|Home | About | Journals | Submit | Contact Us | Français|
Cryptococcal immune reconstitution inflammatory syndrome (C-IRIS) may present as a clinical deterioration or new presentation of cryptococcal disease following initiation of antiretroviral therapy (ART) and is believed to be caused by recovery of cryptococcus-specific immune responses. We have reviewed the existing literature on C-IRIS to inform the development of a consensus case definition specific for paradoxical cryptococcal IRIS in patients with known cryptococcal disease prior to ART, and a second definition for incident cases of cryptococcosis developing during ART (here termed ART-associated cryptococcosis), a proportion of which are likely to be “unmasking” C-IRIS. These structured case definitions are intended for use in future clinical, epidemiologic and immunopathologic studies of C-IRIS, harmonizing diagnostic criteria, and facilitating comparisons between studies. As with tuberculosis-associated IRIS, these proposed definitions should be regarded as preliminary until further insights into the immunopathology of IRIS permit their refinement.
Cryptococcal disease remains a major cause of morbidity and mortality in persons with advanced HIV/AIDS, particularly in Africa and South-East Asia (1–4). Cryptococcal immune reconstitution inflammatory syndrome (C-IRIS) presents as a clinical deterioration of cryptococcal disease following rapid reversal of immune deficiency (5). This is thought to be mediated by recovery of Cryptococcus-specific immune responses, resulting in exaggerated inflammatory host responses. In HIV, this reversal is driven by antiretroviral therapy (ART), but C-IRIS can also occur after solid organ transplantation, with an estimated incidence of 4.8%, or in pregnancy (6, 7).
The International Network for the Study of HIV-associated IRIS (INSHI) was established in 2006 to promote research collaboration and harmonization among IRIS researchers worldwide. Generic IRIS definitions are limited in their application because of the highly heterogeneous spectrum of underlying disease and clinical features of IRIS. We have undertaken a review of existing literature on the clinical and diagnostic features of C-IRIS, and used this to develop a consensus case definition for C-IRIS that can be used in both resource-limited and well-resourced settings.
Similar to the case definition of tuberculosis (TB)-associated IRIS (8), two distinct modes of presentation of C-IRIS are recognized. First, in up to one third of patients with cryptococcosis diagnosed pre-ART, “paradoxical” C-IRIS occurs following ART initiation. This presents as worsening or recurrence of treated cryptococcal disease in the same or new anatomical sites despite microbiologic treatment success (9–12). Second, new onset of cryptococcosis occurs in up to 1% of patients initiating ART in whom cryptococcosis was not recognized prior to ART (13–17).
Incident cryptococcosis during ART may result from the emergence of previously sub-clinical disease during immune reconstitution (“unmasking C-IRIS”), or simply be associated with persistent immunodeficiency. These two entities may be indistinguishable at present, therefore we favor the term “ART-associated cryptococcosis” for both unmasking C-IRIS and immunodeficiency-related cryptococcosis occurring on ART.
We conducted a Medline search during the period up to June 2010 to identify published cohort studies, case-control studies, case series, and case reports in English describing or enumerating C-IRIS cases. We searched bibliographies of articles focusing on IRIS for other relevant reports. Studies and reports were categorized by whether they used a case definition for IRIS (Table 1) or simply reported clinical and laboratory features (Table 2). Unpublished data and conference abstracts were not included. Where possible, the original source data was obtained from authors of cohorts.
There were 11 published cohort studies focusing on C-IRIS, comprising 142 cases from Brazil, France, Italy, South Africa, Thailand, Uganda, and the United States (U.S.) (9–12, 14, 18–24). A further 5 cohort or case-control studies that also reported other IRIS-associated conditions described an additional 23 cases of C-IRIS or ART-associated cryptococcosis (15, 16, 25–27), and case reports and series reported 48 cases (28–54). Overall, 183 of the 213 reported cases (85.9%) were paradoxical C-IRIS, while the remainder were either ART-associated cryptococcosis (n=25) or in individuals where the pre-ART history was unknown (n=3).
Almost all structured case definitions used to diagnose C-IRIS (9–12, 15, 21, 26, 52) focused on paradoxical presentations, and share three common criteria: (1) new or worsening clinical disease after ART initiation, (2) evidence of ART effectiveness, and (3) exclusion of alternative etiologies. The clinical and laboratory case definition criteria for C-IRIS were inconsistent between studies and were frequently subjective or imprecise (e.g. “evidence of immunological and/or virological response”) (12, 21). Six of thirteen studies that used an IRIS definition required fungal cultures to be negative in paradoxical C-IRIS (9–12, 18, 19).
In seven studies including a total of 598 patients with known cryptococcal disease initiating ART, the reported incidence of C-IRIS was 8–49% (9–12, 14, 19–21, 23, 24, 27). Of 171 cases in which both the site and mode of presentation were published, 126 (73.7%) involved mainly meningeal disease (Table 1 and and2).2). Other common presentations were complications of central nervous system (CNS) disease (19 cases; 11%), lymphadenopathy (19 cases; 11%), pneumonitis (9 cases; 4.5%), multifocal disease (7 cases; 4.1%) and soft tissue disease (2 cases; 1.8%). Specific CNS features included intracranial cryptococcoma or abscess (20, 22, 35, 41), spinal cord abscess (38), recalcitrant raised intracranial pressure (22, 34, 43), optic disk swelling (50), cranial nerve lesions (21, 45), dysarthria (26), hemiparesis (21) and paraparesis (14, 53). Non-CNS manifestations of paradoxical C-IRIS include fever (28, 31, 48), eye disease (9), suppurating soft tissue lesions (28, 33, 36), hypercalcemia (36) and pulmonary disease, including cavitating or nodular lesions (9, 11, 22, 26, 36, 52).
The time of onset of paradoxical C-IRIS after ART varies widely, from 4 days up (31) to around 3 years (12, 38), with a median time ranging from 1 to 10 months (9–12, 14, 18–21) in cohort studies. Reported CD4 counts prior to ART were typically below 50 cells/mm3 (summarized in Table 1) but pre-ART viral loads (VL), virologic outcomes and follow-up CD4 counts were inconsistently reported. In case reports, individual data were usually reported, giving a median baseline CD4 count of 28 cells/mm3 and VL of 5.6 log10 copies/mL, and CD4 count at event of 162 cells/mm3 and VL of 2.4 log10 copies/mL.
One study reported comparable data for controls (21), although follow-up data were recorded at 6 months and not at the time of symptoms. In this South African study of 11 cases and 54 controls, the median baseline CD4 count in 6 cases was 28 cells/mm3, rising to 162 cells/mm3 after 6 months of ART, compared to 41 cells/mm3 at baseline in controls, rising to 187 cells/mm3 after 6 months (21). Baseline VL in cases was median 110,000 copies/mL, with a minimum fall of 1.6 log10 at the time of event, and all cases had VL <1000 copies/mL at 6 months. In controls, baseline VL was 150,000 copies/mL and at least 75% achieved VL <50 copies/mL at 6 months.
Mortality from paradoxical C-IRIS and ART-associated cryptococcosis ranges from 27–83% in Africa (9, 14, 15, 21, 22) and 0–20% in North America, Europe and South-East Asia (10, 12, 19, 20, 23). In sub-Saharan Africa, paradoxical C-IRIS is considered to be an important contributor to early mortality following ART (9, 14, 18, 22).
Risk factors for paradoxical C-IRIS have been reported in six studies (10–12, 23, 55), and retrospective cohorts indicate these may include higher HIV-1 viral load prior to ART (in a study including both ART-experienced and naïve patients) (10), earlier initiation of ART (10, 12) and greater CD4 count increase in the first 6 months of ART (21). However, in three prospective cohorts, HIV-1 viral load, time to start ART, and baseline CD4 were not risk factors for C-IRIS (21–23). Markers of fungal burden may also be risk factors, i.e.; the presence of fungemia; and higher serum cryptococcal antigen (CrAg) titer at pre-ART diagnosis (23). However, two studies examining markers of fungal burden did not find an association between number of colony-forming units in the CSF and IRIS (21, 22). In another prospective cohort, lack of initial CSF inflammation (CSF protein <50 mg/dL and WBC <25 cells/µL) prior to ART was associated with a seven-fold increase in IRIS risk (22).
Infection with Cryptococcus neoformans can remain latent for years after initial exposure (56), and active disease may remain subclinical for some time in patients with advanced immunodeficiency, so it is not surprising that clinical cryptococcosis may emerge for the first time after ART initiation. The incidence of ART-associated cryptococcosis ranged from 0.2% to 1.6% in 6 studies including a total of over 2000 patients without evidence of cryptococcosis before initiating ART (13–17, 27). However, the incidence may be as high as 33% in individuals with subclinical cryptococcal antigenemia without fluconazole preemptive therapy (57, 58). Subclinical antigenemia is therefore the overwhelming risk factor for ART-associated cryptococcosis (17, 57).
Of 25 cases of ART-associated cryptococcosis in which clinical features were described, the clinical spectrum and relative frequency of different manifestations were similar to that reported in paradoxical C-IRIS. Meningitis and/or CNS complications occurred in 17 cases (68.0%), skin or soft tissue lesions in 3 cases (12.0%), lymphadenopathy in 2 (8.0%), lung disease in 2 (8.0%), and disseminated disease in 1 (4.0%). In 13 cases where individual values were reported, the baseline median CD4 count was 19 cells/mm3 and VL was 5.5 log10 copies/mL, and follow-up median CD4 count was 65 cells/mm3 and VL was 2.6 log10 copies/mL.
One feature of ART-associated cryptococcal meningitis is the accelerated onset of severe illness over a few days (14, 17, 31, 53) from the onset of symptoms, compared with the 1 to 2 week subacute course typically seen with cryptococcal meningitis (CM) in patients not receiving ART (9, 59). We compiled all individually reported times of symptom onset: the median time on ART was 9 weeks (n=54, inter-quartile range (IQR) 2–26 weeks) in paradoxical C-IRIS cases and 4 weeks (n=19, IQR 2–10 weeks) in ART-associated cryptococcosis cases, although this was not statistically different (P=0.12, Wilcoxon rank-sum test).
In patients who develop new clinical cryptococcosis during ART, it may be very difficult to differentiate between IRIS-associated disease (caused by restoration of specific immune responses) and progression of untreated, occult cryptococcosis in the context of persisting immunodeficiency (60). The analogous situation in ART-associated tuberculosis has led to the use of the term “unmasking TB-IRIS” being reserved for a subset of cases with “heightened intensity of clinical manifestations” or “rapid, destructive necrotic inflammation” (8, 61–65). It may be that unusual ART-associated cryptococcosis cases with florid inflammatory features are more likely to occur with unmasking IRIS.
However, the clinical spectrum of ART-associated cryptococcosis does not appear to differ significantly from that seen in the pre-ART era, when 75–90% of cryptococcosis presented as meningitis and/or encephalitis, and the remainder were mainly pneumonitis, lymphadenopathy, or cutaneous lesions (66). Therefore, while we acknowledge that most post-ART cases of cryptococcosis may be suspected unmasking C-IRIS, we propose that such cases should be reported as ART-associated cryptococcosis until further evidence becomes available to support meaningful clinical discrimination between C-IRIS and immunodeficiency disease.
We have developed case definitions for paradoxical C-IRIS and ART-associated cryptococcosis based on published data on the clinical and diagnostic features. The case definition for paradoxical C-IRIS (Table 3) applies to patients with recognized pre-ART cryptococcal disease who subsequently deteriorate while on ART. The definition of ART-associated cryptococcosis (Table 4) applies to those without recognized cryptococcosis at the time of initiating ART, who develop cryptococcosis on ART. While these definitions are focused on HIV-related IRIS, we anticipate their modification for use in other, non-HIV situations where there is reversal of immunosuppression (5–7).
From our literature review, we identified five main clinical syndromes in paradoxical C-IRIS and ART-associated cryptococcosis: (1) meningitis, accounting for around 70% of cases; (2) space-occupying CNS lesions; (3) lymphadenopathy, typically necrotic; (4) pneumonitis; (5) soft tissue or subcutaneous mass lesions, typically suppurative. Other localized sites such as bone, prostate and peritoneum have been described in non-HIV-infected individuals and may be possible sites of C-IRIS disease, but have not been reported in ART-treated individuals.
There may be vigorous inflammatory signs with granulomatous lesions and/or evidence of necrosis, with or without organisms on fungal staining and culture (28, 32, 33, 38, 39, 42, 46). Unpublished data suggest that elevated C-reactive protein (CRP) and IL-6 concentrations precede the development of C-IRIS, compared with ART-treated individuals with cryptococcosis who do not experience IRIS (67). At time of C-IRIS, a marked type-1 CD4 T-helper (Th1) response is present in serum and CSF (22, 55). Prior to ART, persons with increased Th2 responses (e.g. IL-4) may be at increased risk of subsequent C-IRIS (unpublished data). The emerging paradigm of IRIS pathogenesis suggests that persons at risk for C-IRIS have a paucity of inflammation and ineffective antigen clearance prior to ART (22), followed by antigen presentation on ART and a robust and probably dysregulated antigen-specific and generalized pro-inflammatory response (55, 60, 68–71). A central pathogenic role for an alteration in Th1/Th2 balance has also been proposed (72, 73).
The CSF profiles in paradoxical C-IRIS cases frequently show an increased CSF white blood cell count (WBC) and opening pressure (OP) ≥25 cm of CSF, but the range of values overlaps considerably with those observed in patients with CM prior to ART or non-IRIS relapses of CM due to therapeutic failure (21, 22). In one series of 14 paradoxical C-IRIS cases and 45 controls with non-IRIS CM, higher CSF opening pressures (OP) (median 45 vs. 31 cm CSF) and CSF WBC (56 vs. 12 cells/µL) were reported in IRIS than in non-IRIS cryptococcal meningitis (10). However, the interpretation of these findings is limited because the control group for comparing WBC counts included pre-ART as well as post-ART cases of cryptococcal meningitis, and high CSF OP and WBC formed part of the diagnostic criteria for IRIS. In addition, these findings were not confirmed by three other studies of similar or larger size (18, 21, 22). Therefore, at present it is not possible to recommend a pre-defined threshold in CSF OP or WBC to reliably distinguish IRIS from non-IRIS CM, and further studies are needed in this area. Cytokine profiles at time of clinical deterioration may distinguish C-IRIS from CM-relapse with increased pro-inflammatory cytokines (e.g. IFN-γ, TNF-α, IL-12, IL-17) present in the CSF of C-IRIS (22).
We propose an upper time limit of 12 months after ART initiation is based on the reported time of onset in published studies. It appears that C-IRIS has a relatively delayed onset compared to other diseases such as TB, where the median onset of paradoxical TB-IRIS is as early as 2–4 weeks of ART (74, 75). Rare, late presentations of cryptococcal disease can occur (12, 19, 38, 42, 45), in which cases it becomes essential to exclude ART failure and relapse with fluconazole-resistant organisms.
Confirmation of a virologic response to ART is recommended, but not essential, for the diagnosis of C-IRIS. A specific VL threshold (reduction in VL of <1 log10 (15, 25, 52, 62) at the time of clinical event) has been suggested in the literature as an exclusion criterion for IRIS. However, there has never been any experimental validation of the best time to perform VL assays, or the best threshold value. In contrast, studies designed to validate the INSHI TB-IRIS clinical case definition (8) have shown that the absence of CD4 and VL criteria does not significantly impact on the case definition’s performance (76, 77). Several other major studies examining the epidemiology of IRIS have also omitted CD4 or VL criteria in their case definition (15, 25, 26, 78–81).
While the lack of a virologic response to ART may support an alternate diagnosis of treatment failure, suboptimal responses to combination ART may still be consistent with IRIS (the syndrome was originally described during zidovudine monotherapy (82)). ART-naïve patients with good adherence typically have excellent virologic responses to ART in the first 6 months of therapy (83). In the largest prospective cohort to date, only 5 out of 144 (3.5%) probable IRIS events had virologic failure (27). Therefore one may reasonably assume that ART-naive individuals who are reasonably adherent will have had an early virologic response and partial immune reconstitution within days or weeks of initiating ART, in the absence of another reason, and the lack of available virologic data does not exclude IRIS. In settings where viral load is readily available, a threshold of 1 log10 reduction in VL at the time of clinical deterioration may be employed. In all settings, an assessment of adherence to ART is essential in patients presenting with suspected C-IRIS.
Our C-IRIS case definitions do not include any criteria based on the patient’s CD4 count. As with VL changes, some authors have incorporated a rise in CD4 count as a diagnostic criterion for IRIS (15, 52, 62), but several major studies have omitted this measure (15, 25, 26, 78–81). The main arguments for this omission have been presented above as well as in a published review (8). The number of CD4+ T cells measured in peripheral blood does not necessarily reflect function nor how many CD4 cells are actually present at the site of an opportunistic infection. In addition, there may be functional immunological improvements even in the absence of a CD4 count rise. In one prospective C-IRIS cohort study, 30% (11 of 36) persons with C-IRIS and with sterile CSF cultures and virologic responses had <25 cell/µL CD4 increase from their pre-ART baseline (22). Furthermore, data from cohort studies on IRIS associated with non-tuberculous mycobacterial infection suggest that restoration of a pathogen-specific cellular immune response may occur without an increase in the circulating CD4+ T-cell count (84).
With regard to exclusion of other causes of disease, we have emphasized the importance of adherence to ART and antifungal therapy, successful response to ART, and exclusion of co-morbidity such as other common pathogens or malignancy. However, in some settings this may be problematic because of limited diagnostic capabilities. As a minimum requirement, TB (using staining for acid fast bacilli) and bacterial disease should be excluded. Indeed IRIS associated with dual co-infection is possible, and ART-associated pulmonary TB has been reported in patients with pre-ART CM (9).
Discriminating C-IRIS from drug toxicity is not usually problematic as there are few drug effects likely to mimic the syndrome. Two rare but plausible considerations are hepatic cryptococcal lesions mimicking drug-related hepatotoxicity (85), and aseptic meningitis induced by trimethoprim/sulfamethoxazole (86).
Paradoxical C-IRIS can occur after initiation ART in treatment-naïve individuals, re-initiation of ART, or switching regimen after virologic failure. To meet the antecedent criteria of paradoxical C-IRIS, the pre-ART diagnosis of cryptococcosis may be defined by locally-available diagnostic facilities, e.g. India ink stain, fungal culture, and CSF or serum cryptococcal antigen detection. Two scenarios may fall outside the definition of paradoxical C-IRIS: patients with known cryptococcosis who have discontinued antifungal therapy prior to initiating ART, and cases where a provisional pre-ART diagnosis of cryptococcosis is confirmed after ART initiation.
In an ideal setting, quantitative CSF culture should be used to assist the diagnosis of C-IRIS presenting with meningitis. Yet, in contrast with several published definitions (9–12, 18, 19), our view is that a negative cryptococcal culture is not an absolute requirement for a diagnosis of paradoxical C-IRIS. Our reasoning is that the timing of CSF culture sterility is variable, with approximately 50% of patients becoming culture negative within 2 weeks, 66% by 4 weeks, and 80% by 6 weeks of amphotericin-based therapy (9, 87–91). The reality in sub-Saharan Africa is that many cryptococcus-infected patients only receive oral fluconazole (18, 92–95) therapy, and CSF sterilization is delayed compared with amphotericin B (89). Even in patients receiving high-dose fluconazole therapy (800–1200mg daily), up to 20% may have persistent positive CSF cultures after 12 weeks (96, 97). Thus if patients commence ART shortly after antifungal therapy, they may still be CSF culture positive when they present with paradoxical IRIS. In the largest prospective cohort to date, 32 of 33 subjects with meningitis C-IRIS had negative CSF cultures (22). The single person with culture-positive C-IRIS had a quantitative CSF culture with a very low 70 colony forming units(CFU)/ mL (22).
While accepting the difficulties in diagnosing C-IRIS when cultures remain positive, we favor an approach informed by the expected range of therapeutic responses to antifungal therapy. The management implications of this are discussed below. Although the likelihood of a positive fungal culture and the risk of C-IRIS both decrease over time, we believe a positive fungal culture after 3 months of antifungal therapy is likely to indicate failure of antifungal therapy and therefore propose excluding such events from the paradoxical C-IRIS definition.
Serum and CSF CrAg titers are unhelpful in the diagnosis of C-IRIS. In the above prospective Ugandan cohort (n=101), 25% of patients had <4-fold decrease in CSF CrAg titer at time of their C-IRIS event (22).
Recurrence of cryptococcal symptoms may be caused by the host immune system (i.e. IRIS), ongoing immunosuppression, sequelae of existing disease, or failure of antifungal therapy. Conceptually it is easier to consider these mechanisms in distinction from one another, but more than one may exist concurrently. For example, suboptimal antifungal therapy or suboptimal adherence may attenuate the decline in pathogen burden prior to ART and provoke a more vigorous cryptococcus-specific immune response on ART than would have occurred with more effective antifungal treatment. The development of paradoxical TB-IRIS in cases of undiagnosed drug-resistant TB has recently been highlighted (98). Furthermore, whether the antigen is derived from live organisms, dead intact organism, or cellular debris, may not be critical in propagating an immune response leading to IRIS.
As already discussed, there is inherent uncertainty in differentiating ART-associated cryptococcosis caused by restoration of a cryptococcus-specific immune response (i.e. unmasking C-IRIS) from cryptococcosis caused by persisting immunodeficiency on ART. Clinical management may be influenced by this distinction, and both causes are important in regions of high cryptococcal prevalence.
Given the relative lack of evidence, our criteria to distinguish between immune deficiency-related cryptococcosis and unmasking IRIS are provisional at this stage. While specific criteria or clinical cut-offs may improve the objectivity of our definition, they introduce unnecessary arbitrariness into the diagnostic process. As with tuberculosis (61) and other OIs (60), there may be a continuous spectrum between immunodeficiency-associated cryptococcosis and C-IRIS, rather than two distinct entities.
Although clinically recognizable cryptococcosis should be absent at ART initiation, the presence of mild symptoms prior to ART should not preclude the diagnosis if appropriate screening for cryptococcal infection was initially negative. Occult cryptococcal antigenemia can occur in asymptomatic individuals prior to ART, particularly in resource-limited settings with high cryptococcal prevalence (57, 99). Although patients with untreated antigenemia are more likely to develop clinical cryptococcosis (57, 99–101), we have observed ART-associated cryptococcosis in persons who were serum CrAg-negative prior to ART in South Africa and Uganda, albeit rarely (unpublished data). Similarly, a placebo-controlled study of fluconazole primary prophylaxis in Uganda reported ART-associated cryptococcosis in 1.0% of individuals receiving placebo and ART, all of whom were serum CrAg-negative at a median of 11 weeks pre-ART (13). Therefore, neither a positive nor negative serum CrAg prior to the start of ART is an exclusion criterion for ART-associated cryptococcosis. However, pre-ART screening for cryptococcal antigenemia may be a useful strategy for identifying and treating subclinical infection and reducing the incidence of ART-associated cryptococcosis in high prevalence regions.
Prevention of paradoxical C-IRIS has been a justification for delaying ART initiation, yet there is unclear evidence supporting such rationale. In two retrospective studies, earlier ART initiation (<4–8 weeks) was associated with increased risk of C-IRIS (10, 12), but in two prospective observational cohorts, timing of ART initiation was not associated with C-IRIS (21, 22). Two randomized controlled trials have been completed to address the question of when to start ART, using mortality as their primary outcome. In the ACTG a5164 trial, delaying ART to 6 weeks after starting OI therapy did not decrease the incidence of IRIS compared to starting ART within 2 weeks, but it did increase mortality (102). Only 35 (12%) subjects in the trial had cryptococcosis, compared to 63% with pneumocystis, and only aggregated data were reported (102). In a second trial of 54 subjects from Zimbabwe, initiation of ART at a median of 24 hours after starting antifungal therapy was associated with increased mortality compared to a delay of 10 weeks (103). The causes of death were unclear and C-IRIS was not assessed (103). We do not support either approach to the timing of ART in this study as recommended clinical practice. A clinical trial is currently planned to definitively answer when to start ART after CM, and whether earlier ART is associated with excess C-IRIS risk (NCT01075152).
For treatment of cryptococcosis, published expert guidelines do not consider IRIS in detail (104, 105). Intensification of antifungal therapy is indicated in all severe C-IRIS cases (e.g. intracranial space-occupying lesions or extracranial disease impinging on vital structures), any with positive culture, and patients suspected to have received a suboptimal antifungal regimen.
Case reports have noted beneficial responses of C-IRIS to immune modulating therapies, including corticosteroids (12, 37, 47, 52), NSAIDs (33), and thalidomide (12) a randomized controlled trial of high-dose oral prednisone in mild to moderate TB-IRIS reported an overall reduction in inpatient days (106)and outpatient therapeutic procedures, but there is no evidence of therapeutic benefit of steroids in either C-IRIS or non-IRIS cryptococcal meningitis (107). Potential risks associated with corticosteroids in immunosuppressed patients include development of Strongyloides hyperinfection, worsening of Kaposi’s sarcoma, or inappropriate administration during culture-positive cryptococcal relapse. Aggressive control of raised intracranial pressure by therapeutic lumbar punctures (59, 105) and optimization of antifungal therapy should take priority, regardless of whether one is able to confirm or exclude the diagnosis of C-IRIS.
Our structured case definitions for C-IRIS and ART-associated cryptococcosis provide tools for future clinical, epidemiologic and immunopathologic studies of C-IRIS, allowing investigators to harmonize diagnostic criteria and facilitating comparison of studies, pooling of data and meta-analyses. Specifically, we recommend future reports avoid pooling paradoxical C-IRIS and ART-associated cryptococcosis cases together as one entity, but to report each separately.
Evaluation of these case definitions will require comparison with expert opinion, given current absence of an objective gold standard. The use of cut-offs in CSF parameters and inflammatory biomarkers in C-IRIS requires particular evaluation (108). Evidence for specific diagnostic cut-offs is conflicting and the entity of “unmasking IRIS” remains controversial, so future alterations to our case definitions are anticipated. In developing case definitions of rheumatologic and other complex syndromes, researchers evaluated the sensitivity and specificity of detailed candidate criteria, and used data-mining techniques to identify the best combination of criteria (109–111). As with TB-IRIS case definitions, these proposed C-IRIS definitions should also be regarded as preliminary until further insights into the immunopathology of IRIS and the development of diagnostic tools permit their refinement.
We acknowledge all participating members of the International Network for the Study of HIV-associated IRIS (INSHI), also special thanks to Mohamed-Yunus Moosa, Tihana Bicanic, Samuel Shelburne, Danielle Cohen, Rosalind Parkes-Ratanshi, David Lalloo, David Meya, Andrew Kambugu, and Tom Harrison for discussion and contributions of unpublished data.
No specific funding was provided for this work. DRB receives support from the United States NIH National Institute of Allergy and Infectious Diseases (K23AI073192), PRB (R34AI081554-01). SDL and GM are funded by the Wellcome Trust, London, UK. LJH is funded by the Medical Research Council, UK (G0700530).
Conflict of interest
No specific approval was sought for this review paper.