Cryptococcus neoformans infection in organ transplant recipients: variables influencing clinical characteristics and outcome.

Unique clinical characteristics and other variables influencing the outcome of Cryptococcus neoformans infection in organ transplant recipients have not been well defined. From a review of published reports, we found that C. neoformans infection was documented in 2.8% of organ transplant recipients (overall death rate 42%). The type of primary immunosuppressive agent used in transplantation influenced the predominant clinical manifestation of cryptococcosis. Patients receiving tacrolimus were significantly less likely to have central nervous system involvement (78% versus 11%, p =0.001) and more likely to have skin, soft-tissue, and osteoarticular involvement (66% versus 21%, p = 0.006) than patients receiving nontacrolimus- based immunosuppression. Renal failure at admission was the only independently significant predictor of death in these patients (odds ratio 16.4, 95% CI 1.9-143, p = 0.004). Hypotheses based on these data may elucidate the pathogenesis and may ultimately guide the management of C. neoformans infection in organ transplant recipients.

Invasive fungal infections have been reported in 5% to 59% of organ transplant recipients (1)(2)(3)(4). Infections due to Cryptococcus neoformans, while less common than those due to Candida and mycelial fungi, are also an important posttransplant complication. The incidence of invasive candidiasis has declined in subsets of organ transplant recipients (e.g., liver transplant patients) as a result of fluconazole use and technologic advances in surgery (5). However, the risk factors and pathogenesis of C. neoformans infection in the transplant setting are poorly understood, and fluconazole prophylaxis is generally not used in the late posttransplant period when cryptococcal infections usually occur. Thus, the incidence and impact of cryptococcal infection in organ transplant recipients are unlikely to diminish in the foreseeable future. Indeed, as incidence of C. neoformans infection in HIV-infected patients has declined, organ transplant recipients have become the group of immunocompromised patients at highest risk for cryptococcosis. The overall death rate in transplant recipients with cryptococcal infection has been 20% to 100% (6)(7)(8)(9). While the predictors of outcome in patients with C. neoformans have been well documented in nontransplant settings (10)(11)(12), predictors in transplant recipients are largely unknown. The unique neurotropism and predilection of C. neoformans to cause central nervous system (CNS) infections are well recognized; CNS has been the most common site for cryptococcal infections. However, 67% of our liver transplant recipients with cryptococcosis who received tacrolimus as primary immunosuppression had cutaneous or osteoarticular lesions; 17% had meningitis (8). Small sample size and lack of comparison with patients on other immunosuppressive regimens, however, precluded meaningful interpretation of these data.
Given the limited number of transplant recipients with C. neoformans infection at individual institutions, accumulating a sufficiently large sample was difficult, so we turned to reports and analyses of cases for valuable data. This review summarizes unique epidemiologic and clinical characteristics of C. neoformans in transplant recipients, as well as variables influencing the outcome of cryptococcal infections after transplantation.

Cases of C. neoformans infection in transplant recipients
were identified with a MEDLINE search through 1998 by cross-referencing the keywords "Cryptococcus neoformans" and "transplantation" or "transplant." Reference lists of original articles and textbooks were reviewed for additional cases. A patient was considered infected if C. neoformans was cultured from a clinical specimen in the presence of signs or symptoms of cryptococcus infection. The onset of infection after transplantation was determined on the basis of detailed case studies; summarized data providing only a mean or range for the group of transplant recipients were excluded. Cryptococcal infection was considered early-onset if it occurred within 12 months and late-onset if it occurred >12 months after transplantation. Predictors or risk factors for death were assessed only in detailed cases for which the variables to be analyzed were explicitly stated.
Unique clinical characteristics and other variables influencing the outcome of Cryptococcus neoformans infection in organ transplant recipients have not been well defined. From a review of published reports, we found that C. neoformans infection was documented in 2.8% of organ transplant recipients (overall death rate 42%). The type of primary immunosuppressive agent used in transplantation influenced the predominant clinical manifestation of cryptococcosis. Patients receiving tacrolimus were significantly less likely to have central nervous system involvement (78% versus 11%, p =0.001) and more likely to have skin, soft-tissue, and osteoarticular involvement (66% versus 21%, p = 0.006) than patients receiving nontacrolimus-based immunosuppression. Renal failure at admission was the only independently significant predictor of death in these patients (odds ratio 16.4, 95% CI 1.9-143, p = 0.004). Hypotheses based on these data may elucidate the pathogenesis and may ultimately guide the management of C. neoformans infection in organ transplant recipients.
Technologies, Cambridge, MA). The χ3 or the Fisher exact test was used to compare categorical variables. Continuous variables (e.g., time of onset) were compared by using the Student t test or the Mann-Whitney U test. Multiple comparisons were done by analysis of variance and the Kruskal-Wallis test. A multiple regression model was used to examine the risk factors for death.

Results
A total of 178 cases of C. neoformans infection in organ transplant recipients were identified (1,(6)(7)(8)(9). Of these, 96 cases were individually detailed, and 82 were summarized in reports containing 2 to 22 cases. Of 178 cases, 145, 20, and 10 were in renal, liver, and heart transplant recipients, respectively. Three cases were reported in lung transplant recipients, and none were described in bowel or pancreas transplant recipients. Patients were 12 to 67 years of age (median 44 years); 78% were male. The mean incidence of C. neoformans infection was 2.8 per 100 transplants (0.3 to 5.3 per 100). The overall incidence was 2.4% in liver, 2.0% in lung, 3.0% in heart, and 2.8% in renal transplant recipients.
Of 127 transplant recipients who could be evaluated, 100 (79%) had azathioprine as the primary immunosuppressive agent, 9 (7%) had tacrolimus, 11 (9%) had cyclosporine, and 7 (6%) had cyclosporine and azathioprine. Of these 127 patients, 78 were also receiving prednisone in various dosages, 5 were not receiving prednisone, and data on prednisone use were unavailable for 44 patients. The incidence of cryptococcosis was 4.5 per 100 transplants in patients who received tacrolimus, 2.4 per 100 transplants in patients who received cyclosporine, and 3.4 per 100 transplants in patients who received azathioprine. These rates did not differ significantly. Rejection episodes preceding cryptococcal infection were documented in 17 (25%) of 67 patients; rejection had occurred a median of 7 months (from 5 days to 49 months) before onset of infection. Eleven (18%) of 62 patients had received augmented immunosuppression (predominantly corticosteroids) within 6 months of onset of cryptococcosis; two patients had received antilymphocyte preparations or OKT3 monoclonal antibodies for the treatment of allograft rejection.
The time to onset varied significantly for different types of organ transplant recipients. The median time to onset after transplantation was 35 months for kidney, 25 months for heart, 8.8 months for liver, and 3 months for lung transplant recipients (p = 0.001). Overall, cryptococcosis developed in 100% of the lung, 75% of the liver, 33% of the heart, and 30% of the kidney transplant recipients within 12 months of transplantation (p = 0.002) ( Table 1).
C. neoformans infection tended to occur later in patients who received azathioprine than in patients who received tacrolimus or cyclosporine (p = 0.16). The median time to onset was 11.4 months after transplantation in patients who received cyclosporine, 9.2 months in patients who received tacrolimus, and 27 months in patients who received only azathioprine-based immunosuppression (p = 0.16). Patients from the northeastern United States were more likely to have early-onset cryptococcosis (i.e., infection within 12 months of transplantation) than other patients (67% versus 31%, p = 0.004). Age, cytomegalovirus infection, or prior rejection episodes did not correlate with early-versus late-onset cryptococcal infection (Table 1).

Death Rate
The overall death rate among organ transplant recipients with cryptococcal infection was 72 (42%) of 172. The death rate was 8 (40%) of 20 for liver, 57 (41%) of 139 for kidney, 6 (60%) of 10 for heart, and 1 (33%) of 3 for lung transplant recipients. Death rates did not differ between patients on tacrolimus and patients on other primary immunosuppressive regimens (33% versus 38%, p >0.05). CNS infection (p = 0.04), renal failure (defined as serum creatinine >1.5 mg/ dL on admission, p = 0.005), and abnormal mental status (p = 0.03) were significant predictors of death in univariate analysis (Table 3   status and absence of headache (p = 0.07) correlated with poor outcome in patients with CNS cryptococcal infection ( Table 4). Presence of fever, CSF pleocytosis, positive blood cultures, and CSF cryptococcal antigen titer did not correlate with outcome (Table 4).

Discussion
C. neoformans infection was documented in 2.8% of the organ transplant recipients, with an overall death rate of 42%. A number of findings in our study have previously not been fully appreciated in the context of cryptococcal infections after transplantation. For example, the type of primary immunosuppression after organ transplantation may influence the predominant clinical manifestation. Patients receiving tacrolimus were less likely to have CNS involvement and more likely to have skin, soft tissue, or osteoarticular involvement due to C. neoformans than patients who received nontacrolimus-based immunosuppression. Furthermore, both tacrolimus and cyclosporine were less likely to be associated with CNS involvement and more likely to be associated with cutaneous infection than azathioprine.
A number of biologic plausibilities exist for this observation. Tacrolimus is a natural macrolide antifungal product (59,60). Although its immunosuppressive effect outweighs its antifungal action in vivo, tacrolimus is toxic to C. neoformans in vitro by inhibition of calcineurin (59)(60)(61). Furthermore, tacrolimus suppresses the growth of C. neoformans at 37 o C but not at 24 o C, which suggests that the target of tacrolimus, calcineurin, is required at higher body temperatures (59,61). Thus, temperature-dependent inhibition of cryptococci by tacrolimus may prevent CNS infection but allow growth of fungus at cooler body sites, e.g., skin, soft tissue, and bone. Cyclosporine also possesses in vitro antifungal activity by inhibition of calcineurin (60,61). However, cyclosporine does not effectively penetrate the CNS, while tacrolimus crosses the blood-brain barrier (61,62). Thus, the relative rarity of meningitis compared with extraneural manifestations of cryptococcosis in patients receiving tacrolimus may merely be due to high cerebrospinal fluid levels of tacrolimus.
Strains of C. neoformans known to be selectively dermatotropic and rhinotropic have been demostrated in animal models (63,64). In addition, C. neoformans serotype D is more likely to be associated with cutaneous lesions (65). However, the precise reason for dermatotropism or the propensity of these strains to occur in transplant recipients receiving calcineurin-inhibiting agents (e.g., cyclosporine and tacrolimus) has not been elucidated.
The immunosuppressive agents (cyclosporine, tacrolimus, and rapamycin) have in vitro activity against fungi, including C. neoformans (59,61,66,67). The antifungal activity of cyclosporine and tacrolimus is mediated by fungal homologs of calcineurin and that of rapamycin through complexes with TOR kinase (61,66). Mutations in calcineurin A and B genes have been shown to confer resistance to cyclosporine and tacrolimus and in FKBP12 gene, to tacrolimus and rapamycin in vitro (66). In addition, TOR I mutants of cryptococci have been identified that are resistant only to rapamycin (66). Despite high seroprevalence of cryptococcal antibodies in early childhood (68), cryptococcal infection is rare in transplant recipients. These data suggest that the immunosuppressive agents currently used may be conferring some degree of protection against Cryptococcus. Whether C. neoformans infections in patients receiving these immunosuppressive agents represent breakthrough infections due to resistant mutants, however, remains to be determined.
Although the susceptibility of transplant recipients to C. neoformans is well recognized, it is not known whether cryptococcal infection in these patients is newly acquired or a reactivation of latent infection. That cryptococcal disease may be due to a reactivation of latent infection is suggested by the following observations in the nontransplant setting: 1) autopsy studies have documented pulmonary granulomas containing C. neoformans in patients who had no history of C. neoformans infection (69); 2) molecular typing in African patients residing in Europe indicated that cryptococcosis resulted from a reactivation of latent infection (70); 3) serologic evidence of C. neoformans infection was documented in most children in New York City in early childhood, even though symptomatic infections were rare (68).
We previously reported that transplant recipients from the northeastern United States were more likely to have cryptococcosis than transplant recipients from other regions of the United States (8). This review shows that cryptococcal infections in patients from the Northeast developed significantly earlier after transplantation than in other patients. Although, there is incontrovertible evidence of primary acquisition of cryptococcosis in isolated case reports (71), our data suggest that C. neoformans may have a predilection for certain geographic areas and that most cryptococcal infections in transplant recipients may result from a reactivation of latent infection.
Epidemiologic studies of C. neoformans have been hampered by lack of sensitive and specific immunologic tests to evaluate the prevalence of latent infection. New immunoblotting assays (68,72), however, have unique implications not only for discerning whether cryptococcal infections result from reactivation or primary acquisition but also for identifying patients at high risk for reactivation or patients never exposed (who may therefore be vulnerable to primary infection).
The relative rarity of cryptococcal infections in pediatric organ transplant recipients has been noted (55). However, the precise reason for this is not known. If cryptococcosis represents reactivation of latent infection in a transplant setting and primary cryptococcal infection is acquired asymptomatically in childhood, it is plausible that pediatric transplant recipients may not yet have acquired the infection. C. neoformans infection is also strikingly rare in bone marrow transplant recipients, possibly because fluconazole prophylaxis is used widely for candidiasis or because thymic regeneration in bone marrow transplant recipients may render T cells more efficacious against cryptococci than T cells present in solid organ transplant recipients (Heitman J, pers. comm.).
Although various clinical manifestations have been described, molluscum contagiosum-like lesions are characteristic of cutaneous cryptococcosis in HIV-infected patients. In the transplant setting, cutaneous cryptococcal infection most frequently mimicked (and was clinically indistinguishable from) bacterial cellulitis. A unique propensity for the extremities to be the site of cutaneous cryptococcosis in transplant recipients was noted in this review; 94% of the patients with cutaneous C. neoformans infections had lesions on upper or lower extremities. Cutaneous cryptococcosis, however, represents disseminated infection and should be treated with systemic antifungal agents.
Elevated CSF pressure without evidence of obstructive hydrocephalus, believed to result from basilar meningitis and impaired reabsorption of CSF across arachnoid villi, has recently been recognized as an important complication of cryptococcal meningitis (73). HIV studies have shown that high baseline opening pressure in patients with cryptococcal meningitis correlated inversely and independently with survival. CSF opening pressure was recorded infrequently in organ transplant recipients. However, all 17 patients in whom such a measurement was conducted had intracranial pressure >140 mm of H 2 O; the death rate in these patients was 8 (47%) of 17. These data underscore the need for assessing intracranial pressure in all patients with cryptococcal meningitis, including organ transplant recipients.
Overall, 72 (42%) of 172 of the transplant recipients with C. neoformans infection died. Preexistent renal failure was an independently significant predictor of death in transplant recipients with cryptococcosis. Renal failure has been proposed to increase the risk for cryptococcosis (62). Uremia decreased lymphocyte transformation and chemiluminescence by splenic cells in C. neoformans-infected mice (74).
This review summarizes the overall impact and highlights the key features of C. neoformans infection in organ transplant recipients. These include the effect of primary immunosuppressive agents on the clinical manifestations of cryptococcosis; geographic diversity in the incidence and onset of infection posttransplantion; and variables influencing outcome, specifically in the transplant setting. More importantly, however, we have identified a number of outstanding questions with implications relevant to elucidating the pathogenesis of C. neoformans infection.
These questions involve the biologic basis of tissue tropism, reasons for the predominance of dermatotropic strains in recipients of tacrolimus, the role or virulence of immunosuppressive-agent resistant C. neoformans mutants in the transplant setting, and the relative rarity of cryptococcal infections in pediatric and bone marrow transplant recipients. We caution that a retrospective study may carry unknown bias. In this regard, our data may be considered hypotheses generating.
Dr. Husain is an infectious diseases fellow at the University of Pittsburgh Medical Center. His research interests include infections in immunocompromised hosts, in particular fungal infections in organ transplant recipients.