Human milk is the recommended nutrition for preterm infants given its reported benefits, such as protection against infectionsReference Furman, Taylor, Minich and Hack1–Reference Schanler, Shulman and Lau3 and necrotizing enterocolitis (NEC),Reference Schanler, Shulman and Lau3–Reference Sisk, Lovelady, Dillard, Gruber and O’shea5 and improved long-term neurocognitive developmentReference Bier, Oliver, Ferguson and Vohr6, Reference O’Connor, Jacobs and Hall7 and cardiovascular health outcomes.Reference Singhal, Cole, Fewtrell and Lucas8, Reference Singhal, Cole and Lucas9 Banked human milk (BHM) is commonly used in countries around the world for premature infants when mother’s own milk is not available or is insufficient.Reference Gartner, Morton and Lawrence10, Reference Arslanoglu, Ziegler and Moro11 However, BHM could be associated with inherent infectious risks, even when pasteurized. These risks are usually thought to be extremely small and to be largely outweighed by the benefits resulting from BHM feeding.Reference Kim and Unger12 One possible exception that has been raised by some is the risk of systemic infection with Bacillus cereus.Reference Decousser, Ramarao and Duport13–Reference Landers and Updegrove15
Bacillus cereus is a motile, gram-positive, aerobic or facultative anaerobic, spore-forming rod of the family Bacillaceae that is widely distributed in air, dust, and water. Its capacity to sporulate allows the bacterium to resist the usual cleaning procedures used in the food industry and in hospitals.Reference Decousser, Ramarao and Duport13, Reference Bottone16–Reference Lebessi, Dellagrammaticas, Antonaki, Foustoukou and Iacovidou18 As a human pathogen, the organism is known for its role as a mediator of self-limited foodborne illnesses and its association with severe local and systemic infections in immunosuppressed hospitalized patients and neonates. In this context, premature infants and very-low-birth-weight neonates in neonatal intensive care units (NICUs) are highly susceptible to infection because of their immature innate and adaptive immune systems and the invasive procedures performed, such as mechanical ventilation, intravascular catheters, gastric feeding tubes, etc.Reference Hilliard, Schelonka and Waites17, Reference Ramarao, Belotti and Deboscker19 Nevertheless, infections due to B. cereus in preterm infants are extremely rare, but when they occur, the mortality rate is high.Reference Ramarao, Belotti and Deboscker19–Reference Van Der Zwet, Parlevliet and Savelkoul21
The pathogenic spectrum of B. cereus ranges from harmless to lethal and highly toxic strains.Reference Wiedermann22 This complex pathogenic potential represents an additional challenge when it comes to identifying the cause of contamination and diagnosing potential infections in immunocompromised neonates. Moreover, because of the ubiquitous distribution of B. cereus in the environment, the source of contamination may be difficult to identify.Reference Hilliard, Schelonka and Waites17 Nevertheless, the ability of B. cereus to resist the pasteurization process has led some researchers to hypothesize that BHM could be a possible source of contamination.Reference Lotte, Herisse and Berrouane20 However, to our knowledge, no case has been documented of B. cereus infection of preterm infants caused by BHM ingestion. Moreover, to address this theoretical risk, milk banks have established safety standards that define the acceptance or rejection criteria prior to distributing pasteurized milk,23 including postpasteurization testing.23–Reference Fischknecht, Wälchli, Annen, Fuhrer, Gianoli and Stocker27
We report 2 cases of B. cereus infection in neonates, suspected to be related to the provincial maternal milk bank involving 2 different hospitals in the province of Quebec. In this study, we aimed to provide some insights on B. cereus mode of transmission and pathogenic potential in vulnerable neonates, with a particular emphasis on the putative role of BHM ingestion as the source of infection and to compare clinical isolates with those found in the involved BHM.
Methods
Established in 2014 by Héma-Québec, the provincial milk bank provides pasteurized human milk to premature infants born at 32 weeks of gestation or earlier admitted to hospitals of the province of Québec, Canada (total population, 8.4 million). The Héma-Québec milk bank collects, pasteurizes, tests, and distributes ~4,000 L of human milk that feeds ~1,000 preterm neonates annually. Production procedures, such as collection, preparation and qualification follows the Human Milk Banking Association of North America (HMBANA) standards.23
In 2014 and 2016, BHM was suspected as a possible source of B. cereus infection in 2 premature neonates admitted in NICUs in 2 hospitals in the province of Quebec. Milk batch recall procedure was launched and banked milk production was stopped during the investigation.
Results
Case 1
In September 2014, an extremely-low-birth-weight (ELBW) preterm infant (gestational age: 25 weeks; weight: 590 g) was born vaginally and admitted to a NICU in the province of Quebec, Canada. Prematurity resulted from premature rupture of membranes. At birth, no signs of infection (clinical or biological) were noted. Nonetheless, prophylactic antibiotic therapy (ampicillin, gentamicin and cloxacillin) was prescribed. On the day of NICU admission and for the entire duration of hospitalization, intensive care support, including umbilical catheterization (arterial and venous), gastric feeding, intubation, and total parenteral nutrition were provided. At 3 days after admission, intensive care included inhaled nitric oxide (iNO) and blood transfusions. Oral trophic feeding with BHM and expressed breast milk using a gastric tube began on the day of admission. The trophic administration scheme was set to 1 mL every 2–4 hours. The BHM came from 2 distinct lots and was stopped after 5 days of administration in favor of exclusive maternal milk gavage. 4 days after birth, clinical deterioration manifested as respiratory distress (hyaline membrane disease and pulmonary hypertension), hyperglycemia, and anemia. At day-of-life (DOL) 4, intraventricular bilateral hemorrhage lesion (grade I) and septic shock were diagnosed. Despite extensive antimicrobial therapy using linezolid, meropenem, and vancomycin, the baby died 10 days after birth. Based on 3 blood cultures positive for B. cereus, the cause of death was identified as sepsis.
Despite the absence of bacteria was found with standard postpasteurization bacteriological testing, the putative role of BHM was raised as a source of sepsis in this case, and microbiological investigation was performed on batches that were used to feed the infant. A nonstandard protocol for postpasteurization bacterial culture, using a large volume of BHM (ie, corresponding to a 25-mL sample), was processed to recover any bacteria in the retention lot. A B. cereus was isolated and identified using 16S rRNA sequencing method in implicated batches. To genetically discriminate the B. cereus strains isolated from the neonate from that of BHM, genetic typing method using pulsed-field gel electrophoresis (PFGE) analysis was performed. The PFGE analysis showed that the 2 B. cereus strains isolated from the infant’s blood cultures were indistinguishable to each other, sharing a unique PFGE profile (profile A), but the B. cereus isolated from the BHM was genetically different, displaying a unique PFGE profile (profile B) (Fig. 1). Therefore, no causal link could be established between the pasteurized BHM and the infant’s infection. Finally, despite additional clinical microbiological analyses in the NICU, investigations failed to identify the source of infection.
Dendogram resulting from cluster analysis of PFGE fingerprints from SmaI digestion including blood cultures (n = 2) from case 1 and BHM Bacillus cereus isolates. Profile A: clinical B. cereus involved in the sepsis (based on 2 blood cultures); profile B: B. cereus isolated from the BHM; profiles C and D: B. cereus used as controls of the protocol.
Case 2
In August 2016, a second case of B. cereus septicemia in an ELBW premature infant (gestational age, 24 weeks; weight, 560 g) fed with BHM, expressed breast milk (EBM) and human milk fortifier from a different NICU in the province was reported. Prematurity was secondary to placenta abruption with chorioamnionitis and preterm rupture of membranes. At birth, no clinical signs of infection or positive cultures were noted, but in the context of a respiratory distress syndrome and suspected maternal chorioamnionitis, antibiotics were administered prophylactically (IV ampicillin and tobramycin). Bag mask ventilation, bubble continuous positive airway pressure, central 3.5-Fr umbilical vein catheters, and 2.5-Fr umbilical artery catheters were used at birth. Trophic feeding with EBM and BHM began the day after admission, and on DOL 6, gastric tube feeding of EBM and BHM was started (human milk fortifier was provided at DOL 17 when feedings reached 160 mL/kg/day). A blood culture was positive for B. cereus at DOL 6. On DOL 7, BHM feeding was stopped, and a lumbar puncture was performed because of suspected meningitis, given the positive blood culture. Meningitis was suspected but never confirmed since cerebrospinal fluid analysis did not support the diagnosis. Clinical status worsened, with evolving chronic lung disease (CLD) and patent ductus arteriosus. A peripherally inserted central catheter was used at DOL 8. After clinical deterioration, the DART (Dexamethasone: A Randomized Trial) protocol (corresponding to dexamethasone dose of 0.15 mg/kg/day) was started for CLD and cholestasis on DOL 16 and stopped 5 days later because of a suspected infection and important respiratory deterioration. The baby was in septic shock and, despite broad-spectrum antimicrobial therapy (including ampicillin, tobramycin, vancomycin, meropenem, piperacillin-tazobactam, fluconazole, amphotericin and trimethoprim-sulfamethoxazole), the baby died on DOL 48 of sepsis diagnosed with 1 B. cereus–positive blood culture and 1 P. aeruginosa–positive blood cultures.
BHM was suspected as the source of the B. cereus infection. Therefore, high-sensitivity microbiological cultures (as done in case report 1) were performed on retained samples from BHM batch suspected of contamination, and the suspected BHM was from a single batch. The culture was positive for Brevibacillus spp using 16S rRNA sequencing method, which is a spore-forming, gram-positive Bacillus bacterial species unrelated to the isolate from the baby. No causal relationship could be established between the pasteurized BHM and the infant’s infection.
Discussion
Since the inception of the milk bank reported herein, 2 cases of B. cereus infection in neonates have pointed to BHM as a possible source of infection. However, after thorough microbiological investigations on retained batches of BHM, and in spite of the presence of Bacillus cereus in a single implicated batch, using PFGE analysis, no molecular epidemiological relationship and causality could be established between the ingestion of contaminated BHM and neonatal infections. Strains found in the BHM and in the infected neonates were different, thereby providing evidence that BHM was unlikely to be the source of contamination in these infants. PFGE has proven to be a reliable tool for the differentiation of species and strains and has thus become the gold standard in epidemiological studies for several pathogenic bacteria such as Bacillus spp.Reference Maslow, Slutsky and Arbeit28, Reference Castiaux, N’Guessan, Swiecicka, Delbrassinne, Dierick and Mahillon29
Cases of infections due to B. cereus have previously been reported in preterm neonates.Reference Bottone16 However, given the small number of published cases (Table 1), the pathophysiology of these infections remains elusive and poorly understood. Furthermore, the assessment of the origin of contamination by B. cereus, an environmentally ubiquitous bacterium, may not lead to an obvious and unique source.Reference Hilliard, Schelonka and Waites17 However, B. cereus infection appears to often arise as outbreaks, suggesting a common source of aerial contamination or direct and indirect contact routes.Reference Birch, Perera and Hyde30–Reference Youngs, Roberts, Kramer and Gilbert33 Furthermore, in the NICU context, premature infants and ELBW neonates are highly susceptible to infections because of their immature innate and adaptive immune systems and their prolonged exposure to invasive procedures, such as mechanical ventilation and the frequent use of intravascular catheters.Reference Hilliard, Schelonka and Waites17, Reference Ramarao, Belotti and Deboscker19 In Amsterdam, Netherlands, an outbreak of systemic infections caused by B. cereus occurred in a NICU in 1998. Three children developed sepsis with positive blood cultures, and one died (28 weeks of gestation and birth weight of 895 g). A thorough investigation revealed that all 3 cases of infection were caused by a single clonal type of B. cereus, and the same strain was found to have colonized the respiratory tract of 35 other neonates in the NICU. Moreover, surveillance revealed that colonization occurred exclusively in the respiratory tracts of mechanically ventilated neonates. This epidemic strain of B. cereus was found on the hands of nursing staff and in balloons used for manual ventilation.Reference Van Der Zwet, Parlevliet and Savelkoul21 A report from the United Kingdom showed that 28 of 57 (49%) neonates in a NICU had been colonized asymptomatically in the umbilicus over a surveillance period of 2 weeks, 90% of whom were carrying a B. cereus serotype identical to the strain found in the diaper supply of the NICU.Reference Hilliard, Schelonka and Waites17 Two other reports from Great Britain also described outbreaks of B. cereus in NICUs. Young et al (1985)Reference Youngs, Roberts, Kramer and Gilbert33 described the dissemination of B. cereus over 4 months. In 259 umbilical cord swabs examined, B. cereus was isolated from 48; 88% belonging to a single serotype. However, the source of contamination could not be established because of the widespread B. cereus dissemination in the NICU environment. Finally, Gray et al (1999)Reference Gray, George, Durbin, Ewer, Hocking and Morgan31 reported an outbreak of B. cereus respiratory tract infections affecting 6 ventilated preterm neonates over a 2-week period in a NICU: 4 infants developed clinical illness and 3 died. Water for humidification of ventilator circuits was identified as the source of the infection in these intubated preterm infants.Reference Gray, George, Durbin, Ewer, Hocking and Morgan31 In 2013, Hosein et alReference Hosein, Hoffman and Ellam34 conducted a routine microbiological screening of premature newborns using umbilical swabs in the summers of 2009 and 2010 at a 400-bed UK general hospital in an attempt to uncover a possible relationship between environmental and clinical isolates of B. cereus.Reference Hosein, Hoffman and Ellam34 They found that freshly laundered linens were contaminated with B. cereus, with possible subsequent spread to and colonization of newborns: 65% of newborn umbilical swabs were contaminated with B. cereus and >85% of freshly received linen was contaminated with B. cereus.
Published Case Reports of Bacillus cereus Infection in Preterm Infants
Note. ND, not determined; M, male; F, female; PMT, prematurity; MV, mechanical ventilation; NEC, necrotizing enterocolitis; IUGR, intra-uterine growth restriction; VP shunt, ventriculoperitoneal shunt; IV, intravenous; AMK, amikacin; AMP, ampicillin; AMX, amoxicillin; BEP, benzylpenicillin; CAZ, ceftazidime; CFZ, cefozopran; CHL, chloramphenicol; CIP, ciprofloxacin; CLI, clindamicyn; CTX, cefotaxime; CYC, cyclamycin; DXM, dexamethasone; ERY, erythromycin; FLQ, fluoroquinolone; GEN, gentamicin; IPM, imipenem; LIZ, linezolid; MER, meropenem; MTZ, metronidazole; PAN, panipenem; PIP, piperacillin; SMZ, sulfamethoxazole; SUL, sulbactam; TBM, tobramycin; TEI, teicoplanin; TMP, trimethoprim; VAN, vancomycin.
a The criterion for establishing that the source of infection has been identified is identic between B. cereus strains found in the environment and in the infant.
b Idem: The suspected source of infection corresponds to the proven source of infection.
In addition, several studies have also assessed the putative implication of human milk in cases of B. cereus infection in premature infants (Table 1); however, clinical strains of B. cereus never matched strains found in banked milk. Therefore, to date, the traceback of B. cereus infections to contaminated BHM hasnever been demonstrated.Reference Decousser, Ramarao and Duport13, Reference Wendelboe, Smelser, Lucero and McDonald35 Similarly to our cases presented herein, between September and December 2016, 6 preterm infants hospitalized in four different Parisian hospitals (France) presented B. cereus infection, and 3 died because of B. cereus septicemia (30 weeks of gestation and birth weight of 750 g; 29 weeks of gestation and birth weight of 1,075 g; and 26 weeks of gestation and birth weight of 750 g). All of those preterm neonates were fed by human milk from the Lactarium d’Île-de-France, a regional milk bank located at the Hôpital Necker-Enfants Malades, Paris, France, that distributes maternal milk for one-third of all preterm infants born in France. The concomitant occurrence of these cases, and the common infection with B. cereus, pointed to the pasteurized human milk supplied by the Île-de-France milk bank as a possible common source of contamination. Moreover, during the outbreak (August 2016), the Île-de-France milk bank had experienced an abnormally high rate of postpasteurization milk batch rejection due to Bacillus spp contamination.Reference Fournier, Faraut-Derouin and Casetta36 Milk-bank production was then stopped, and 500 L of conformed BHM were discarded. The provincial Lactarium, like the other 35 French human milk banks, use the French rules for the collecting, preparing, qualifying, treating, storing, distributing, and dispensing (by medical prescription) human milk.26 Specifically, each batch of BHM is subjected to bacteriological testing before and after pasteurization, and BHM is only delivered once bacterial cultures after pasteurization are negative. After a retrospective microbiological investigation distinct from the standard bacteriological analysis, 2 of the pasteurized milk batches used to feed 2 infants were found to be contaminated with B. cereus. However, the strains found in the infants’ blood were distinct by molecular typing from those that were cultured from the banked milk batches.Reference Fournier, Faraut-Derouin and Casetta36 Thus, no causal association could be made between the banked milk and the neonatal infections. Moreover, post hoc microbiological investigation led to >100 environmental bacteriological samples made by the Parisian hospitals. In addition to BHM, B. cereus colonies were found in diapers, benches, baby bath, linens, and lipid emulsions.Reference Fournier, Faraut-Derouin and Casetta36 In 2013, Decousser et alReference Decousser, Ramarao and Duport13 described 2 cases of severe intestinal tract illness in preterm infants fed with BHM. In their investigation, they isolated B. cereus strains from both the intestine and the BHM. However, clinical strains of B. cereus did not match milk strains; thus, no causal epidemiological connection between banked milk contaminated with B. cereus and preterm sepsis could be established.Reference Decousser, Ramarao and Duport13 Lotte et alReference Lotte, Herisse and Berrouane20 (2017) investigated the source of B. cereus infections after the death of 2 very preterm neonates in the same intensive care unit. Breast milk and freeze-dried breast milk were collected and sent for microbiological analysis. Swabs of bench surfaces where BHM bottles were prepared were positive for B. cereus. Strains isolated from the 2 newborns revealed that patients were infected by 2 different strains; both were also distinct from the environmental strains.Reference Lotte, Herisse and Berrouane20 The possible role of food contamination in the occurrence of cluster cases of necrotizing enterocolitis in a NICU was investigated by Wendelboe et alReference Wendelboe, Smelser, Lucero and McDonald35 (2010). These researchers detected the presence of B. cereus in 2 of 4 lots of human milk fortifier, but they could not establish a cause-and-effect relationship between the presence of Bacillus cereus in the milk fortifier and the outbreak.Reference Wendelboe, Smelser, Lucero and McDonald35
To date, only 50 cases of preterm infections with B. cereus have been reported in the English scientific literature (1 published in French) in 40 years (1977–2018). In none of these 50 cases has related contamination of BHM been demonstrated (Table 1). When retrospective microbiological analyses have been assessed, the data have revealed that the most probable origin of the majority of infections remains the contamination of the NICU environment. Indeed, B. cereus contamination is present and persistent in the hospital environment, and reported sources of infection in NICUs such as respiratory support equipment, umbilical catheterization, gastric feeding tubes, dried formulas, linens, heating, ventilation and air conditioning systems have been documented.Reference Hilliard, Schelonka and Waites17, Reference Birch, Perera and Hyde30, Reference Hosein, Hoffman and Ellam34, Reference Barrie, Hoffman, Wilson and Kramer37–Reference Turabelidze, Gee and Hoffmaster39 However, B. cereus should not be considered only as an environmental contaminant because it can induce a delay in treatment that may compromise the clinical outcome.
The feeding route, frequently investigated in the past, is generally not retained as a source of contamination, especially for pasteurized and bacteriologically controlled BHM. Nevertheless, because of the recent outbreak of B. cereus neonatal infections in the Paris region and the concern about the possible causal link with BHM, French authorities, based on benefit–risk analysis, have adopted more sensitive microbial detection protocols as a precautionary measure.Reference Rigourd, Barnier and Ferroni40 Despite the absence of evidence of an actual causal link between BHM and neonatal sepsis, improved detection and decontamination methods might help to preserve the public’s confidence in this vital biological product for infants whose mothers cannot breastfeed. However, to fully prevent B. cereus infection in NICUs, more emphasis should be placed on the direct patient environment.
In conclusion, to our knowledge, no association has yet been confirmed between banked milk contamination and infections in preterm neonates. Therefore, further research is required before establishing a causal relationship between the ingestion of pasteurized human milk contaminated with B. cereus and the infection of immunocompromised infants. If the risk exists, it appears to be extremely low and to be largely mitigated by postpasteurization microbiological criteria applied to define BHM acceptance for distribution. Given the widespread dissemination of B. cereus, more emphasis should be given to the immediate NICU environment, including invasive care equipment, linens and contaminated heating, ventilation, and air conditioning systems, rather than pasteurized human milk.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2019.110
Acknowledgments
We thank Milagros Gonzales and Zineb Laghdir, research assistants of Dr Quach, for the identification and the description of the Canadian cases.
Financial support
This work was supported by internal funds from Héma-Québec.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
