]. The combination is best avoided when possible. In patients with significant renal dysfunction, Use of tenofovir should be avoided where possible and dosing of ethambutol, NRTI drugs, some quinolones (ofloxacin and levofloxacin) and certain other second-line antituberculosis drugs (including cycloserine, para-aminosalicylic acid, clofazamine and linezolid) needs to be adjusted.

TB Immune reconstitution inflammatory syndrome (IRIS)

Two major forms of TB immune reconstitution syndrome (TB-IRIS) are recognized and these are called paradoxical TB IRIS and unmasking TB-IRIS and case definitions have been published [120]. Paradoxical TB-IRIS is an important cause of morbidity in patients known to have HIV-associated TB and occurs within the first weeks of ART [120, 121]. The typical clinical course of paradoxical TB-IRIS is as follows. Initiation of TB treatment in a patient with HIV infection and newly diagnosed TB results in clinical stabilization or improvement. However, subsequent introduction of ART is accompanied by recurrence or exacerbation of TB symptoms with new or worsening clinical signs of TB that often have a marked inflammatory component [120, 121].

While seldom life-threatening, deaths due to paradoxical TB-IRIS have been described. Two major risk factors identified in observational studies [122–125] and in clinical trials [55, 95, 126] are a low CD4 count prior to ART and a shorter interval between starting TB treatment and ART. There is no diagnostic test for TB-IRIS; the diagnosis is based on clinical presentation and exclusion of alternative diagnoses such as bacterial infection or drug resistant TB [120]. However, drug-resistant TB is not only in the differential diagnosis as an alternative cause of the clinical deterioration but may also be a risk factor for the development of paradoxical TB-IRIS [127].

The second major form of TB-IRIS is commonly referred to as ‘unmasking’ TB-IRIS. This occurs when active TB is present but remains undiagnosed at the time of starting ART [120, 128]. Subsequent immune recovery triggers the overt symptomatic presentation of TB. In a proportion of cases, unusual inflammatory features may also develop and such cases are regarded as having ‘unmasking’ TB-IRIS. Risk of unmasking TB-IRIS is therefore directly related to the efficiency of the pre-ART screening process and the resulting prevalence of undiagnosed disease.

Both types of TB IRIS have a wide range of clinical features often with involvement of multiple organ systems, reflecting widespread dissemination of M. tuberculosis in those with profound immunosuppression. Common features include fever, recurrence of respiratory symptoms with worsening infiltrates on chest radiographs, enlargement of lymph nodes (often with suppuration), formation of tuberculous abscesses and serous effusions [120, 121]. There are many case reports of unusual and diverse complications, including granulomatous nephritis with renal impairment, parotitis, epididymo-orchitis, granulomatous hepatitis, splenic enlargement and abscess formation, psoas abscess, peritonitis, ascites and intestinal involvement [120, 121]. Neurological TB-IRIS is particularly severe, manifesting with tuberculomas, tuberculous abscesses, cerebral edema, meningitis and radiculomyelopathy [99, 129, 130]. Neurological TB IRIS has a much poorer outcome compared to other forms, with a mortality of 13% to 75% [99, 129, 130].

In most cases, the onset of paradoxical TB-IRIS is within the first 4 weeks of ART (median 14 days (IQR, 8 to 23) in 1 series [127]) but can occur within a few days. The proportion of patients affected ranges widely from 0% to over 40% [120] and this may relate to differences in risk factors and case definitions. In a meta-analysis, the summary risk estimate was 15.7% [131]. Of these, 3.2% died, representing approximately 1 in 200 patients with HIV-associated TB who start ART. The median duration of TB-IRIS symptoms has been reported to be 2 to 3 months [124, 125] but a minority of cases have a protracted course which may last for more than 1 year [120, 124, 132]. Such protracted cases typically have persistent or recurrent suppurative lymphadenitis or abscess formation. However, the majority of cases have a favorable long-term outcome [133].

TB-IRIS is not an indication for stopping ART, although this should be considered in life-threatening cases such as those with cerebral edema and depressed level of consciousness or severe respiratory failure. In mild cases, no specific treatment is usually required; the patient should be treated symptomatically and counseled regarding the need to continue ART and TB treatment. Corticosteroids should be considered if symptoms are more significant. In a randomized placebo-controlled trial, prednisone used at a dose of 1.5 mg/kg/day for 2 weeks followed by 0.75 mg/kg/day for 2 weeks was associated with reduced morbidity (duration of hospitalization and need for therapeutic procedures) [134]. Symptom improvement was more rapid and there was no excess risk of other severe infections [134]. Although no mortality benefit was demonstrated, patients with immediately life-threatening TB-IRIS were not enrolled in view of ethical considerations. Indeed, most experts recommend steroid therapy for life-threatening TB-IRIS, especially IRIS involving the CNS. A subgroup of patients in this trial (approximately one in five) relapsed after stopping prednisone and required a further and more prolonged course to control symptoms [134]. Similarly, in other settings, TB-IRIS has relapsed in up to 50% of patients after stopping steroids [133] and thus the duration of therapy must be tailored according to the clinical response.

Non-steroidal anti-inflammatory drugs (NSAIDs) have also been used in the treatment of TB IRIS although no clinical trial data exist to support their use. Other forms of immunomodulatory therapy such as thalidomide, azathioprine and tumor necrosis factor α blockers (such as adalumimab) have been used in cases refractory to steroid therapy with anecdotal reports of benefit [135]. In patients with suppurative lymphadenitis or abscesses, needle aspiration may provide a pus sample to exclude drug-resistant TB as well as bringing symptomatic relief.

There is no evidence base for pharmacological prevention of TB-IRIS. However, this needs to be considered in view of the recommendation within guidelines for early ART initiation in TB patients with advanced HIV [11]. Adjunctive immunomodulatory therapies might reduce the risk or severity of TB-IRIS in such patients. A randomized placebo-controlled trial of prednisone for prevention of TB-IRIS in high-risk patients (CD4 counts <100 cells/mm³ starting ART within 30 days of TB treatment) is underway [136]. Until results from this trial are available corticosteroids cannot be recommended for prevention of TB IRIS with the exception of patients with TB of the CNS for whom adjunctive steroids form part of the standard of care [137]. However, in such patients, TB IRIS occurs in approximately 50% of patients with CNS TB starting ART despite receipt of corticosteroids [99].

Other agents that have been proposed for prevention of TB IRIS are vitamin D, statins and the C-C chemokine receptor type 5 (CCR5) blocker maraviroc [135]. Vitamin D has modulating effects on both the adaptive and innate immune responses [138, 139]. Statins have anti-inflammatory properties and there is precedence for using these agents for autoimmune inflammatory disorders in an experimental model [140, 141]. However, neither vitamin D nor statins have yet been tested in clinical studies. Maraviroc, however, was shown not to prevent IRIS in a placebo-controlled trial conducted in Mexico and South Africa [142].