OBJECTIVES. Trichophyton tonsurans is the single most common cause of pediatric dermatophytoses in North America and is observed with increasing frequency in other countries. This investigation was designed to gain insight into the natural course of T tonsurans infection.
PATIENTS AND METHODS. This 2-year prospective, longitudinal study evaluated all preschool-aged children attending a single child care center. Scalp cultures were collected monthly from each child in attendance, and the presence of disease symptoms recorded at each visit. Dermatophyte genotype was assigned based on the combination of stable sequence variations (2 length variants, 8 single-nucleotide polymorphisms, a 10-base pair insertion, a 14-base pair deletion) present in 2 gene loci.
RESULTS. A total of 3541 scalp cultures were collected from 446 children during 24 months. Twenty-two percent to 51% of scalp cultures per month were positive, contributing 1390 fungal cultures of which 1048 were typeable. Among children with multiple typeable isolates, 51% exclusively carried the same strain, 37% demonstrated a single predominant strain with secondary strains transiently acquired, and 12% harbored a different strain of T tonsurans with each typeable culture. The probability that the same strain persisted in subsequent months was 0.898 and unlikely to have arisen by chance. Rates of symptomatic disease were significantly different between exclusive, predominant, and transient carriers of T tonsurans.
CONCLUSIONS. In contrast to dermatophyte infections in older individuals, where symptomatic disease seems to be a consequence of pathogen acquisition and carriers can be traced to an index case, in this preschool-aged population infection was endemic, and symptomatic disease seemed to represent activation of a single strain that persisted on the scalp.
Although identification of the infectious etiology of dermatophytoses dates back to the early 1800s,1 accumulation of knowledge on the natural course of these diseases has been limited and markedly outpaced by investigations into diseases that are bacterial and viral in origin.1,2 Consequently, there has been no attenuation in infection rates. Dermatophytes remain the single most common cause of human fungal infections worldwide.2 They afflict millions of individuals annually and contribute to hundreds of millions of dollars in associated drug costs (estimates that do not consider the added cost of physicians visits and lost time from work and school).3,4
Among children, tinea capitis is the principal dermatophytosis, with symptomatic infection estimated to occur at a rate of 3% to 8% (ie, in ∼1 of every 20 children).5–9 The anthropophilic Trichophyton tonsurans has displaced nearly all other dermatophyte species as the leading cause of tinea capitis in North America,5,6 and the organism is recovered with increasing frequency from infections in European and East Asian countries.10–12 In addition to its central role in symptomatic disease, T tonsurans can be recovered from hosts that are devoid of symptoms at rates comparable to or in excess of the rates observed for clinical infection.7,13–16 Ostensibly, the undefined factors that permit T tonsurans to avoid clearance and remain on the human host in a subclinical state contribute to the ability of this organism to effectively persist and thrive in the population.
To date, few studies have followed young children with sufficient frequency and duration, and none have used techniques of sufficient discriminatory power to identify whether recurrent positive cultures represent transient acquisition of the most common environmental strain types or persistent colonization with a single strain (ie, a true carrier state). As a result, it remains unclear whether active disease represents acquisition or activation of the pathogen. Our investigation was designed to characterize the natural course of T tonsurans infection by evaluating the degree of genetic relatedness for serial isolates acquired from an individual host and elucidating whether a change in disease phenotype arises as a result of the acquisition or loss of unique strain types.
Subjects and Study Design
This study was conducted as a prospective, longitudinal evaluation of preschool-aged children (1–6 years of age) attending a single urban child care center. Among the factors taken into consideration in selecting the participating study site were the size and demographic constitution of the population, as well as the physical organization of the center. The participating center ranks among the largest in the city, caring for ∼200 children of preschool age. The majority of child care center attendees were black, representing the most suitable population in which to examine the course of infection given the predilection of T tonsurans for racial minorities. Finally, the children attending this child care center were distributed among 10 classrooms determined principally by age. This facilitated the large-scale sampling efforts and introduced the opportunity to evaluate the impact of microenvironments within the child care center on infection patterns.
Study investigators visited the child care center monthly over the course of 24 months. At every visit, fungal scalp samples were obtained from each child who was in attendance (100% participation), and the children were examined for clinical indicators of tinea capitis including alopecia, scaling, crusting, erythema, inflammation, and lymphadenopathy. A finding of scaling and/or alopecia in addition to 1 or more of the other clinical indicators was required to classify the child as symptomatic. Given that this was a noninterventional study, treatment was not provided as part of the investigation; however, the child care staff were informed of each child who was symptomatic, and the children were managed according to the policies in place at the center.
Children were enrolled without discrimination for race or gender under a protocol that was reviewed and approved by the University of Missouri-Kansas City Pediatric institutional review board. The parents or legal guardians of each child were provided with detailed information on the study procedures and offered the opportunity to refuse enrollment of their child in the study for any reason.
A sterile, soft-head toothbrush was massaged over the entire scalp to obtain fungal elements and immediately plated onto solid culture medium (Sab-C; Beckton Dickinson, Cockeysville, MD). Plated cultures were maintained at 25°C for 4 weeks, and those positive for fungal growth were subcultured onto a Sab-C plate to provide a sufficient quantity of material for speciation, DNA isolation, and genetic characterization. Isolation of fungal DNA was accomplished by using the DNeasy Plant Mini Kit (Qiagen Inc, Valencia, CA) as described previously.17
A second toothbrush was applied to the scalp in the same fashion and placed into a sterile test tube containing 5 mL of aqueous culture medium (yeast nitrogen base without amino acids and ammonium sulfate [Beckton Dickinson], keratin [ICN Pharmaceuticals, Aurora, OH], chloramphenicol ([Sigma-Aldrich, St Louis, MO], and cycloheximide [Sigma-Aldrich]). Aqueous cultures were incubated with agitation at 32°C for 5 days. Fungal material isolated from aqueous cultures was used in the event that the corresponding solid cultures were negative.
For the purposes of this investigation, culture-positive indicates samples with fungal growth where macroscopic morphology was consistent with a dermatophyte. Samples that were positive for growth of nondermatophytic fungi were not included among those considered culture-positive. The designation T tonsurans-positive indicates that a sufficient quantity of DNA was available from the culture to fully strain-type the organism.
Previously characterized sequence variations within the nontranscribed spacer of the ribosomal RNA (rRNA) region and the alkaline protease-1 (ALP-1) gene locus were used to assign a genotype to the clinical isolates.17,18 Both gene regions were amplified, and the sequence variations were determined by nested polymerase chain reaction (PCR) and PCR restriction fragment length polymorphism (PCR-RFLP) of the long PCR product. In total, 8 single-nucleotide polymorphisms (SNPs), a 10-base pair (bp) insertion, a 14-bp deletion, and 2 length variants were incorporated into the genotyping scheme. Details on the amplification conditions and enzymes used to determine each sequence variation have been previously published.17,18 For strains in which the composite PCR and PCR-RFLP patterns indicated a previously uncharacterized genotype, the analyses were repeated and subsequently confirmed by sequencing. Sequence analysis was conducted with the DYEnamic ET Dye Terminator Kit for MegaBACE (Amersham Biosciences, Piscataway, NJ), and the sequences were analyzed on a MegaBACE 500 capillary array sequencer.
Prototype isolates bearing each of the sequence variations investigated herein underwent serial passage over 36 months to confirm their stability. All SNPs and length variations were stable over this time frame, suggesting that these loci provide a sufficiently sensitive means for longitudinal genotyping in T tonsurans isolates.
The probability of strain persistence was determined by identifying the rate at which the identical strain type was observed in the same child on subsequent months. A randomization test incorporating 1000 shuffled estimates was used to examine the likelihood that the observed persistence rate arose by chance.19 The randomization approach took all strains in the data set at the rate they actually appeared and assigned them in random order to individual patients and months. This process was repeated 1000 times, providing a baseline rate of persistence that could be expected on chance alone.
A generalized linear mixed model (GLMM)20 was used to examine the relationship between culture results or symptom status and fungal sequence variations. This model was also used to explore the relationship between demographics and culture results or symptom status. In some models, an adjustment for calendar month was incorporated to correct for temporal trends observed in the environment. Subject number was used as a random effect in the GLMM. This produced an analysis that properly accounted for the correlation produced by taking multiple observations on the same subject. Failure to do so would lead to seriously biased results.21
All statistical analyses were performed using R 2.2 with the glmmML and EpiTools libraries (R Foundation for Statistical Computing, Vienna, Austria). All P values were 2-sided, with those <.05 considered to indicate statistical significance.
Subjects and Isolates
A total of 446 children (3.9 ± 1.0 years of age) participated in this investigation over the 24-month study period, with the number of participants ranging from 106 to 174 per month. The median number of visits per child was 7 and ranged from 1 to 24. There was an equal distribution of children by gender (234 girls, 212 boys); however, black subjects represented the majority of children in our population (381 black, 47 white, 12 mixed race, 2 Asian American, 1 Native American, and 3 other).
In total, 3541 scalp cultures were collected over the 24-month period. The proportion of scalp cultures in which fungal growth was morphologically consistent with T tonsurans ranged from 22.4% to 51.3% per month (Fig 1), contributing 1390 fungal samples for strain typing. Of these, 1048 were confirmed to be T tonsurans by molecular analysis. In the remaining isolates there was an insufficient quantity of DNA to fully strain-type the organism. As a result, we were able to confirm T tonsurans-positive infection in 13.7% to 43.8% of all scalp cultures (35%–89% of total positive cultures) that were obtained from children at the child care center (Fig 1).
Among the 1048 typeable clinical isolates, sequence variations present in the rRNA region combined to form 11 distinct patterns, and those in the ALP-1 gene resulted 6 different patterns. When considered together, the composite profiles constituted 24 unique T tonsurans strain types in this population (Fig 2). Consistent with our earlier report,17 we observed 5 length variants in the nontranscribed spacer region; however, 2 sequence variations previously reported for this locus were not observed in any of the present strains. In addition, a new ALP-1 length variant was identified along with the previously reported variations in this gene.18
Genotype frequency was estimated on the basis of the number of times the composite strain type appeared in the population. Frequencies were arbitrarily assigned as high, moderate, or low if they occurred at a rate of >10%, 1% to 10%, and <1%, respectively. As depicted in Fig 2, there were 3 high-frequency genotypes, which comprised 65% of the isolates recovered in this investigation. A total of 9 moderate-frequency and 12 low-frequency strain types were observed in the population of T tonsurans isolates recovered from the child care center, and these comprised 29.4% and 5.6% of the total isolates, respectively.
In 264 children (2605 scalp cultures), at least 1 strain of T tonsurans was typeable. There were 2 or more typeable observations in 173 children (1909 scalp cultures). Of the children with multiple typeable isolates, 89 exclusively carried the same strain or pair of strains over a period of time spanning 2 to 24 months. High-frequency genotypes were observed in 62 of 89 children, moderate-frequency genotypes in 22, and low-frequency genotypes in 4. In 1 child, a dual infection was repeatedly observed and was comprised of a single high- and moderate-frequency strain. In 64 children, 1 predominant strain was observed over the course of the investigation; however, additional strains seemed to be transiently acquired and lost. In only 20 children did each typeable culture reveal a different strain with no predominant strain observed on sequential visits. Figure 3 offers representative examples of the 3 patterns of infection.
When all children with multiple typeable isolates were evaluated collectively, the probability that the identical strain type was observed on a subsequent month (ie, persistence probability) was 0.898 (Table 1). A randomization test incorporating 1000 shuffled estimates predicted an average persistence probability of 0.19, and none of the estimates were as large as 0.898 (range: 0.16–0.22), indicating that a proportion this large almost certainly could not be because of chance. Notably, the duration of time that elapsed between positive typeable cultures seemed to have no bearing on the likelihood of observing strain persistence in any given child (Table 1).
In attempts to distinguish whether strain persistence was uniquely host-specific or subject to modulation by the child care center environment, we evaluated the degree to which changes in classroom assignment within the center influenced a child's strain type. For children remaining in the same classroom on subsequent visits, there was an 8% chance that they would experience a change in strain type. The likelihood that the strain type changed when the child graduated to another classroom between subsequent visits increased to 12% (risk ratio [RR]: 1.41; 95% confidence interval [CI]: 1.13–1.76; P = .002), suggesting that modifications to the child's classroom environment had a small but significant impact on the infecting strain type.
To assess the contribution of external environmental factors (ie, the home environment) on infection and strain persistence, we took a closer look at the sibling sets that attended the child care center. In total, 68 sets of siblings participated in the investigation over the course of 2 years. Among these, there were 17 sibling sets for which each child had multiple typeable isolates that were available for comparison. In 9 of 17 sets, the predominant strain type carried by each child was the same; however, in 5 sets the predominant isolate carried by each sibling was different. Of note, in 4 of 5 pairs where the predominant strain differed, both children were in attendance at the child care center at the same time. In only 1 of these pairs was child care sequential, such that the older sibling advanced to primary school before their younger sibling was enrolled in preschool. In the remaining 3 of 17 sets, the relationship between sibling strain types remained inconclusive, because 1 child in the sibling pair demonstrated a different strain with each typeable isolate.
There were an additional 13 sibling pairs in which 1 child demonstrated multiple typeable isolates during the course of the study, whereas their sibling remained persistently culture-negative. In all but 1 of these 13 sibling sets, the children attended the child care center concurrently.
Strain Persistence and Symptomatic Disease
A notable trend in the rate of symptomatic disease was observed based on the patterns of carriage described above. Among the children exclusively carrying the same strain at every visit (n = 89; 902 visits), a symptomatic disease rate of 18.7% was observed. Symptomatic disease rates were 14.0% in children (n = 64; 744 visits) where a single strain predominated but intermittent acquisition of secondary strains could be observed, and 6.5% in the children (n = 20; 263 visits) where a different strain type was observed with each positive culture. In the children who presented with only a single typeable strain of T tonsurans over the course of the study (n = 91; 696 visits), in those who presented with a positive fungal culture but no typeable strain (n = 67; 466 visits) or in those who were persistently culture negative (n = 115; 467 visits), rates of symptomatic disease were 4.4%, 4.3%, and 4.1%, respectively.
As indicated above, the presence of symptoms was not wholly predicted by the status of the fungal culture. The sensitivity and specificity of a positive culture were 66.1% and 66.5%, respectively. The presence of a typeable T tonsurans isolate demonstrated reduced sensitivity (61.2%) but increased specificity (76.7%) over culture alone.
Symptomatic Disease and Strain Type
Given that the majority of culture-positive children were asymptomatic, we evaluated whether a change in disease phenotype (ie, symptomatic versus asymptomatic) occurred coincidentally with the acquisition of a different strain type. Among children where isolates could be genotyped, the probability that a transition in disease phenotype occurred was comparable whether the same strain persisted (35%) or a new strain was acquired (31%). Consistent with this finding, a random effects GLMM revealed no relationship between sequence variations in the rRNA locus and disease phenotype. Although there seemed to be an association between the number of repeat elements in the 5′-untranslated region (UTR) of ALP-1 and disease phenotype, the association did not achieve statistical significance (P = .08).
Temporal Infection Patterns
There were no systematic changes in the rate of symptomatic disease over the course of this investigation, with 10% of children, on average, presenting with findings that were consistent with active infection (Fig 1). In contrast, a negative trend was observed in the proportion of positive cultures (r2 = 0.41; slope = −0.727; P = .001) and recoverable T tonsurans isolates (r2 = 0.71; slope = −1.061; P < .001) during the course of the study (Fig 1).
Demographics and Disease
A clear association between gender and race was observed for culture findings and disease phenotype. Boys were more likely to be culture positive (RR: 1.71; 95% CI 1.21–2.41) and more likely to be symptomatic than were girls (RR: 1.67; 95% CI: 1.11–2.51). This finding may be because of generally shorter hairstyles facilitating both scalp sampling and the identification of symptomatic infection. White children and children of other ethnic backgrounds were far less likely to be symptomatic (white vs black: RR: 0.13; 95% CI: 0.05–0.31; other vs black: RR: 0.21; 95% CI: 0.09–0.50) or culture-positive (white vs black: RR: 0.07; 95% CI: 0.02–0.22; other vs black: RR: 0.08; 95% CI: 0.01–0.48) than were black children. Although the predilection for racial disparity with this infection is well recognized, the reasons remain unclear.
In contrast to gender and race, age did not seem to influence rates of symptomatic disease, because children were equally likely to demonstrate signs of infection along the continuum of ages evaluated in this preschool population. Preliminary analysis alluded to the possibility of an inverse relationship between age and the presence of a positive culture; however, correction for the decline in culture-positive rates observed over the course of the study revealed no effect of age on the likelihood of presenting with a positive fungal culture.
Among school-aged children, symptomatic T tonsurans infection occurs with an estimated incidence of 3% to 8%.7–9 Rates among preschoolers are less well defined, although infections likely occur with comparable frequency given a reported median age of 4 years for the disease.8 Prevalence estimates for the rate of fungal isolation from individuals that are asymptomatic are more widely variable and highly dependent on the nature of the environment from which the sample population arises. Within a typical pediatric population, ∼5% of individuals are described as carriers, whereas the numbers are upward of 30% when sampling from a population of individuals who are in close contact with an index case (eg, family members, high contact sports team mates, caregivers in long-term health care facilities).10,22–26
Given the nature of a preschool environment, where the structure permits appreciably more contact among children than does a primary or secondary school, it was not surprising that the rate of fungal recovery in our population was reflective of a high-contact environment. On average, more than one-quarter of the children in our child care center were harboring a fungal isolate on their scalp, with T tonsurans conservatively confirmed in at least 14% and as many as 44% of the total population in any given month. Of note was the reduction in the overall the proportion of recoverable isolates during the course of this investigation. Although it is conceivable that this wane reflected early peak infection rates that regressed to the population mean during the course of the study, the finding is likely accounted for by specific changes that occurred at the child care center unrelated to our study.
Early during the course of this investigation, pillows, stuffed animals, and upholstered items were removed from the child care center (in a staggered fashion) in attempts to decrease asthma risk in the population. In addition, the cleaning strategy at the child care center was modified late during the course of this investigation because of a Shigella outbreak. Although we are unable to confirm that the aforementioned interventions are directly responsible for the reduction in infection rates observed, both changes could contribute to a reduction in fomite contamination and overall spore load at the child care center. We are also unable to address whether the sustained presence of investigators over a 2-year period contributed to an increase in the recognition and subsequent treatment of tinea capitis, thus influencing the observed culture-positive rates. However, one would expect that an increase in treatment incidence would concurrently reduce the rate of symptomatic disease, which was not the case in this investigation.
The prevalence of positive T tonsurans cultures observed in the current investigation reflects a striking fungal burden in this environment and raises the question of whether comparable centers across the United States, caring for high-risk populations, exhibit similar statistics. Affirmation across a spectrum of young children would portend a fungal population with significant evolutionary potential. This putative capacity for adaptation may explain the large degree of genotype diversity we observed in the population of T tonsurans isolated from our child care center. Whether selective pressure is the principal driver for the heterogeneity we observed remains to be determined. The degree of variation may simply reflect the anthropogenic influence of human migration on the dispersal of unique strain types beyond their natural limits.
Irrespective of the origin of the genetic diversity that was observed, the degree of diversity allowed us to characterize the nature of persistently positive cultures. Taken together, the data seem to indicate that persistent infection is less likely a result of intermittent fungal acquisition and more likely the reflection of a true carrier state. Foremost, this assertion is supported by the extremely high rate with which we observed the same strain on sequential positive cultures. Were transient acquisition to play a more prominent role, infection patterns would seem more random than were observed. In addition, we might expect to find a smaller fraction of sibling pairs carrying different isolates despite sharing both a home and school environment. Moreover, we would not expect to have observed a proportionally large number of children who never acquired a pathogen over the course of the study. Each of these elements lends support to the assertion that a unique relationship exists between T tonsurans and its pediatric host, which seems to be influenced only to a minor extent by the home and school environment.
Intriguingly, the data reveal that the natural course of disease is not identical for all carriers. In our population, carriers of the same strain exclusively and/or predominantly demonstrated rates of symptomatic infection that were twofold to threefold higher than carriers of random isolates, and in excess of fourfold higher than children with a single isolate or no evidence of infection. Although the reason for this finding remains unclear, it is possible that unique, as of yet undefined, host factors enhance the susceptibility of the pediatric carrier to clinical disease. Alternatively, the strains observed on our carriers may be no more likely to cause infection but more likely to be sustained on the host in a carrier state, which in and of itself would predispose clinical infection.
Our observations are not entirely consistent with the current model of infection determined from older children and adults, wherein high carrier rates seem to be linked with an index case. Rather, this preschool-aged population seems to exhibit a stable endemic pattern of infection. Consequently, these data support the need for a critical examination of the current treatment strategies used in the management of T tonsurans infections in our youngest patients. Reevaluation of existing treatment approaches needs to occur in the context of questions which explore whether treating isolated cases of symptomatic infection in a population with an extraordinary fungal burden is satisfactory, whether current therapies have a long-term influence on the carrier state, and whether existing treatments contribute to the endemic nature of dermatophytoses by way of selecting for strain types with increased fitness. Investigations designed to address these issues may reveal that broadly applied eradication strategies offer the greatest likelihood of benefit in the management of this disease.
This work was supported by National Institutes of Health grants 1 R21 AR053234-01 and 5 U10 HD031313-13.
We acknowledge the assistance of Georgann Meredith, RN, Jami Penny, LPN, and Michael Venneman, RN, in the collection of fungal isolates and Natasha Ahmed and Matthew Maloney in the preparation of samples for strain typing. We also thank Drs M. J. Blake, J. Bozue, J. S. Leeder, and A. A. Mitchell for critical review of this manuscript. Finally, we extend sincere appreciation to the children and staff at the participating child care center.
- Accepted August 29, 2006.
- Address correspondence to Susan Abdel-Rahman, PharmD, Division of Clinical Pharmacology and Medical Toxicology, Children's Mercy Hospitals and Clinics, 2401 Gillham Rd, Suite 0411, Kansas City, MO 64108. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
This work was presented, in part, at the 106th General Meeting of the American Society for Microbiology; May 21–25, 2006; Orlando, FL.
- ↵Zakon SJ, Benedek T. David Gruby and the centenary of medical mycology, 1841–1941. Bull Hist Med.1944;16 :155– 168
- ↵Birren B. Fungal Genome Initiative: A White Paper for Fungal Genomics. July 10, 2004. Available at: www.broad.mit.edu/annotation/fungi/fgi/July2004FGIWP.pdf. Accessed September 1, 2004
- ↵Williams JV, Honig PJ, McGinley KJ, Leyden JJ. Semiquantitative study of tinea capitis and the asymptomatic carrier state in inner city school children. Pediatrics.1995;96 :265– 267
- ↵Gaedigk A, Gaedigk R, Abdel-Rahman SM. Genetic heterogeneity in the ribosomal RNA gene locus of Trichophyton tonsurans. J Clin Microbiol.2003;41 :5478– 5487
- ↵Good PI. Resampling Methods: A Practical Guide to Data Analysis. Boston, MA: Birkhauser; 1999
- ↵Faraway JJ. Extending the Linear Model With R Generalized Linear, Mixed Effects and Nonparametric Regression Models. Boca Raton, FL: Chapman and Hall/CRC Taylor and Francis Group; 2005
- ↵McCulloch CE, Searle SR. Generalized, Linear and Mixed Models. New York, NY: John Wiley and Sons, Inc; 2001
- Copyright © 2006 by the American Academy of Pediatrics