Discussion
The incidence of neutropenic enterocolitis in the current study was 7.4%. Fever and pain abdomen were the most common symptoms. However, diarrhoea, haematochezia and mucositis were observed at a higher frequency than in other studies (Table 1).
All the patients had neutropenia, severity and duration of which correlated with various outcome parameters. Significant hypoproteinaemia and hypoalbuminemia observed in 60% and 78% of children respectively can be explained by the loss of serum proteins and albumin through the inflamed gastrointestinal mucosa. Similar findings were noted by User et al14 and Moran et al30, where hypoalbuminemia was observed in 50% and 46% respectively.
No definite risk factors for neutropenic enterocolitis could be identified in the current study. Clostridiodes difficileinfection, found as a risk factor by El-Matary et al10, was not isolated in any patient. Anthracyclines, etoposide and cytarabine based chemotherapy were administered in half of children with neutropenic enterocolitis; however, this association was not found to be statistically significant. The degree of gut dysbiosis was higher in children with NEC, with significantly higher isolation rates of Bacteroides species, nonsignificant higher isolation ofClostridium species and lower growth of Lactobacillusspecies. Though this doesn’t establish the causality of gut dysbiosis, underlying dysbiosis could predispose a susceptible individual to neutropenic enterocolitis, and a ‘healthy flora’ may be protective for development of mucositis and neutropenic enterocolitis.
On etiological evaluation, blood culture positivity rate was 17%, majority being contributed by Klebsiella pneumoniae. Fungal cultures and CMV PCR were non-contributory. Other series have reported culture positivity rates of 8-25%; majority of isolates being gram negative bacilli.8,31,32,33 Stenotrophomonas maltophila , isolated in one patient, has been rarely implicated in the causation of neutropenic enterocolitis34.
The current study found only the presence of free fluid to be an independent adverse prognostic factor predicting mortality. However, factors like female gender, younger age, duration and severity of neutropenia and bowel wall dilatation and thickening were associated with other parameters like longer duration of hospital stay and bowel rest. Notably, isolation of Bacteroides and lack ofLactobacillus were associated with longer requirement of bowel rest, thus suggesting a possible role of gut dysbiosis in pathogenesis of neutropenic enterocolitis. Cartoni et al35 found higher mortality in patients with bowel wall thickening > 10 mm (60% vrs 4.3%). McCarville et al8 also found a similar association with bowel wall thickening and duration of neutropenia. User et al14 found that the presence of hypoalbuminemia, hypokalemia and metabolic acidosis were associated with poor outcome, which was not found in the current study.
Faecal calprotectin was elevated in patients with NEC as compared to patients without NEC and healthy controls. Stringer et al found a similar elevation of faecal calprotectin in patients with chemotherapy related diarrhoea, though the degree of elevation was significantly less.28 Wedlake et al found a significant rise in faecal calprotectin after pelvic radiotherapy.17However, van Vliet et al found faecal calprotectin to be below detectable range in a majority of patients with chemotherapy related mucositis and diarrhoea.18
Though prognostic value of fecal calprotectin couldn’t be demonstrated, it appears to have diagnostic value. Using a cut off value of 75 µg/g, faecal calprotectin as a diagnostic tool yielded a sensitivity of 65% and specificity of 63% with AUC of 0.74.
Various studies have uniformly noted a decrease in alpha diversity as measured by Shannon Index as a marker of gut dysbiosis. They also reported decrease in isolation of Lactobacillus andBifidobacterium . However, there are conflicting reports onBacteroides and other Firmicutes as markers for dysbiosis of faecal microbiota. van Vliet et al, Huang et al and Stringer et al report a decrease in Bacteroides species as a marker of gut dysbiosis, whereas Rajagopala et al36, Montassier et al24 and Zwielehner et al25 reported increase in Bacteroides and decrease in Firmicutes . This difference in patterns of gut dysbiosis may be explained by racial, geographical and dietary differences and difference in age. In the current study, increase in Bacteroides and reduction inLactobacilli were noted in children with NEC. (Table 6)
The relative abundance of each flora did not have significant impact on mortality, this might have been contributed by the small sample size in the non survivors. However, isolation of Bacteroides and absence of Lactobacilli did have an impact on requirement of prolonged bowel rest; isolation of Bacteroides increased the requirement of bowel rest by 2.2 days and presence of Lactobacilli tended to decrease the requirement of intravenous alimentation by 2.4 days. This observation supports the hypothesis and mechanism of role of gut microbiota as proposed by van Vliet et al23. This gives basis for further studies on gut microbiota in the field of pediatric oncology and role of interventions like prebiotics and probiotics to target gut dysbiosis.
Conflict of interests: None