Acute Pulmonary Hemorrhage in Infants Associated With Exposure to Stachybotrys atra and Other Fungi

Background  A geographic cluster of 10 cases of pulmonary hemorrhage and hemosiderosis in infants occurred in Cleveland, Ohio, between January 1993 and December 1994.

Study Design  This community-based case-control study tested the hypothesis that the 10 infants with pulmonary hemorrhage and hemosiderosis were more likely to live in homes where Stachybotrys atra was present than were 30 age- and ZIP code–matched control infants. We investigated the infants' home environments using bioaerosol sampling methods, with specific attention to S atra. Air and surface samples were collected from the room where the infant was reported to have spent the most time.

Results  Mean colony counts for all fungi averaged 29227 colony-forming units (CFU)/m³ in homes of patients and 707 CFU/m³ in homes of controls. The mean concentration of S atra in the air was 43 CFU/m³ in homes of patients and 4 CFU/m³ in homes of controls. Viable S atra was detected in filter cassette samples of the air in the homes of 5 of 9 patients and 4 of 27 controls. The matched odds ratio for a change of 10 units in the mean concentration of S atra in the air was 9.83 (95% confidence interval, 1.08-3×10⁶). The mean concentration of S atra on surfaces was 20×10⁶ CFU/g and 0.007×10⁶ CFU/g in homes of patients and controls, respectively.

Conclusion  Infants with pulmonary hemorrhage and hemosiderosis were more likely than controls to live in homes with toxigenic S atra and other fungi in the indoor air.

PULMONARY hemosiderosis is a rare condition characterized by spontaneous pulmonary hemorrhage, often associated with iron-deficiency anemia.¹ The cause is most often unknown, although previous reports have linked a portion of childhood cases with cardiac or vascular malformations, infectious processes, immune vasculitis, trauma, or known milk protein allergies.²

During January 1993 to December 1994, 10 infants with acute pulmonary hemorrhage and hemosiderosis were seen at Rainbow Babies and Childrens Hospital, Cleveland, Ohio.³ The diagnosis was made by demonstrating alveolar hemosiderin-laden macrophages in biopsy specimens or in bronchoalveolar lavage 3 to 6 weeks after the initial hemorrhage.

All of the infants were black, and all but 1 of the infants were male. During the previous 10 years, 3 cases of idiopathic pulmonary hemosiderosis had been diagnosed among infants and children at this hospital.

The purpose of this investigation was to determine the cause of the high incidence of acute pulmonary hemorrhage and hemosiderosis among infants in Cleveland during these 2 years. The affected infants all lived in an area of eastern metropolitan Cleveland within 6 miles of the hospital.

Among the most striking features of this illness were its severity and its tendency to recur after hospital discharge. In 5 infants, acute hemoptysis necessitating readmission to the intensive care unit recurred within 1 day to 6 months of discharge. All infants' hemorrhages were so severe that they required admission to the pediatric intensive care unit. All but one underwent intubation. One infant died. In a previous report from this investigation, infants with pulmonary hemorrhage and hemosiderosis were found to be 16 times more likely than control infants to live in a house with a history of water damage from roof or plumbing leaks or flooding.⁴

The geographic clustering of the patients' homes, the history of recurrent bleeding, and the history of water damage in the homes led us to hypothesize that the infants had been exposed to a toxic substance in their homes. We found no differences between patients and controls with respect to exposures to household pesticides, infant care products, or cocaine. After ruling out all other known causes of pulmonary hemorrhage (including coagulopathies and abuse), the finding of hemolysis on peripheral blood smears led us to consider the possibility that the infants had been exposed to toxins produced by indoor molds. Specifically, our a priori hypothesis was that infants with pulmonary hemorrhage were more likely than controls to live in homes where Stachybotrys atra was present. This fungus is known to grow in water-damaged homes⁵ and to have toxins that produce hemorrhagic disease and hemolysis in animals.⁶

All infants younger than 1 year who had been admitted to the hospital with idiopathic pulmonary hemorrhage and hemosiderosis between January 1993 and December 1994 were included. For each patient, we generated a list of potential controls from all infants born in Cleveland within 2 weeks of the patient and presently living in an area bearing 1 of the 6 ZIP codes in which all of the patients lived. The list of potential controls was generated from Cleveland birth certificates and records of the hospital continuity clinic. Infants' parents were telephoned to invite their participation in the study. For each patient, the first 3 potential controls whose parents agreed to participate were enrolled. Informed consent was obtained from the parents or guardians of all infants.

A pediatrician (R.A.E., E.M., or D.G.D.) visited the homes of all patients and controls to administer a questionnaire comprising more than 200 items that included questions about the infant's health, infant care practices, and home environment characteristics. The questionnaire included specific questions about the infant's exposure to toxic agents (pesticides, paints, solvents, and gasoline) and structural characteristics such as water damage. A registered sanitarian performed an environmental survey of each home, with special attention to the infants' sleeping and living areas. The pediatrician and the sanitarian were not blinded to the case or control status of the infant.

From December 11 to December 19, 1994, at a separate visit, industrial hygienists unaware of case or control status of the homes performed environmental sampling to look specifically for the presence of S atra in the air and on surfaces. Residents were not at home during the sampling.

Bioaerosol sampling was performed at the homes of patients and controls to determine the presence of S atra and other fungi. All air and surface samples collected from each site for viable microorganisms were refrigerated at approximately 4°C before analysis.

The air samples were collected during 1 to 2 hours from the room where the infant was reported to have spent the most time. Various activities, including vacuuming carpets, pounding on furnace ducts several times, and walking on carpets, were performed at each residence in an effort to simulate household activities that could release dusts from ventilation systems and household surfaces. Air samples were collected to test for S atra spores and viable fungi.

Airborne conidia (spores) were collected using total dust sampling on cellulose ester membrane filters. Samples were collected using a Gilian pump (Gilian Instrument Corporation, Wayne, NJ) at a flow rate of 1.0 L/min for 6 to 8 hours. After sampling, each filter was positioned on a glass slide, and the entire area of each filter section was scanned using brightfield microscopy (approximate magnification, ×200) to identify the presence of S atra spores. A standard reference slide of S atra spores was prepared in the same manner to aid in the identification of spores.

Continuous samples for viable fungi were collected using the CAMNEA filter method.⁷ Fungal propagules were collected on polycarbonate filters using a Gilian pump at a flow rate of 2.0 L/min for approximately 1 to 2 hours. By culturing serial dilutions of the filter washings, fungal spores from these samples were enumerated.⁸ Diluted filter fluids were plated on the following media: rose bengal streptomycin agar,⁹ cellulose agar (Czapek-Dox agar with sucrose and FeSO₄ omitted, containing 20 g/m powdered cellulose and 50 mg/L rose bengal and adjusted to pH of 8.0), 2% malt extract agar, and dichloran glycerol agar.¹⁰ The plates were then incubated at 24°C for 10 days. Colonies were classified into the following categories: Aspergillus, Cladosporium, Penicillium, Stachybotrys, and other. The other category included all other fungi observed. Concentrations are reported as colony-forming units per cubic meter of air sampled (CFU/m³).

Samples were collected from areas of suspected mold growth in homes of patients and controls by scraping surface materials into sterile centrifuge tubes or plastic bags. Serial 10-fold dilutions were prepared after adding 0.5-g portions of the sample to 49.5 mL of phosphate-buffered saline containing 0.1% polysorbate 80. Aliquots of these dilutions were plated as described, except that 2% malt agar was not used. The plates were incubated at 24°C for 10 days, the colonies were counted, and results were expressed as CFU per gram.

Mean concentrations of fungi in the air and on surfaces were calculated for the homes of patients and controls by dividing the total number of CFU by the number of plates from each home. LogXact¹¹ was used to calculate the matched odds ratio (OR) for a change of 10 units in the mean concentration of S atra in the homes of the patients compared with the homes of the controls. Mean concentration of S atra is a continuous predictor in the logistic model. It follows that the slope coefficient for S atra gives the change in log OR for an increase of 1 unit in mean concentration. Since this was unlikely to be of interest biologically, we decided to consider a change of 10 units in mean concentration of S atra.¹² To test for interaction with environmental tobacco smoke, a multivariate logistic model was constructed that also controlled for the matching.

Acute pulmonary hemorrhage occurred in infants who were previously in excellent health. Parents or caregivers noted that the infant abruptly stopped crying, became limp and pale, and then coughed up blood, started grunting, and stopped breathing (Table 1).⁴

Patients and controls appeared to be from relatively comparable socioeconomic settings. For example, mothers of 80% of patients and 83% of controls were receiving Medicaid assistance. Mean maternal ages (21.2 vs 24.3 years), mean maternal education (11.4 vs 11.5 years), and use of an air conditioner (25% vs 29%) were also comparable in both groups.⁴

We were able to gain entry to homes of 9 of the 10 patients and 28 of 30 controls. Patients lived in homes that were an average age of 76 years (range, 59-89 years), whereas controls lived in homes that were an average age of 75 years (range, 35-95 years).

Microscopic analyses of dust for airborne S atra spores detected S atra spores in homes of 7 of 9 patients vs 9 of 28 controls.

The relative concentrations of the various categories of fungi in filter cassette samples from homes of patients and controls are shown in Table 2. An unmatched analysis shows that mean CFU counts for all fungi averaged 29227 CFU/m³ in homes of patients vs 707 CFU/m³ in those of controls. The mean concentration of S atra was 43 CFU/m³ in homes of patients and 4 CFU/m³ in homes of controls when averaged across all media. Stachybotrys atra was detected in filter cassette samples from homes of 5 of 9 patients vs 4 of 27 controls. The matched OR for a change of 10 units in the mean concentration of S atra on the filters was 9.83 (exact 95% confidence interval [CI], 1.08-3×10⁶). In other words, if there was a 10-CFU/m³ increase in the concentration of S atra in the air the infant breathed, then the infant was 9.83 times more likely to be a patient (Table 3).

To test for interaction with environmental tobacco smoke, a multivariate matched analysis assessed the impact of S atra concentration and exposure to environmental tobacco smoke and showed an OR of 21 (95% CI, 1.07-7.5×10⁶) for an increase of 10 units in the mean concentration of S atra in the presence of environmental tobacco smoke.

The relative concentrations of the various categories of fungi in surface samples from homes of patients and controls are shown in Table 4. An unmatched analysis shows that the mean concentration of S atra was 20×10⁶ CFU/g and 0.007×10⁶ CFU/g in homes of patients and controls, respectively. Stachybotrys atra was found in excess of 10⁸ CFU/g in surface samples in homes of some patients. Stachybotrys atra was detected in surface samples from homes of 4 of 7 patients vs 10 of 19 controls (Table 5 and Table 6). The matched OR for a change of 1 million units in the mean surface concentration of S atra was 1.35 (exact 95% CI, 0.99-1.5×10¹⁴).

Homes of only 1 of the 9 patients failed to reveal the presence of S atra using at least 1 of these detection methods, in contrast to those of 12 of 28 controls (OR, 4.9 [95% CI, 0.5-255.6]) (P=.26).

The results of our study suggest that infants with acute pulmonary hemorrhage were more likely than controls to live in homes that had molds, including S atra, in the air. The spores of S atra contain potent mycotoxins, and we hypothesize that pulmonary hemorrhage occurred after the infants inhaled these spores.

A unique set of circumstances in Cleveland may explain, in part, why this cluster of cases of pulmonary hemorrhage occurred there. The neighborhood in which most of the patients resided consisted of older homes, some in poor repair. Roof and plumbing leaks and flooding with standing water in many of basements were commonly reported, resulting in conditions suitable for the growth of a variety of fungi, including toxigenic S atra. Many of the forced-air heating systems of these homes were designed so that return air for the furnaces was pulled from the basements. Because of limited resources, the patients' caregivers reported that water-damaged items were not removed from the homes.

The numbers of CFU per cubic meter of air sampled for all categories of fungi studied were consistently higher in homes of patients than those of controls when the air samples were collected using filter cassettes (Table 2), suggesting that conditions in these homes favored exceptional levels of fungal contamination.

In some homes, we were able to culture S atra from the air but did not find it on surfaces. This may be because the fungus was growing in areas not visible to the investigators, such as behind wallpaper or inside walls. In other homes, we cultured S atra from surfaces but did not find it in the air. This may be because the fungus, which is slimy and not easily aerosolized, was not in the air at the time of our sampling. Aerosolization may be an intermittent phenomenon.

Although S atra was found on surfaces in homes of 10 controls, we think that it may be necessary to aerosolize spores to put an infant at risk for pulmonary hemorrhage. The presence of this fungus on surfaces therefore may not be clinically relevant unless it is disturbed or becomes aerosolized.

Stachybotrys atra requires water-saturated, cellulose-based materials for growth in buildings.⁵ Its spores contain a variety of toxins,¹³ including the most potent members of a large family of mycotoxins called trichothecenes.¹⁴ Two specific trichothecenes produced by Stachybotrys, satratoxins G and H, are among the most potent protein synthesis inhibitors known.¹⁵

A study of the toxigenic potential of strains of S atra from the homes of Cleveland patients grown in the laboratory in pure culture on rice demonstrated that these isolates of S atra produced satratoxins G and H and a variety of other trichothecene mycotoxins.¹⁶

Species of the genera Aspergillus and Penicillium were abundant in the homes studied, which suggests the possibility that metabolites of S atra and of other fungi may be present together. Some of these species are also known to produce mycotoxins, eg, Aspergillus ochraceus, Aspergillus versicolor, Penicillium aurantiogriseum, and Penicillium chrysogenum.¹⁷ However, the matched ORs in Table 3 demonstrate that, in our study, there were no differences in concentrations of Aspergillus or Penicillium between patient and control homes.

Young or immature animals are more susceptible to the toxic effects of trichothecenes than adults, and hemorrhage and karyorrhexis are conspicuous in rapidly dividing cells.¹⁸ It is possible that very young infants may be unusually susceptible because their lungs are growing rapidly. Conceptually, local inhibition of protein synthesis during the formation of the endothelial basement membrane is likely to lead to capillary fragility and subsequently to stress hemorrhage. Male animals may be more susceptible to these mycotoxins than female animals.¹⁹

In an earlier report from this investigation, exposure to environmental tobacco smoke appeared to increase the risk for acute pulmonary hemorrhage. Nine (90%) of the 10 patients were exposed to tobacco smoke in the home, whereas 16 (53%) of 30 controls were so exposed. In a matched analysis, exposure to tobacco smoke in the home showed an OR of 7.9 (95% CI, 0.9-70.6).⁴ Although no association between pulmonary hemosiderosis and environmental tobacco smoke exposure has been reported previously, idiopathic hemosiderosis has been linked to active smoking²⁰ in a 15-year-old boy. Secondary stressors such as tobacco smoke or other illnesses may play an important role in triggering overt pulmonary hemorrhage.

A variety of investigators have described the effects of exposure to S atra among adults.^21,22 Bloody nasal discharge has been documented among adults with occupational exposure to S atra. Forgacs and Caril⁶ described "severe pharyngitis, or burning sensation in the nose accompanied by bloody nasal discharge and a moderate to severe cough" in workers in whom illness developed after inhaling dusts from Stachybotrys -contaminated straw.

In animals, exposure to trichothecenes has been associated with hemorrhaging and anemia,²³ but this has not been reported previously in human infants. In mature mice, studies of intranasal administration of S atra spores demonstrated severe alveolar and interstitial inflammation with hemorrhagic exudate in the alveoli.^24,25 Studies of the effects of inhalation exposure of another trichothecene, a biological warfare agent called T-2 toxin, have been made in several animal models. The effects of inhalation were noted to be much greater (>10 times) than those of intravenous exposure.²⁶

Stachybotrys atra is thought to be uncommon in North American homes. A study in California found about 3% of 70 homes to have this fungus.²⁷ A study of 52 homes in eastern Canada found S atra in 1 home.²⁸ A recent Canadian study surveyed 401 single-family homes in Wallaceberg, a largely rural community of 12000 in southern Ontario, during the winter of 1994. Approximately 280 species of molds were recovered from dust samples collected in the living areas of the homes. Stachybotrys was found in 3 homes.²⁹ Thus, in large surveys of residential environments, Stachybotrys has not been listed among the most common fungi found indoors.

There are several limitations of our study. Home sampling for fungi occurred after the infants' hemorrhages, and the conditions at the time of sampling may not have reflected conditions during development of the hemorrhage. However, the fact that none of the patients' parents reported clean-up of water damage suggests the presence of long-standing mold problems. In previous studies, the concentrations of fungi in the air of residences was shown to differ considerably from week to week.³⁰

Since each home was sampled only once, it is possible that we may have misclassified some homes as negative for S atra when in fact they were positive. Air spore counts are known to increase with construction work and vacuuming of carpets.³⁰ It is well known that spores released in 1 part of a home can rapidly spread throughout the home on air currents.³¹ Since the sampling was performed with the environmental hygienists unaware of case status, however, any misclassification would have made an association between the presence of S atra and infant pulmonary hemorrhage less likely. Another limitation is that we cannot rule out association of pulmonary hemorrhage with exposure to other toxigenic fungi that we did not uniformly speciate and quantify.

Further research is needed to determine whether this association is causal. Although the association meets several of the epidemiologic criteria for causation (ie, strength of the association, specificity, biologic plausibility, and coherence), other criteria (ie, temporality and consistency with other studies) have not yet been fulfilled.³² Additional research is needed to determine whether exposure to toxigenic fungi such as S atra is associated with acute pulmonary hemorrhage in infants in other areas. Such work would be aided by the development of methods to detect spores of Stachybotrys or trichothecene metabolites in human tissue.

Accepted for publication March 27, 1998.

The use of trade names or commercial products is for identification only and does not imply endorsement by the US Public Health Service or the US Department of Health and Human Services or recommendation for use.

We thank Vera Anna Hofmeister, PhD; Philip R. Morey, PhD; Michael Infeld, MD; Gregory Wagner, MD; Raymond Biagini, PhD, Barbara A. Bowman, PhD, Paul Smith, DO; Beverly Dahms, MD; Mary O'Connor, MD; Timothy Horgan, MPH; Robert Staib; B. Kim Mortensen, PhD; Roberta H. Hilsdon; and the parents and grandparents of the infants who participated in this investigation.

Editor's Note: The epidemiology story is great reading. Rather than "Eleven Blue Men," we have 10 blue infants.—Catherine D. DeAngelis, MD

Corresponding author: Ruth A. Etzel, MD, PhD, 1400 Independence Ave SW, Room 3718 Franklin Ct, Washington, DC 20250-3700; 202-501-7373 (e-mail: RUTH.ETZEL@USDA.GOV).