Optimum timing of antiretroviral therapy in HIV infected patients with concurrent serious opportunistic infections

The optimal time to initiate antiretroviral therapy (ART) in HIV-1 infected patients with serious opportunistic infections has, until recently, been undefined. Initiation of early ART is associated with the risk of the immune reconstitution inflammatory syndrome (IRIS), complex drug interactions and a high pill burden; but deferral risks advancing immunosuppression and mortality. A number of recent clinical trials have now better informed this issue. Two studies have shown benefit from earlier ART initiation [1, 2]. A study of patients with a range of opportunistic and bacterial infections, apart from tuberculosis (TB), demonstrated that early ART initiation (median 12 days after diagnosis of infection) resulted in fewer AIDS progression events or deaths (a combined secondary endpoint) compared with patients who initiated after acute infection treatment (median 45 days) [1]. Subanalysis showed that in patients with fungal infections there was also significant reduction in AIDS progression/death with early initiation. In TB, patients starting ART during TB treatment had reduced mortality compared with those deferring until TB treatment was completed [2]. The pendulum seemed to be swinging in favor of earlier ART initiation in the context of opportunistic infections. However, in a study of patients with TB meningitis there was no difference in mortality between patients who started ART at the same time as TB treatment and those who deferred 2 months, but there were significantly more severe adverse events in the first 2 months in the immediate arm [3].

A study by Makadzange and colleagues published in this edition [4] evokes further concern. 54 ART-naïve patients diagnosed with cryptococcal meningitis (CM) were randomized in an open-label clinical trial to receive early ART (within 72 hours of CM diagnosis) or to defer ART initiation until after 10 weeks of fluconazole treatment. The data and safety monitoring board stopped the study because of significant excess mortality in the early ART arm (88 vs 54% 3-year mortality). Thus while ART appears to improve survival in patients with CM [5, 6] very early initiation under some circumstances may be harmful.

Three factors need to be considered when interpreting these results that may limit their generalizability. Firstly, in the early treatment arm, patients were commenced on ART just 72 hours after the diagnosis of CM: very early ART initiation. Secondly, IRIS in the CNS has been shown in several settings to be the form most associated with mortality [7, 8]. Thirdly, the treatment of CM was with fluconazole 800mg daily without initial amphotericin B. Amphotericin B has greater early fungicidal activity resulting in more rapid clearance of viable cryptococci [9].

Almost all deaths occurred within the first four weeks of the study, reflecting mortality during early ART in the early arm and prior to ART in the delayed arm. In both groups the majority of deaths were attributed to CM. The authors hypothesise some of the excess mortality in the early ART arm may have been caused by cryptococcal IRIS. Raised intracranial pressure associated with cryptococcal IRIS may have played a role. However, it is difficult to be certain of the contribution of IRIS to the excess mortality for several reasons. Firstly, acute inpatient management remained with the primary medicine service and the study team had no input into management including recurrent recording and reduction of raised intracranial pressure: the first study visit after enrollment was at 2 weeks and approximately half of the deaths had occurred before this. Secondly, there appeared to be no attempt to ascertain cryptococcal IRIS cases using a case definition as other investigators have done [10-12], repeat lumbar punctures at deterioration were not routinely performed to measure opening pressure or to determine whether cell count was increasing (suggestive of IRIS), and there appeared no management plan for IRIS.

Typically cryptococcal IRIS manifests with recurrent or new symptoms and neurological manifestations of meningitis during the first weeks of ART in patients who have been diagnosed with CM prior to ART. If the interval between starting antifungal treatment and ART is very short then cryptococcal IRIS may manifest as progressive deterioration given that there has been no interval for improvement. The incidence of cryptococcal IRIS among patients with CM starting ART reported from sub-Saharan Africa is 17-33% [5, 10, 13]. Raised intracranial pressure is common in cryptococcal IRIS cases: median opening pressure at lumbar puncture was 29 cmH₂O (IQR 19-31) in one study [10]. Mortality of up to 66% has been reported [13].

The optimal management of cryptococcal IRIS has not been studied in a clinical trial. Evidence that adjunctive corticosteroid therapy reduces hospitalization and outpatient therapeutic procedures and improves symptoms exists for paradoxical TB-IRIS [14]. There are anecdotal reports of response to corticosteroids in patients with cryptococcal IRIS [10, 12, 15]. Therapeutic lumbar punctures to control raised intracranial pressure have been advocated as an important part of the management of CM [16, 17].

Studies have demonstrated that a shorter interval from starting antifungal treatment to initiation of ART is associated with cryptococcal IRIS [11, 12]. This suggests that persistent viable cryptococci in the CNS may be a risk factor for cryptococcal IRIS. Given that the CSF clearance of organism with fluconazole is slower than with amphotericin B [9], it is plausible that initial treatment with fluconazole may have predisposed patients to higher rates of IRIS than seen in cohorts where initial amphotericin B treatment is used.

Against these potential deficits in trial design it must be acknowledged this study was conducted under busy clinical service conditions in Harare, Zimbabwe, at a time of political and socio-economic turmoil in that country. The investigators are commended for attempting any important clinical trial under such difficult circumstances. The antifungal treatment of fluconazole reflects standard practice throughout much of sub-Saharan Africa. Although amphotericin B and flucytosine combination is advocated as the optimal initial treatment for CM [18], amphotericin B is expensive and difficult to administer and monitor in many under-resourced settings and flucytosine is unavailable. This may partly account for the very poor outcomes in CM in sub-Saharan Africa despite ART availability. It is therefore important that efforts to advocate and research ways to provide better initial CM treatments in these settings continue. Making amphotericin B and flucytosine available, and research regarding higher doses of fluconazole [19] are critical. Recent data suggest a combination of fluconazole and flucytosine may provide better CSF clearance than fluconazole alone [20] and this combination may provide a better oral option where amphotericin B treatment is not possible. Broadening ART access is also critical in improving outcomes of HIV-associated CM, but this study [4] cautions that very early ART in the context of fluconazole therapy alone may be associated with harm.

Further specific studies on this subject appear necessary. Issues that need to be addressed are what the impact of starting ART slightly later at 2 weeks would be, and whether amphotericin B as an initial treatment would prevent the excess mortality associated with early ART. Such studies also need to have in place procedures for early intensive monitoring and active ascertainment of cases of cryptococcal IRIS, and when IRIS occurs active management of the inflammatory component and the associated raised intracranial pressure. Clinical trials to define the optimal management of cryptococcal IRIS are also required.