Abstract
Background.Risk factors for Nocardia infection in organ transplant recipients have not been formally assessed in the current era of transplantation.
Methods.We performed a matched case-control study (1 : 2 ratio) between January 1995 and December 2005. Control subjects were matched for transplant type and timing. Univariate matched odds ratios were determined and conditional logistic regression was performed to identify independent risk factors. Clinical and microbiological characteristics of all case patients were reviewed.
Results.Among 5126 organ transplant recipients, 35 (0.6%) were identified as having cases of Nocardia infection. The highest frequency was among recipients of lung transplants (18 [3.5%] of 521 patients), followed by recipients of heart (10 [2.5%] of 392), intestinal (2 [1.3%] of 155), kidney (3 [0.2%] of 1717), and liver (2 [0.1%] of 1840) transplants. In a comparison of case patients with 70 matched control subjects, receipt of high-dose steroids (odds ratio, 27; 95% confidence interval, 3.2–235; P = .003) and cytomegalovirus disease (odds ratio, 6.9; 95% confidence interval, 1.02–46; P = .047) in the preceding 6 months and a high median calcineurin inhibitor level in the preceding 30 days (odds ratio, 5.8; 95% confidence interval, 1.5–22; P = .012) were found to be independent risk factors for Nocardia infection. The majority of case patients (27 [77%] of 35) had pulmonary disease only. Seven transplant recipients (20%) had disseminated disease. Nocardia nova was the most common species (found in 17 [49%] of the patients), followed by Nocardia farcinica (9 [28%]), Nocardia asteroides (8 [23%]), and Nocardia brasiliensis (1 [3%]). Of the 35 case patients, 24 (69%) were receiving trimethoprim-sulfamethoxazole for Pneumocystis jirovecii pneumonia prophylaxis. Thirty-one case patients (89%) experienced cure of their Nocardia infection.
Conclusions.Receipt of high-dose steroids, history of cytomegalovirus disease, and high levels of calcineurin inhibitors are independent risk factors for Nocardia infection in organ transplant recipients. Our study provides insights into the epidemiology of Nocardia infection in the current era, a period in which immunosuppressive and prophylactic regimens have greatly evolved.
Nocardia species are ubiquitous environmental microorganisms that are found worldwide and belong to a diverse group of bacteria known as aerobic actinomycetes [1, 2]. Thus far, >50 species of the genus Nocardia have been characterized [1], with at least 16 species being implicated in human infection [2]. The most common of these include Nocardia asteroides sensu stricto, Nocardia brasiliensis, Nocardia farcinica, Nocardia nova, Nocardia otitidiscaviarum, and Nocardia transvalensis.
In most instances, Nocardia is an opportunistic pathogen, with the majority of infections occurring in immunocompromised hosts, including those with long-term corticosteroid exposure, malignancy, HIV infection, and a history of transplantation [3]. Primary cutaneous infection with N. brasiliensis after a traumatic injury is an exception; the majority of affected individuals are immunocompetent. The frequency of Nocardia species infection in organ transplant recipients is 0.7%–3%, with the majority of infections being described in kidney, heart, and liver transplant recipients [4–7]. Over time, changes in immunosuppressive regimens (most importantly, the introduction of cyclosporin) have led to a significant reduction in the number of Nocardia infections in heart and kidney transplant recipients [8, 9]. Such changes most likely relate to an improved specificity of immune suppression and a reduction in corticosteroid exposure. The first year after transplantation carries the greatest risk for Nocardia species infection, with several reports indicating a temporal association with episodes of allograft rejection [6, 10, 11]. Cytomegalovirus (CMV) infection, which has immunomodulatory effects [12], is also thought to be an important cofactor in the development of Nocardia infection [4], but this has yet to be confirmed in a formal analysis.
Risk factors for Nocardia infection in organ transplantation have not been formally assessed for 25 years [6], a period in which significant advances have been made in immunosuppressive regimens and antimicrobial prophylactic strategies. Therefore, we performed a study with the following objectives: (1) to systematically assess the risk factors for Nocardia infection in organ transplant recipients during the modern era, using a 1 : 2 matched case-control study design; and (2) to describe the clinical, radiological, and microbiological characteristics of the largest number of Nocardia infections in organ transplant recipients yet reported from a single institution.
Materials and Methods
Study population. The study was conducted at the University of Pittsburgh Medical Center (Pittsburgh, PA), a tertiary care institute specializing in organ transplantation, including liver, kidney, heart, lung, pancreas, intestinal, and multivisceral transplantation. All patients with culture-proven Nocardia infection from January 1995 through December 2005 were included in the analysis. Patients were initially identified from the hospital's microbiology database. This was followed by a detailed review of the patients medical records. Nocardia infection was defined as a positive culture in the presence of clinical and/or radiological features of infection. Disseminated infection was defined as involvement of ⩾2 noncontiguous organs, disease of the CNS, or the presence of Nocardia species in blood samples. The clinical significance of each isolate was determined by 2 transplant infectious diseases physicians blinded to the other's decision. Concordance was obtained for all isolates. Ten of the cases of Nocardia infection have been previously described [13]. The study was approved by the institutional review board of the University of Pittsburgh Medical Center.
For the purposes of risk factor analysis, matched control subjects were obtained from the same institution. Two control subjects were included for each case patient with Nocardia infection. The control subjects were matched for 2 characteristics: (1) transplant type and (2) time of transplantation. The recipients who received transplants immediately before and after the index case patient and who survived at least as long as the time to diagnosis of Nocardia infection in the index case patient were categorized as control subjects.
Clinical data. Data were systematically collected from the hospital's computer-based medical records. For the case patients, the “time of event” was defined as the time of Nocardia infection; for the control subjects, the time of Nocardia infection in the corresponding case patient was used. Variables analyzed for both case patients and control subjects included patient demographic data, transplant type, prior transplantation, presence of diabetes mellitus, renal function at the time of the event, maintenance immunosuppression, an elevated mean calcineurin inhibitor level in the preceding 30 days (>15 µg/mL for tacrolimus and >300 ng/mL for cyclosporin), history of high-dose prednisone therapy in the preceding 6 months (⩾20 mg of prednisone for⩾1 month or >2 pulses of 1 g of intravenous methylprednisolone), receipt of a lymphocyte-depleting antibody (alemtuzumab [CD-52 monoclonal antibody], rabbit antithymocyte globulin, muromonab-CD3, or CD-3 monoclonal antibody and rituximab [CD-20 monoclonal antibody]) or lymphocyte-modulating antibody (daclizumab [IL-2 receptor antagonist]) in the preceding 12 months, prophylactic antimicrobials at the time of event, allograft rejection in the preceding 6 months, treatment for allograft rejection, history of CMV infection and/or disease in the preceding 6 months, history of another opportunistic infection in the preceding 3 months, mean lymphocyte and neutrophil counts within 30 days of the event, and mortality at 6 months after the event. All case patients had detailed clinical and microbiological information collected. Two separate investigators reviewed all case patients and control subjects independently.
CMV infection was defined as the detection of viremia by pp65 antigenemia in the absence of symptoms. CMV disease was defined as symptoms and/or signs of an organ-specific illness in conjunction with the detection of CMV in tissue by culture, immunohistochemical analysis, or in situ hybridization [14]. Other opportunistic infections were defined according to current guidelines [14–16].
Microbiology. Organisms were initially identified as Nocardia species by the microbiology department at our institution on the basis of their ability to grow aerobically, their colonial and microscopic morphology, and demonstration of partial acid-fast staining. Organisms were then sent to the University of Texas Health Science Center (Tyler, TX) for confirmation and susceptibility testing. From 2003, in vitro susceptibility testing was performed by broth microdilution according to CLSI guidelines [17]. Prior to this, broth microdilution was performed for sulfamethoxazole, and disk diffusion was performed for other antimicrobials.
Statistical analysis. Statistical analyses were performed using SAS software, version 8.2 (SAS Institute). Univariate matched ORs were calculated for the potential risk factors for Nocardia infection. Multivariable conditional logistic regression was performed using the SAS procedure PHREG to determine independent risk factors for Nocardia infection. Variables with a P value of <.2 on univariate analysis were included in the multivariable model. Variables were assessed for confounding and collinearity. All P values were 2-tailed, and a value of <.05 was considered to be statistically significant.
Results
Epidemiology.A total of 35 Nocardia infections were identified among 5126 transplant recipients at the University of Pittsburgh Medical Center from January 1995 through December 2005. This represented 0.6% of all transplant recipients, and included 18 (3.5%) of 521 lung recipients, 10 (2.5%) of 392 heart recipients, 2 (0.1%) of 1840 liver recipients, 3 (0.2%) of 1717 kidney recipients, and 2 (1.3%) of 155 small bowel or multivisceral recipients. There were no cases of Nocardia infection among 180 pancreas recipients, 297 simultaneous kidney and pancreas recipients, or 24 heart and lung recipients. The median number of Nocardia infections diagnosed each year was 3 (range, 0–6 cases), with 21 case patients (60%) receiving a diagnosis in the latter half of the study period. The majority of case patients (22 [63%]) received a diagnosis within 1 year after transplantation, with 5 (14%) receiving a diagnosis >5 years after transplantation.
Risk factors forNocardiainfection.The risk factors analyzed for the 35 case patients and 70 matched control subjects are shown in table 1. On univariate analysis, allograft rejection (OR, 6.7; 95% CI, 2.2–20; P < .001), high-dose prednisone in the preceding 6 months (OR, 26; 95% CI, 3.4–195; P = .002), receipt of a lymphocyte-depleting antibody in the preceding 12 months (OR, 6.3; 95% CI, 1.8–23; P = .005), a high median calcineurin inhibitor level in the preceding 30 days (OR, 4.1; 95% CI, 1.6–11; P = .004), and receipt of antifungal prophylaxis (OR, 3.0; 95% CI, 1.01–8.9; P = .047) were significantly associated with Nocardia infection. Patients with Nocardia infection had significantly higher prednisone dosages (P = .007), lower lymphocyte counts (P = .003), and higher neutrophil counts (P = .006) at the time of infection, compared with control subjects.
Univariate analysis of risk factors for Nocardia infection in organ transplant recipients.
After performing a conditional logistic regression model (table 2), receipt of high-dose prednisone (OR, 27; 95% CI, 3.2–235; P = .003) and history of CMV disease (OR, 6.9; 95% CI, 1.02–46; P = .047) in the preceding 6 months and having a high median calcineurin inhibitor level in the preceding 30 days (OR, 5.8; 95% CI, 1.5–22; P = .012) remained as independent risk factors for Nocardia infection.
Independent risk factors for Nocardia infection as determined by multivariable conditional logistic regression analysis.
Of the 35 case patients, 24 (69%) were receiving trimethoprim-sulfamethoxazole (TMP-SMZ) for prophylaxis. The majority were prescribed a single-strength tablet (80 mg/400 mg), administered either 3 times per week (for 11 [46%] of the case patients), twice per week (9 [38%]), once per week (2 [8%]), or once per day (1 [4%]). One case patient (4%) received a double-strength tablet (160 mg/800 mg) twice per week. The median duration of TMP-SMZ prophylaxis before the diagnosis of Nocardia species infection was 264 days (range, 29–4990 days). On univariate analysis, receipt of TMP-SMZ was not shown to be protective for Nocardia infection (P = .5) .
Clinical characteristics of patients withNocardiainfection.The clinical characteristics of the 35 transplant recipients with Nocardia infection are shown in table 3. Of the 27 transplant recipients with pulmonary disease only, 6 (22%) had night sweats, fevers, or constitutional symptoms with no respiratory symptoms. Seven transplant recipients (20%) had disseminated infection, including 4 lung transplant recipients (3 with brain abscesses and 1 with bacteremia and septic arthritis), 1 heart transplant recipient (pulmonary and cutaneous infection), 1 liver transplant recipient (bacteremia), and 1 intestinal transplant recipient (bacteremia and prostatitis) [18].
Clinical characteristics of 35 organ transplant recipients who developed Nocardia infection from January 1995 through December 2005 at the University of Pittsburgh Medical Center (Pittsburgh, Pennsylvania).
Coexisting infections and temporal relationship to allograft rejection.Seven recipients (20%) had a coexisting infection (table 3). The clinical pattern of Nocardia infection in this subgroup was similar to that for the remaining case patients, although a higher rate of treatment in an intensive care unit (3 [43%] of 7 patients vs. 3 [11%] of 28 patients; P = .08) and a significantly greater mortality rate (3 [43%] of 7 patients vs. 2 [7%] of 28 patients; P = .04) were seen among patients with a coexisting infection. Of the 35 case patients, 17 (49%) experienced an episode of allograft rejection in the 6 months preceding the Nocardia infection. The median time from the last episode of rejection to Nocardia infection was 107 days (range, 1–176 days).
Microbiology and susceptibility profiles.The species distribution and susceptibility profile of the Nocardia isolates are shown in table 4. N. nova was the most common species (17 isolates [49%]), followed by N. farcinica (9 isolates [28%]), N. asteroides (8 isolates [23%]), and N. brasiliensis (1 isolate [3%]). No significant difference in rates of disseminated disease was noted between Nocardia species. No antimicrobial had susceptibility test results indicating that all isolates were susceptible. Amikacin (97% susceptible), imipenem (94% susceptible), and sulfamethoxazole (97% susceptible) had the greatest in vitro activity. Ceftriaxone had poor activity against N. farcinica (13% susceptible). Varied results were seen for the remaining antibiotics (table 4).
Species distribution and susceptibility profile of the Nocardia isolates obtained from January 1995 through December 2005 at the University of Pittsburgh Medical Center (Pittsburgh, Pennsylvania).
Treatment.Treatment data were available for 34 (97%) of the transplant recipients. All case patients received empirical combination therapy. The most common backbone of empirical therapy was intravenous ceftriaxone and oral or intravenous TMP-SMZ (administered to 28 case patients [82%]). This was administered either alone (6 case patients [18%]) or in combination with imipenem (15 case patients [44%]), imipenem and minocycline (1 case patient [3%]), meropenem (1 case patient [3%]), amikacin (1 case patient [3%]), amikacin and ciprofloxacin (1 case patient [3%]), ciprofloxacin (1 case patient [3%]), levofloxacin (1 case patient [3%]), or minocycline (1 case patient [3%]). The second most common backbone of empirical therapy was imipenem and oral or intravenous TMP-SMZ (administered to 4 case patients [11%]). This was administered either alone (1 case patient [3%]) or in combination with linezolid (2 case patients [6%]) or amikacin (1 case patient [3%]). The final 2 case patients were treated with ceftriaxone, imipenem or meropenem, and minocycline. All case patients except 1 received empirical treatment with an antibiotic that was active against the corresponding isolate. A lung recipient with N. asteroides infection was treated with empirical imipenem (to which the infecting strain had intermediate susceptibility) and oral TMP-SMZ (to which the infecting strain was resistant). Despite this, the patient recovered well, and when susceptibility results were available, therapy was changed to doxycyline and gatifloxacin.
Among the 29 case patients who received definitive therapy after susceptibility results were known, oral TMP-SMZ was the most common therapy (administered to 22 case patients [76%]). This was administered either alone (15 case patients [52%]) or in combination with minocycline (4 case patients [14%]), amoxicillin-clavulanate (2 case patients [7%]), or linezolid (1 case patient [3%]). Four patients received ceftriaxone, either alone (1 case patient [3%]) or in combination with imipenem and TMP-SMZ (1 case patient [3%]), imipenem and minocycline (1 case [3%]), or linezolid (1 case [3%]). Two further case patients received monotherapy with amoxicillin-clavulanate and clarithromycin, respectively. Of the 2 case patients who survived with CNS involvement, 1 received definitive therapy with TMP-SMZ alone, and 1 received a combination of TMP-SMZ and linezolid. Treatment duration data were available for 32 case patients (91%). The median duration was 6 months (range, 3 weeks to 12 months), and this was the most common treatment duration in the study population (18 case patients [56%]). Treatment duration was longer for those with disseminated disease, who had a median treatment duration of 7.5 months (range, 6–12 months). Nine case patients (26%) developed a drug-related adverse event requiring treatment modification.
Outcome.Thirty-one case patients (89%) experienced cure of Nocardia infection, including 6 of 7 case patients with disseminated disease. Five case patients (14%) died during therapy. In 4 of these cases, death was at least partly attributable to Nocardia infection: 2 of the patients who died were lung transplant recipients (1 of whom had pulmonary infection due to N. farcinica and 1 of whom had multiple brain abscesses due to N. nova), 1 was a heart transplant recipient with pulmonary infection due to N. nova, and 1 was a multivisceral transplant recipient with pulmonary infection due to N. asteroides. Compared with 70 matched control subjects, recipients with Nocardia infection had a significantly greater 6-month mortality rate (5 [14%] of 35 patients with Nocardia infection died, compared with 1 [4%] of 70 matched control subjects; P = .015).
Discussion
The frequency of Nocardia infection in organ transplant recipients at our institution was 0.6%, a figure that is consistent with recently published literature [4, 19]. The highest frequency was in lung transplant recipients, followed by heart, small bowel, kidney, and liver transplant recipients. Such data conflict with data reported previously, in which kidney, heart, and liver transplant recipients have had the highest rates of Nocardia infection [20–22]. Two key points may explain these epidemiological differences. First, transplant centers are now performing transplantation of a greater breadth of organ types. Also, the absolute number of transplantations is increasing, especially for relatively newer organ types, such as lungs and intestines. Lung transplantation has only emerged in the past 20 years, with 33 case patients recorded in the United States by the Organ Procurement and Transplantation Network in 1988, compared with 1406 case patients in 2005 [23]. Second, much of the literature regarding the susceptibility of organ transplant recipients to Nocardia infection predates the expansion of these newer transplant types [20], and such studies are most often derived from single-organ transplantation programs [5–7]. In the present study, we evaluated, to the best of our knowledge, the broadest range of transplant types reported in the literature to date, and we provide an update on the frequency of Nocardia infection in the modern era of transplantation.
To our knowledge, there has only been 1 other matched case-control study published in the literature that assesses the risk factors for Nocardia infection in organ transplant recipients [6]. This study, published over 25 years ago, involved 21 cardiac transplant recipients with Nocardia infection from a single institution. These patients were compared with all other cardiac transplant recipients without Nocardia infection, as well as a time-matched cohort (1 : 1 ratio). No risk factors were identified in this analysis [6]. Risk factors for Nocardia infection in kidney transplant recipients have repeatedly been described in previous reviews [3, 7, 21], although significant limitations exist with the data. Many of the referenced studies were designed to assess overall infection risk among kidney transplant recipients, rather than risk of Nocardia infection alone [24, 25]. One study reported the infectious etiologies causing death in kidney transplant recipients, with only 2 case patients with Nocardia infection included [26]. A further study identified that multiple early-rejection episodes were a risk factor for fungal and nocardial infections [10], although only 4 case patients with Nocardia infection were included. Finally, such studies are over 28 years old and, given the changes in prophylactic and immunosuppressant regimens used in transplantation, a reevaluation of the frequency, risk factors, and outcomes of Nocardia infection was necessary.
After performing a matched case-control study (1 : 2 ratio), we determined that high-dose prednisone and a history of CMV disease in the preceding 6 months and an elevated mean calcineurin inhibitor level in the preceding 30 days were independent risk factors for Nocardia infection in organ transplant recipients. Such variables have clear biological plausibility. Prednisone is a broadly acting immunosuppressant, affecting both the innate and adaptive immune responses, and has previously been associated with the development of Nocardia infection [2, 7, 20]. The immunomodulatory effects of CMV are increasingly being recognized [12], with previous reports associating it with other viral, bacterial, fungal, and protozoan infections [27–29]. To our knowledge, its role as a risk factor for Nocardia infection has not been formally assessed, and our data further support the importance of aggressive management of CMV disease in preventing the “indirect” effects of CMV infection. The calcineurin inhibitors, cyclosporin and tacrolimus, inhibit T cell activation by binding to cyclophilin and FK-binding protein 12, respectively [30]. T cells are essential for an adequate host response against Nocardia infection, primarily through the activation of macrophages and the stimulation of a cellular immune response [20]. Allograft rejection has often been thought of as a predisposing variable for Nocardia infection in organ transplant recipients [6, 10, 11], but confounding variables, such as augmented immunosuppression, are often not accounted for. Our data illustrate the importance of this, whereby allograft rejection was a highly significant risk factor on univariate analysis (P < .001), but, after adjusting for confounders in the multivariable model, it was no longer significant.
At our institution, all organ transplant recipients are prescribed TMP-SMZ for Pneumocystis jirovecii pneumonia prophylaxis, most commonly at a dosage of 1 single-strength tablet 3 times per week. Twenty-four (69%) of the 35 transplant recipients with Nocardia infection were receiving prophylactic doses of TMP-SMZ at the time of their infection. All breakthrough Nocardia isolates (except 1) were susceptible to this agent. After analyzing this variable using matched control subjects, TMP-SMZ prophylaxis was not shown to be protective against Nocardia infection. Breakthrough nocardial infections that occur while patients are receiving TMP-SMZ have previously been reported in both solid-organ [8] and bone marrow transplant recipients [31]. In patients with HIV infection, use of TMP-SMZ as P. jirovecii pneumonia prophylaxis has often been postulated as a reason for the low incidence of Nocardia infection [21]. However, the dose administered in this population has traditionally been higher than the dose administered to our organ transplant recipients [32]. These data indicate that the use of TMP-SMZ in low dosages for P. jirovecii pneumonia prophylaxis is unreliable in preventing Nocardia infection in organ transplant recipients, despite favorable susceptibility results.
Despite the absence of randomized, controlled trials investigating the treatment of Nocardia infection, combination therapy is frequently used. The reasons for this include the diversity of species and their susceptibility nuances, the prolonged period before susceptibility results are known (if susceptibility testing is performed at all), and the severity of illness, especially in cases of disseminated disease. The ideal choice of agents is less established. The most common drug regimen for empirical therapy in our cohort was therapy with TMP-SMZ, ceftriaxone, and imipenem. Such a combination had excellent activity against all isolates. Murine models of CNS and pulmonary nocardiosis have indicated that the combination of amikacin and imipenem is most effective [33, 34]. Unfortunately, aminoglycoside use in organ transplant recipients is greatly limited by the high rate of preexisting renal dysfunction (49% in our study). Linezolid, the first agent in the class of oxazolidinones, has shown excellent in vitro activity against all of the clinically relevant species of Nocardia [35]. Clinical success has also been reported [36]. Unfortunately, bone marrow and other toxicities, which were seen in one-half of our patients, limit the use of linezolid for prolonged therapy. However, its role as initial therapy for the stabilization of severely unwell, immunocompromised patients with disseminated disease is attractive.
In conclusion, we have formally assessed the risk factors for Nocardia infection in the current era of organ transplantation, showing that high-dose steroids, CMV disease, and high levels of calcineurin inhibitors are independent risk factors. The deficiencies of low dose TMP-SMZ, as used for P. jirovecii pneumonia prophylaxis, for the prevention of Nocardia infection are well illustrated. Given the rapidly increasing population of immunocompromised patients, Nocardia infection is likely to continue to be an important opportunistic disease.
Acknowledgments
We thank the clinicians who cared for the involved patients and the staff of the Microbiology Department of the University of Pittsburgh Medical Center (Pittsburgh, PA), for processing the specimens.
Potential conflicts of interest.S.H. serves on the speakers' bureaus for Pfizer and Schering Plough and has received grant support from Astellas and Enzon. D.L.P serves on the speakers' bureaus for Merck, Roche, Elan, and Cubist Pharmaceuticals and has received grant support from Merck, AstraZeneca, and Elan Pharmaceuticals. All other authors: no conflicts.
Comments