We performed a multicenter retrospective cohort study of patients evaluated at Mayo Clinic facilities in Minnesota, Florida, and Arizona. We reviewed all culture-positive microbiologic specimens of Nocardia spp. during December 2011–November 2018. A previous study from Mayo Clinic Florida published clinical outcomes for patients with invasive nocardiosis during 1998–2018 (13); the Florida cohort for this study included all patients identified during December 2011–November 2018 from that study. We defined invasive nocardiosis diagnosis as culture-positive for Nocardia spp. together with clinical or radiographic evidence of organ involvement. We defined disseminated disease if infection was identified by culture or radiographic imaging in >2 noncontiguous organs, with Nocardia isolated from >1 site, or by a single positive blood culture.

Of 268 patients with invasive nocardiosis during the study period, 129 (48.2%) were enrolled at the Mayo Clinic site in Minnesota, 87 (32.4%) in Arizona, and 52 (19.4%) in Florida. We recorded data on demographics, coexisting conditions, and organs transplanted by study site (Table 1). N. nova complex, 43 (33.4%) cases, was the predominant species in Minnesota and N cyriacigeorgica, 36 (41.4%) cases, in Arizona. In Florida, the most common species were N. brasiliensis, 9 (17.3%) cases, and N. cyriacigeorgica and N. farcinica, 8 (15.4%) cases each (Table 2). Other species accounted for 12.4% of cases in Minnesota, 8% in Arizona, and 26.9% in Florida. In Arizona, N. wallacei accounted for 13.8% and N. transvalensis complex for 11.5% of isolates, but they each constituted <2% of isolates in Florida and Minnesota.

Of the 268 patients, 82 (30.6%) were transplant recipients; 70 received solid organ and 12 hematopoietic stem cell transplants. Median time from transplant to diagnosis of invasive nocardiosis was 12 months (IQR 5–54 months). Kidneys, 33 (40.2%) cases, were the most common transplanted organ; more than half (17/33, 51%) of patients underwent transplantation in Arizona. Most heart recipients (9/13, 69%) also underwent transplantation in Arizona. Most (12/13, 92%) lung recipients underwent transplantation in Florida; however, during the study period, lung transplants were not performed at Mayo Clinic Arizona. Overall mean age of all transplant recipients was 63.8 years (SD 15.1). Chronic lung disease (58.6%) and diabetes (17.9%) were 2 of the most common coexisting conditions (Table 1).

We found isolated nocardial pulmonary involvement in 84.4% of non–transplant patients and 73.2% of transplant patients. Rate of dissemination was 12.3% in the total cohort: 9.7% in non–transplant patients, 18.3% in transplant patients (Table 2). We documented central nervous system (CNS) involvement in 12.2% of transplant recipients and 3.2% of non–transplant patients. (Table 2). The most commonly identified species among transplant recipients were N. farcinica (20.7%, n = 17) and N. cyriacigeorgica (19.5%, n = 16) and among non–transplant patients, N. cyriacigeorgica (38.0%, n = 52), N. nova complex (23.7%, n = 44), and N. farcinica (16.1%, n = 30).

When we analyzed the involvement of diseased organs in transplant patients, we found that Nocardia caused isolated pulmonary disease in 60/82 (75.0%) case-patients. Isolated lung disease was the predominant manifestation for most species: N. cyriacigeorgica caused isolated lung disease in 86.7% of transplant patients and combined lung and CNS disease in 13.3% of cases, and N. farcinica resulted in isolated lung disease in 76.5% of transplant patients and disseminated disease to the lungs, CNS, and skin/soft tissue in 11.8% of patients. Similarly, N. nova complex caused isolated pulmonary disease in 83.3% and isolated skin/soft tissue disease in 16.7%. In contrast, N. brasiliensis resulted in isolated skin/soft tissue disease in 60% of transplant patients and isolated lung disease in only 20%. Only 16/268 (6.0%) of all patients and 10/82 (12.2%) transplant patients had nocardial CNS involvement; N. farcinica (42.8%, n = 6), N. cyriacigeorgica (14.3%, n = 2), and N. wallacei (14.3%, n = 2) were the species most often associated with CNS disease. Rates of CNS involvement among transplant patients were similar across all 3 sites: 4.7% in Minnesota, 5.7% in Arizona, and 9.6% in Florida.

Antimicrobial susceptibilities varied among the most commonly isolated Nocardia spp. (Table 3). All Nocardia isolates were 100% susceptible to linezolid. Analysis of species-specific patterns of susceptibility revealed almost no differences by geographic location in susceptibility to antimicrobials across the 3 Mayo Clinic sites. Trimethoprim/sulfamethoxazole (TMP/SMX) susceptibility was 99.5% among non–transplant patients, 96.3% among transplant recipients, and 99% for the total cohort. Exceptions among Nocardia spp. susceptibility to TMP/SMX were 94% for N. abscessus and 92.0% for those listed collectively as other species. Susceptibility to ceftriaxone ranged from 0% for N. brasiliensis to 100% for N. brevicatena/N. paucivorans. Similarly, susceptibility to imipenem varied; N. cyriacigeorgica (97%), N. nova complex (98%), and N. brevicatena/N. paucivorans (100%) were the most susceptible species (Table 3). For all antimicrobial drugs administered during initial empiric therapy, we calculated the percentage retained in subsequent treatment regimens after analysis of Nocardia antimicrobial susceptibilities (Table 4).

We also analyzed overall Nocardia susceptibility patterns in transplant and non–transplant patients. Susceptibility was 51.9% for moxifloxacin and 32.1% for ciprofloxacin in transplant patients and 34.1% for moxifloxacin and 20.2% for ciprofloxacin in non–transplant patients. Imipenem was 72.8% susceptible in transplant patients and 80.3% in non–transplant patients; linezolid was 98.8% susceptible in transplant patients and 100.0% in non–transplant patients. TMP/SMX was 96.3% susceptible in transplant patients and 99.5% in non–transplant patients; ceftriaxone was 32.1% susceptible in transplant patients and 43.7% in non–transplant patients. Among nocardiosis patients, 155/186 (90.6%) non–transplant patients and 66/82 (80.5%) transplant recipients survived after 1 year. Rates of all-cause mortality at 1 year after diagnosis were similar for all 3 Mayo Clinic sites and different Nocardia spp.

We identified 268 patients (82 transplant recipients and 186 non–transplant recipients) with invasive nocardiosis during 2011–2018 for this retrospective cohort study. Participants were patients at 3 Mayo Clinic tertiary care and high-volume transplant sites in Minnesota, Arizona, and Florida within the United States, which provided a unique opportunity to assemble a large cohort of Nocardia infected case-patients from within the entire Mayo Clinic database.

We described clinical, microbiologic, drug susceptibility, and outcome data in patients with nocardiosis at each geographic site. Nocardia spp. varied by geographic site where diagnosis occurred, correlating with published differences in geographic distribution (2). The most common species of Nocardia isolated in Minnesota were N. nova complex, N. farcinica, and N. cyriacigeorgica. In Arizona, N. cyriacigeorgica, N. farcinica, and N. wallacei were the most common species. N. brasiliensis was the most common species in Florida, which might reflect the organism’s preference for tropical climates (4). Some patient demographics and clinical characteristics differed by site and might contribute to differences among species in occurrence, susceptibilities, and clinical outcomes.

Isolated lung involvement was the most common manifestation of Nocardia infection at all 3 sites: 92% in Arizona, 82% in Minnesota, and 60% in Florida. Possibly the arid climate of the desert southwest in Arizona presents greater opportunity for airborne dispersal of Nocardia and subsequent inhalation and pulmonary infection (similar to the transmission kinetics for coccidioidomycosis). Primary skin/soft tissue infection was present in 19.2% of cases in Florida, more common than in the other locations, likely related to a higher rate of N. brasiliensis infection (17.3%) in Florida; N. brasiliensis is known to cause isolated cutaneous infection (3,14). Patients in Florida had the highest rate of disseminated nocardiosis (21.2%). Rates of death at 1 year were similar among the 3 Mayo Clinic sites. Rates of death based on the species of Nocardia were also similar, which is contrary to a recent study indicating a higher death rate for N. farcinica (15). N. farcinica was most commonly associated with CNS involvement in our cohort, which is consistent with previous reports (16,17).

All Nocardia isolates in our study were susceptible to linezolid, in agreement with other studies (18,19). The 3 most common Nocardia spp. implicated in invasive infection in this study, N. cyriacigeorgica, N. farcinica, and N. nova complex, were 100% susceptible to TMP/SMX. Susceptibility profiles for >1,200 Nocardia isolates reported in another study also noted universal susceptibility to linezolid. TMP/SMX resistance was rare (2%), except among N. pseudobrasilensis (31%) and N. transvalensis complex (19%) (20). In our study, we noted only 3.7% resistance to TMP/SMX among Nocardia isolates in transplant recipients, which is in agreement with other studies that have reported low overall resistance rates (21–23). However, ongoing surveillance for such trends is paramount. In patients with Nocardia infection, especially those receiving substantial immune-suppression treatments, using combination antimicrobial therapy before availability of susceptibility testing results is imperative. Combination therapy should always include either linezolid or TMP/SMX. Based on our susceptibility data, an appropriate second antimicrobial choice would be imipenem, which has the added advantage of excellent CNS penetration. We also calculated the percentage of each antimicrobial in our initial empiric regimen that was retained in subsequent regimens after susceptibilities to all antimicrobials were known (Table 4).

TMP/SMX dosage for Pneumocystis jirovecii pneumonia prophylaxis after organ transplantation is not necessarily protective against invasive Nocardia infection (9,13). Guidelines addressing Nocardia infections in solid organ transplantation suggest that TMP/SMX prophylaxis might be helpful in preventing primary Nocardia infection or relapse after treatment, although infections can occur despite prophylaxis. For secondary prophylaxis, a TMP/SMX dosage of 1 double-strength tablet daily (dosage-adjusted for renal function) has been offered as a consideration, although that recommendation was weak and supported by low-quality evidence (16).

A recent study reported a death rate after 1 year of 16.8% in solid organ transplant recipients with Nocardia infection. Independent risk factors for death included liver transplantation and time from symptom onset to seeking treatment; disseminated infection was not associated with increased death (15). Our study found death rates after 1 year of 19.5% among transplant recipients and 9.4% among non–transplant patients, an observation concordant with findings in a larger study from Europe of nocardiosis in solid organ transplant recipients (24). However, another recent publication noted similar death rates for transplant and non–transplant recipients (13). Retrospective designs, different underlying risk factors for infection, and variable definitions used in published studies make drawing firm conclusions difficult.

Among the limitations of our study, we did not collect detailed clinical or radiographic results or match specific treatment information with outcomes. However, those data have been elaborated in previous studies, including from Mayo Clinic (13,15). The primary purposes of our study were to determine regional differences among patients with risk factors, geographic distribution of Nocardia spp., organs involved by geographic location and species of Nocardia, and differences in antimicrobial susceptibility. Second, even though patients were from distinct geographic locations, they were all primarily managed within the Mayo Clinic system. Patients from other regions and institutions were not represented. However, the substantial climatic and environmental differences among the 3 regions studied—midwestern, southeastern, and southwestern United States—led us to decide on identifying unique trends in Nocardia spp. and disease manifestations for those study sites. A larger prospective study encompassing patients and institutions from across the United States and other parts of the world could uncover additional nuances in data pertaining to both pathogens and hosts. Third, data on residential ZIP codes of patients (Figure) was collected at time of data extraction and may differ slightly from the residences of patients at time of diagnosis and treatment. Mayo Clinic’s research data platform does not store historical residential data. However, the primary geographic variable used for analysis was site of diagnosis and treatment; ZIP codes were used only to illustrate the large regional representation of patients seeking treatment at Mayo Clinic sites as destination healthcare centers. It is worth noting that 88.2% of patients still resided within the region of diagnosis and treatment at time of data extraction, which further validated our conclusions about geographic distribution and variability of Nocardia spp. Fourth, univariate analysis of 1-year survival did not account for possible confounders in the relationship between transplant status and all-cause death. Fifth, we did not record the number of Nocardia isolates excluded because of lack of clinical or radiographic evidence representing potential Nocardia colonization. Sixth, we collected limited data on coexisting conditions and socioeconomic factors and did not record specifics of immune suppression besides organ transplantation. Finally, we recognize inherent biases associated with retrospective studies, such as selection bias and center effect.

In summary, our study provides information on differences in geographic distribution, patient characteristics, disease manifestations, and antimicrobial susceptibility patterns related to Nocardia spp. in noncontiguous regions of the United States with varied climatic conditions. Similar investigations of patients with invasive infections caused by pathogens of environmental origin encompassing broader geographic regions from around the world would help continue to expand knowledge about how to manage and treat such infections effectively.