Coccidioidomycosis: Experience From a Children's Hospital in an Area of Endemicity

Abstract

The incidence of coccidioidomycosis is increasing, and published pediatric experience is limited. This study further characterizes pediatric coccidioidomycosis by describing experience with 64 patients cared for at Children's Hospital of Los Angeles over 20 years, with focus on sites of involvement and severity, as well as serologic profiles of affected children.

Coccidioidomycosis is caused by Coccidioides immitis/posadasii and is an endemic mycosis of the Southwestern United States and certain arid areas of Mexico and Central and South America [1, 2]. Although most infections are asymptomatic or self-limiting, clinical progression may include severe pulmonary compromise with or without dissemination to extrathoracic sites such as bone, joints, skin, soft tissue, or central nervous system (CNS), while dissemination can occur even in the absence of significant respiratory disease [2]. Recent epidemiologic data suggest an increasing incidence and distribution of this infection, and 10% of older adults are diagnosed in nonendemic locales and have no history of residence in an endemic area [3–5]. Coccidioidomycosis in adults has been well studied; recently, growing interest in coccidioidal infection in children has resulted in clinical reports of pertinent data [6–8]. In 2002, over 40% of hospitalized children with coccidioidomycosis were inpatients in a nonendemic area, underscoring the importance of expanding reported experience in pediatrics to inform physicians caring for children [6].

This study was undertaken to provide further characterization of the sites of involvement, predisposing factors, clinical course, and progression in children with coccidioidomycosis.

METHODS

Approval for this study was obtained from the Institutional Review Board of Children's Hospital of Los Angeles (CHLA). The study was performed at CHLA, a children's hospital that serves as a tertiary referral center for Los Angeles County and the Southwestern United States.

For this retrospective descriptive case series, medical records were reviewed for all patients who were seen in inpatient and outpatient settings between January 1993 and January 2013 at CHLA and who carried ICD-9-CM diagnoses 114.0 through 114.9 under the category coccidioidomycosis. Patients thus identified were excluded if, on further review, the records lacked 1 of the following supporting diagnostics: positive coccidioidal serology (at least two thirds run at University of California, Davis, at the laboratory of Dr. Pappagiannis), cultures positive for C. immitis/posadasii, or biopsy with characteristic coccidioidal spherules on histological examination.

Data collected included demographics, clinical course, radiologic findings, complement fixation titers, length of stay, antifungals used, response to antifungal therapy, and outcome when available. Data were de-identified and stored in a secure database prior to analysis. Review of follow-up visits extended to September 2013. Of note, 19 of these patients were partially described in a larger-scale, national study using data from the Pediatric Information Health System [7].

Statistical analysis included odds ratio for categorical variables and Wilcoxon (Mann–Whitney) test for continuous variables, using SPSS software (version 15.0.1, 2006). Differences with P values <.05 were considered significant.

RESULTS

Sixty-four children, 37 with pulmonary disease and 27 with disseminated disease, were identified. All children with disseminated disease and 21 children with pulmonary disease were hospitalized. Of patients with disseminated disease, 10 (37%) had CNS involvement presenting either with headache or seizure, or previous disseminated disease diagnosis. In 5 children, the CNS was the only identified site of dissemination. Mediastinal mass was noted in 1 child with disseminated disease and 1 child with only pulmonary disease. Paratracheal, hilar, or mediastinal lymphadenopathy was a more common finding (5 patients with disseminated disease and 6 patients with pulmonary disease). Cavitary lesions were found in 1 patient with disseminated disease and 9 patients with pulmonary disease.

There was a slight male predominance, and children with extrapulmonary disease tended to be younger, but these differences were not statistically significant (Table 1). Most of our patients came from Southern California, with 20 (31%) originating from high desert areas in Kern, San Bernardino, and Los Angeles counties, but 2 patients were from Phoenix, Arizona, 1 from Las Vegas, Nevada, and 1 from Sonora, Mexico.

Table 1.

Patient Demographics and Disease Distribution

		Disease Distribution
	All Patients	Pulmonary Only	Extrapulmonary	CNS
Demographic	No. (%)	No. (%)	No. (%)	No. (%)
Total	64 (100)	37 (58)	27 (42)	10 (16)
Male sex	38 (59)	22 (59)	16 (59)	5 (50)
Age at diagnosis, median (range), years	10 (0–18)	12 (0.5–17)	9 (0–18)	9.5 (0–18)
Ethnicity/race
Hispanic	22 (34)	13 (35)	9 (33)	4 (40)
Non-Hispanic white	13 (20)	9 (24)	4 (15)	1 (10)
African American	11 (17)	2 (5)	9 (33)	4 (40)
Other	13 (20)	8 (22)	5 (19)	1 (10)
Unknown	5 (8)	5 (14)

Percentages are based on distribution category.

Abbreviation: CNS, central nervous system.

African American children were more heavily represented in the extrapulmonary disease category when compared to overall diversity data at our institution (33% children compared to the 6% typically seen) [9]. Five children listed as having “unknown” ethnicity/race were not included in ethnicity/race-related statistical analysis. African Americans were more likely to have extrapulmonary disease compared to all other children (OR 7.9, CI 1.55–40.1, P = .01) and compared to non-Hispanic children (OR 11, CI 1.98–61.3, P = .006).

Hispanics were more likely to be admitted compared to other non–African American children (OR 6.1, CI 1.19–31.4, P = .03). They were more likely to have a hospital stay > 5 days compared to all other children (OR 5.3, CI 1.22–23.2, P = .02) and even more so when compared to non–African American children (OR 11.7, CI 2.31–59.5, P = .003). African American children were also more likely to have prolonged hospitalizations when compared to non-Hispanic children (OR 15.4, CI 1.47–161, P = .02). Ethnicity/race was not associated with significant increases in readmission, nor in length of treatment required before a decrease in complement-fixing antibody titers.

Patients with disseminated disease tended to have higher complement fixation titers at presentation (Figure 1) and specifically were more likely to have titers of 1:32 or greater at presentation (OR 7.65, CI 2.09-28, P = .002; 77% vs. 31% in pulmonary patients). Patients with titers of 1:128 or greater had longer initial hospitalizations (median 15 days, [range, 3–90 days] vs. median 8 days, [range 1–44 days for patients with titers <1:128], P = .03). Higher initial complement fixation titer was not associated with greater likelihood of readmission.

Figure 1.

Box plot depicting comparison between initial complement fixation (CF) titers in patients with pulmonary disease only and in patients with disseminated disease. Five patients with pulmonary disease had negative initial titers, while 3 patients with disseminated disease had negative initial titers.

The overall percentage of previously healthy patients was 59% and similar for all categories of disease distribution. Among patients with at least 1 admission for coccidioidomycosis, 62% of pulmonary patients and 63% with disseminated disease were previously healthy. The 11 immunocompromised patients included 9 patients on immunosuppressant medications, 1 human immunodeficiency virus (HIV)-infected patient, and 1 asplenic patient, constituting 17% of all patients and 23% of hospitalized patients. Of the patients on immunosuppressants, 3 had undergone solid organ transplant, 4 had autoimmune disease (2 had received tumor necrosis factor inhibitors), and 1 was receiving treatment for leukemia. Immunocompromised status was not associated with greater propensity for disseminated disease; however, intra-abdominal and cardiac involvement (including total cardiac involvement found on autopsy and a previously published report of atrial mass) were more frequently identified in these patients (Table 2) [10]. There was no significant difference in duration of initial hospitalization or in number of coccidioidomycosis-related admissions between known immunocompromised and nonimmunocompromised patients. Immunologic defects, including HIV infection, were not identified in any previously healthy patients. Two patients with recalcitrant disease tested negative for interferon-γ receptor gene defects.

Table 2.

Organ Systems Involved in Patients With Disseminated Coccidioidomycosis^a

	Total	Pulmonaryb	Musculoskeletal	CNSc	Skin	Intraabdominal	Cardiac
Patient Characteristics	No.	No. (%)	No. (%)	No. (%)	No. (%)	No. (%)	No. (%)
All	27	21 (78)	12 (44)	10 (37)	7 (26)	3 (11)	2 (7)
Immunocompromised	5	5 (100)	3 (60)	1 (20)	1 (20)	2 (40)	1 (20)
African American	9	8 (89)	4 (44)	4 (44)	4 (44)	0 (0)	1 (11)
Hispanic	9	7 (78)	5 (56)	4 (44)	0 (0)	2 (22)	1 (11)
Other race/ ethnicity (including non-Hispanic white)	9	6 (67)	3 (33)	2 (22)	3 (33)	1 (11)	0 (0)

Abbreviation: CNS, central nervous system.

^aNine patients had involvement of 2 or more systems apart from pulmonary.

^bOne patient with disseminated disease had a cavitary lesion, compared with 9 of the patients who had only pulmonary involvement.

^cThree of 10 CNS patients had hydrocephalus.

More than half of patients with disseminated disease underwent surgical interventions, most commonly irrigation and debridement in cases of musculoskeletal involvement or ventriculoperitoneal shunt placement with CNS involvement. Most patients with complement fixation titers 1:128 and greater (9/11) required surgical in addition to medical therapy, but when compared to patients with lower titers this was not statistically significant.

Antifungal therapy with fluconazole was the initial therapy for most patients. Other azoles administered were itraconazole, voriconazole, and posaconazole. Of 27 patients with disseminated disease, 6 improved solely with fluconazole, compared to 19 of 29 treated pulmonary patients; 6 received another azole without addition of amphotericin B, and the remainder received an amphotericin B product with or without an azole at some point in their treatment. Overall, 56% of patients with disseminated coccidioidomycosis received therapy with a polyene, compared with 42% of hospitalized pulmonary patients. Six of 16 patients seen in the ambulatory setting received no antifungal treatment.

Intrathecal amphotericin B was used successfully in 1 patient, a previously healthy female who presented at age 11 years with coccidioidal meningitis with severe basilar disease. She received a total of 37 mg of amphotericin B intracisternally along with oral itraconazole prior to developing arachnoiditis. She has been well controlled on itraconazole alone thereafter and has neither neurologic sequelae nor recrudescence observed with last evaluation at age 24.

One patient, a previously healthy 15-year-old African American girl with a strong family history of autoimmune disease, presented with widely disseminated coccidioidomycosis after about 1 month of fever with weight loss. Within several days, laboratory studies became concerning for hemophagocytic lymphohistiocytosis, and the patient began immunosuppressive therapy. Initial evaluation for infectious triggers or underlying rheumatologic disease was negative, but further investigation was limited by critical illness, and the patient died 17 days after admission. Two patients died of other causes.

DISCUSSION

While coccidioidomycosis is usually a self-limited disease, progression to life-threatening disease may occur. This study was undertaken to add to published experience of coccidioidal infection in children. Records of both inpatients and outpatients were examined, but, as CHLA is a referral center, our patient sample is likely biased toward more severe presentations. Less severe pulmonary cases are evaluated and/or treated by community practitioners as opposed to being referred and would not have been represented in this group of patients. However, the percentage of previously healthy patients seen here was similar to the 66% reported by Fisher et al in their study [7], while the percentage of immunocompromised patients hospitalized for coccidioidomycosis (23% in our study) varied greatly in other reports, from 14% in Chu et al to 41% reported by Connelly and Zerella [6, 11]. Limitations of this study included a small sample size, which we attempted to augment by extending the time period examined. Also, the design of the study, which was retrospective and observational, relied on other clinicians' record keeping as opposed to personal interviews or examinations. Nevertheless, our findings are concordant with prior reports in several respects, while contributing additional insights.

Among our patients, African Americans showed greater propensity for disseminated disease and, accordingly, longer hospitalizations, consistent with previous reports [12–14]. Hispanics were more likely to be admitted and to have prolonged hospitalizations, but it could not be determined whether this derived in part from socioeconomic factors rather than severity of disease.

One third of patients with disseminated disease had CNS involvement, underscoring the importance of early consideration and evaluation for presence of CNS disease, especially in the setting of previous coccidioidomycosis diagnosis.

Complement fixation titers of 1:32 and greater were associated with dissemination, as has been reported previously, suggesting a need for investigation for disseminated disease in these patients [1]. In 1 previous study, titers of 1:128 or greater were associated with failure of medical therapy requiring surgical intervention in patients with skeletal disease [15]. Similarly, in our study, children with titers of 1:128 or greater had longer hospital stays, and, of the patients with disseminated disease, titers of 1:128 or greater tended to be associated with need for surgery. Titers of 1:128 or greater may signal particularly severe disease, supporting the importance of aggressive, early antifungal therapy in such patients if further data corroborate these findings. It must be mentioned that, while the majority of our serologic results were obtained from a single laboratory, interlaboratory variability is a factor that may have affected our data.

Our study contributes to a limited but growing body of literature on pediatric coccidioidomycosis by further characterizing the spectrum of system involvement, severity, and serologic profile in affected children. It is clear that additional, carefully designed and prospective clinical research in the pediatric age group is warranted.