Respiratory cryptosporidiosis in Malawian children with diarrheal disease

Introduction

Cryptosporidiosis is a cause of diarrhea [1-3], excess mortality [4, 5], stunting [1, 6], and is associated with malnutrition [6, 7]. The Global Enteric Multicenter Study (GEMS) identified cryptosporidium as second most common cause of diarrhea among infants (0-11 months) in all four African countries studied (The Gambia, Mali, Mozambique, Kenya) regardless of HIV prevalence, and among the top five causes for older children (12-23 months) [4]. Respiratory cryptosporidiosis has been documented in up to a third of children presenting with diarrhea [8]; furthermore, respiratory detection without intestinal involvement has been reported [9, 10], raising the possibility of primary respiratory infection with cryptosporidium, either by inhalation or by contact with fomites [11].

Studies have identified cryptosporidium in sputum but have not looked at whether respiratory involvement is a transient phenomenon or a reservoir for gastrointestinal (GI) disease. This has implications for therapeutic development. Our primary objective was to evaluate whether respiratory involvement of cryptosporidium is a transient phenomenon in diarrheal disease, and to assess for respiratory and GI cryptosporidiosis concurrently and longitudinally in children hospitalized with diarrheal disease.

Methods

Results

From 1 March 2019 to 3 April 2020, 162 children were recruited into the study. Two children did not submit specimens. Of the remaining 160 children, 36 (23%) were positive in any one of the three samples collected (34 stool (21%), 11 sputum (7%), and 4 NP (3%)) and were entered into follow-up. The study was discontinued early due to COVID-19, by which time 27 children had completed at least three of the four follow-up visits (75%).

Among the 162 children who were hospitalized with diarrhea, the mean age was 11.6±5.0 months, and 59% were male (Table 1). Mean height-for-age, weight-for-age, and weight-for-height z scores were -1.41±1.5, -1.02±1.2, and -0.08±1.6, respectively. Nine percent of mothers were HIV-infected, and one-third of those children tested positive by HIV rapid test (3% of total). The participants had a mean of 4.5±1.4 persons in a household, and the majority of children had access to a pit toilet/latrine (95%), piped drinking water (78%), and harbored a residential animal in the compound (66%). Among the 36 children who were cryptosporidium PCR-positive, the mean age was 12±5.0 months, 63% were male and two (6%) were positive by HIV rapid test. No significant differences in characteristics between children who were cryptosporidium-positive versus -negative were noted (Supplementary Table 1).

Induced sputum quality at enrollment was good in 81%, poor in 9% and unreadable in 10%, with similar proportions of good quality sputum among cryptosporidium-positive and - negative participants (81% and 82%, respectively). Over the 8-week period, cryptosporidium was detected in both respiratory and GI tracts (Table 2): most commonly in stool (94% at enrollment to 44% at 8 weeks), followed by sputum (31% at enrollment to 20% at 8 weeks), then NP (11% to 8%). Lowest Ct counts were generally noted in the stool (mean Ct 28.8±4.3 at enrollment to 29.9±4.1 at 8 weeks), followed by sputum (mean Ct 31.1± 4.4 at enrollment to 35.7±2.6 at 8 weeks), and NP ((mean Ct 33.5±1.0 to 36.6±0.7); Figure 1). We detected cryptosporidium in respiratory but not GI tract in 7 study visits, including 2 at enrollment (stool Ct were above the 35 cutoff and therefore considered negative). Detection in the respiratory tract did not always correlate with symptoms (Figures 2A-O). For individual participants in whom cryptosporidium was detected at enrollment, detection was noted most consistently in the GI tract throughout the 8 weeks. Respiratory co-pathogens were detected in 87/104 (84%) visits over the follow-up period.

• Download figure
• Open in new tab

Figure 1.

Comparison of cryptosporidium detection in GI and respiratory tracts (mean cycle threshold with 95% confidence intervals) over 8 weeks in children hospitalized with diarrheal disease

• Download figure
• Open in new tab

Figure 2.

Patient-level dynamics of cryptosporidium detection in GI and respiratory tracts over 8 weeks in a: A) 10 month-old; B) 14 month-old; C) 9 month-old; D) 12 month-old; E) 14 month-old; F) 7 month-old; G) 7 month-old; H) 11 month-old; I) 14 month-old; J) 11 month-old; K) 17 month-old; L) 10 month-old; M) 13 month-old; N) 20 month-old; and O) 15 month-old.

Respiratory symptoms were noted in 72% of cryptosporidium-positive participants at enrollment, the most common symptom being cough (69%), but also included runny nose (46%), crackles (17%) and wheeze (11%). For those entered into the 8-week follow-up period, participants with cryptosporidium positivity in the respiratory tract had respiratory symptoms in 23/43 (53%) of visits.

Among participants with cryptosporidium detection in both the respiratory and GI tract at enrollment (Table 3A) compared to those with cryptosporidium detection in the GI tract only, a larger proportion reported respiratory symptoms (90% v. 65%), and GI shedding of cryptosporidium was longer (14.3±2.1 v. 14.1±0.7 days), but these were not statistically significant. Among participants whom cryptosporidium was detected in both the respiratory and GI tract over the 8-week study period compared to GI tract only (Table 3B), a larger proportion reported respiratory (81% v. 68%) and GI symptoms (62% v. 27%) per study visit, and GI shedding of cryptosporidium was longer (17.5±6.6 v. 15.9±2.9 days), but again these were not statistically significant.

Discussion

This is the first longitudinal study to evaluate the respiratory cryptosporidiosis in pediatric diarrheal disease. In children hospitalized with diarrheal disease, cryptosporidium was detected most frequently in stool, followed by sputum and NP. The most common respiratory symptom was cough. Longitudinally, we detected cryptosporidium in both respiratory and GI tracts through 8 weeks post-enrollment, usually at lowest Ct counts in stool, followed by sputum and then NP. Longer GI shedding of cryptosporidium was noted among those where we detected cryptosporidium in both the respiratory and GI tracts, compared to GI detection only, although findings did not reach statistical significance.

Our cryptosporidium detection rate is far higher than has been reported in diarrhea studies in sub-Saharan Africa, with rates of positivity ranging from 9% in Kenya to 14.7% in Mozambique [4, 8]. Subclinical cryptosporidium infection was noted to be high (6%) in Tanzania [15]. A previous study done at QECH found cryptosporidium to be the third most common cause of diarrhea with a prevalence in stool of 28% among cases [16]. In our study, the positivity rate was 21%, and this would be 12% if we used the GEMS diarrheagenic cutoffs [17]. This thus supports the high prevalence of cryptosporidium infection among young children hospitalized with diarrheal illness in sub-Saharan Africa.

Few studies, however, have evaluated the prevalence of respiratory cryptosporidiosis in pediatric diarrheal disease. A study in Uganda detected cryptosporidium in 35.4% of induced sputum samples from children presenting with diarrhea who were cryptosporidium-positive in the stool [8]. In our study, cough was noted in almost two-thirds of children presenting with diarrhea, and the frequency of respiratory signs and symptoms tended to be higher, but were not significantly different between those with cryptosporidium-positive and -negative stool samples. In addition to sampling the lower respiratory tract with induced sputum, we also collected NP specimens throughout the follow-up period, and documented detection throughout the 8-week period, although in fewer numbers, more intermittently, and at higher Ct counts compared to sputum. Respiratory symptoms were more common among those cryptosporidium-positive in respiratory and GI tract compared to positive in GI tract alone, although again these findings were not statistically significant. In contrast to the Mor study, which only collected sputum samples from children with cough, unexplained tachypnea or hypoxia, we collected respiratory specimens in all participants at enrollment and noted that participants with cryptosporidium positivity in the respiratory tract had respiratory symptoms in just over half of visits. These data support that almost half the participants with respiratory cryptosporidium detection are asymptomatic, and is a lower, rather than upper, respiratory tract pathogen or colonizer [18-21].

The continued detection of cryptosporidium in both GI and respiratory tract over 8 weeks may reflect the young age of this population. A study in Bangladesh noted that children ≤2 years positive for cryptosporidium shed it for a mean of 4.1 weeks, which was significantly longer than those >2 years (mean 1.7 weeks) [22]. However, that was a household transmission study and did not evaluate children hospitalized with diarrhea. For respiratory pathogens, prolonged detection of respiratory viruses has been noted in a community surveillance study in Utah, with significantly longer detection noted in children <5 years (mean 1.9 weeks) compared to other age groups (mean 1.6-1.7 weeks) [23]. To our knowledge, prolonged detection in the respiratory tract has not before been described for a protozoan parasite, which was not associated with higher prevalence of respiratory symptoms.

Prevalence of cryptosporidiosis is higher in cases of persistent v. acute diarrhea (15% v. 6.1% in a study in Guinea Bissau [5]). Persistent diarrhea has been described for cryptosporidiosis with duration of up to 5 months among severely immunocompromised infants [24]. In the Etiology, Risk Factors and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health and Development Program (MAL-ED) study, cryptosporidiosis was the fifth highest attributable pathogen in all pediatric diarrhea in the community setting, and had increased frequency among those with prolonged and severe diarrhea [3]. Among our participants, GI symptoms resolved within 2 weeks of hospitalization for all participants, but cryptosporidium was detected in stool for eight participants throughout the 8-week period. Shedding duration for cryptosporidiosis among children hospitalized with diarrhea in sub-Saharan Africa has not been described previously using molecular techniques, although in Malawian HIV-infected adults we documented consistent shedding for up to 8 weeks [25]. Nutritional status may also account for the prolonged detection of up to 8 weeks that we documented in our study population. Even though the mean weight-for-height z score was within normal, the mean height-for-age and weight-for-age z scores of recruited participants both met WHO criteria for stunting and wasting, and 41%, 22%, and 10% had height-for-age, weight-for-age, and weight-for-height z scores <-2, respectively, indicating mild-moderate malnutrition in this population. Reduced nutritional status could impact host immunity and predisposition to colonization/infection [26-28].

Cryptosporidiosis has been associated with excess mortality in children who had the infection in infancy, and this excess mortality persists into the second year of life [4, 5]. In our study we had one death, which occurred in a 7 month-old breastfeeding, HIV unknown male. This child had anthropometric parameters consistent with severe acute malnutrition (mid-upper arm circumference 11 cm, height-for-age z score -1.85, weight-for-age z score -3.01, weight-for-height z score -2.47, no nutritional edema) at time of enrollment, and an enrollment cryptosporidium Ct of 23 in the stool but negative in sputum and NP. At time of enrollment he was afebrile with normal vitals and had mild dehydration, diarrhea associated with abdominal tenderness and vomiting, as well as runny nose, cough and wheeze, but no shortness of breath. Death occurred two weeks after hospitalization after the child developed a cough and fever and was treated at an outpatient health facility.

This study has several limitations. The sample size was small, and the study population was drawn from a single hospital, and therefore findings may not be generalizable to other settings. Secondly, we did not evaluate for respiratory co-pathogens in our enrollment specimens, thereby limiting our ability ascribe symptoms to a single pathogen, although we did document co-pathogen detection in 84% during the follow-up period. Thirdly, we did not genotype cryptosporidium to ascertain whether re-infections occurred. However, these limitations are balanced out by the longitudinal observational design, which allowed us to document for the first time the dynamics of respiratory cryptosporidium among children hospitalized with diarrheal disease.

Conclusion

In summary, in this study we demonstrated that over 20% of young children hospitalized with diarrheal disease in our setting are positive for cryptosporidium, and they can shed cryptosporidium in the stool for up to 2 months. Concurrent respiratory cryptosporidiosis can be detected in a substantial proportion of young children, and detection continues over a prolonged period, with detection more prominent in the lower rather than upper respiratory tract. This has implications for the development of therapeutics for cryptosporidiosis, which is limited currently to nitazoxanide, and which is poorly efficacious in malnourished children [29, 30]. Since malnourished children are in greatest need for cryptosporidium therapeutics, scientists have been actively trying to find new, more effective drugs to treat cryptosporidiosis in this population. Because safety is of paramount concern in this age group, those developing drugs have debated whether a drug that only resides in the GI tract would be adequate for treating cryptosporidiosis in malnourished children, compared to one with broader systemic distribution [31]. Our findings suggest that the respiratory tract is a significant reservoir for cryptosporidium infection, and that to cure children, may require a drug that distributes to the lungs as well as the GI tract. Thus, future research on therapeutic development should focus on drugs that target not only the GI but also the respiratory tract.

Funding

The work was supported by the Bill & Melinda Gates Foundation (OPP1191165).

Author contributions

Conceived and designed the study: PI; Performed the study: MC, WN, HT, JN; Analyzed the data: MC, JN; Supervised the study and wrote first draft of the manuscript: PI; Reviewed, provided critical feedback, and approved the final draft: all authors.

Potential conflicts of interest

PI has received grants from BMGF outside of the submitted work. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.