DISCUSSION
This study reports on a series of 38 cases of EG in children with hematological disease and cancer. EG is reported mainly in immunocompromised and cancer patients, nevertheless, even in patients with malignancies, this complication is described only as case reports or small case series. An extensive review described a total of 162 cases of EG in immunocompromised and immunocompetent patients of any age, among whom there are several case reports and case series, the largest being of 17 patients1. Our study represents the most extensive series of EG in the hematology and oncology setting, not only pediatric. Based on the wider number of patients in our study, we aimed at outlining the characteristics of presentation of EG in children with malignancies, in order to point out the clinical profile of patients at risk, to early recognize the EG and to treat it properly.
As expected, children in our population were affected mostly (92.1%) by hematological malignancies, since 34/38 patients had acute leukemia and 1 patient severe aplastic anemia (SAA), confirming the role of the underlying disease as risk factor for the infectious disease. In fact, the treatment of acute leukemia is associated with deep alteration of innate and adoptive immunity. Almost all patients in our series had severe neutropenia (97.3%) and the majority was on corticosteroid treatment (71%) at the moment of infection, while 9/38 (23.7%) children had iatrogenic diabetes.
Neutropenia has been reported as the most frequent risk factor for EG in different reports1,3,4, while corticosteroid treatment is known to play a role abrogating host defenses against bacterias18-19. In addition, hyperglicemia influences the risk of infection in hematologic malignancies20.
All these risk factors are known to have detrimental effect on the host immune response due to suppression of phagocyte-mediated killing, humoral immune and T-lymphocyte responses21. The skin and gut dysbiosis, the mucositis and epithelial damage, consequence of the disease treatment, represent an open window for the access and spreading of pathogens4,6. Moreover the endothelial damage likely due to chemotherapy22 might be a facilitating risk factor for the development of EG, which starts typically with microvessel lesion and necrotizing vasculitis caused by bacteria, leading to perivascular invasion and ischemic necrosis of the surrounding tissue. EG is a septic vasculitis in which microorganisms may be demonstrated within the vessel wall and may occlude the vessels lumina23. The microscopic lesion justify the clinical picture of a cutaneous erithematous papula rapidly evolving into the typical destruent and large necrotic lesion (supplemental figure S1).
EG has been described as localized lesion or as bacteremic form often with septicemic dissemination, involving different sites in the body5. In accordance to data reported in the literature1, the majority of patients (57.9%) in our series presented the EG associated with bacteremia, while about in one third (34.2%) of cases the EG presented as single cutaneous lesion without bacteremia.
According to most case reports, the primary involved site was the perineal region (50% of primary site), and this did not differ among patients with or without bacteremia. However we observed also less typical sites of EG such as central venous catheter (CVC) exit, the site of bone marrow puncture at the posterior iliac crest, the periombelical area, the scapula, providing that any area of the body might be involved. As a result, the occurrence of EG in hidden skin sites hampers the early diagnosis. We therefore suggest an accurate inspection of the skin in patients at risk of EG, in order to recognize the initial lesion as early as possible (supplemental figures S1, S2, S4).
Apparently no prevalent age-distribution could be documented in our series, since almost any age is represented, from very young children to older ones. However, a more accurate analysis of the age of presentation suggests some considerations (figure 1): first we confirmed the susceptibility of older children to infectious complications in the pediatric hematological setting as previously reported24-27, in fact near half of children were older than 10 years. In addition, our data demonstrated that over 50% of patients were around the age of 5 or younger and, in particular, 42.1% of patients were in the age 0-3 years and five patients were infants. Moreover in the age group 0-3 years over 80% were females. Thus in our sample younger female children might be at higher risk for this peculiar infection. The occurrence of EG in very young children and newborns is also described in some case reports11-13, 28.
As generally reported for infectious complications in hematological malignancies29, more than 40% of EG were documented in the induction phase of treatment. This is detrimental for the outcome of the malignancy, considering the subsequent delay in the delivery of chemotherapy due to such a devastating skin lesion.
Confirming the epidemiology traditionally reported, P. aeruginosawas identified in the majority of patients (89.5%). S. maltophilia , and E. Coli, also described as etiologic agent of EG1, were cultured in the remaining cases. In 10 patients a co-infection was documented, which influenced negatively the outcome in 1 case.
The distribution of resistance to antibiotics of the gram negative bacteria identified, revealed 31.6% of prevalence of MDR bacteria, comparable to the prevalence of 30% reported in a previous AIEOP survey on P. aeuginosa bacteremias in the same setting of patients30. These data are also corroborated by the recent report of the European Antimicrobial Resistance Surveillance Network (EARS-Net)31, in particular as regard to piperacillina-tazobactam, the antibiotic to whom 1/3 of isolates were resistant in our series.
The subset of 12 patients infected by MDR bacteria, as expected, showed characteristics of more aggressive clinical presentation and outcome, with a significantly different distribution of bacteremia.
The start of targeted antibiotic treatment was prompt in the majority of patients. This is an important key in the treatment of EG32, in keeping with the recommendations of international guidelines33 to overcome the rapid growth of the lesion. The necrosis of the tissue in the outcome of EG is a problem of utmost importance, since delays the recovery from the infection and might challenge the quality of life of this setting of patients, even after healing of the lesion, due to the occurrence of deep scarring (supplemental figures S2, S3, S5, S6).
In order to remove the infected necrotic tissue and to ameliorate and the healing of the wound, in our experience a surgical debridement has been considered in most cases, especially for larger lesions with extensive necrosis and more aggressive disease, confirming a described approach34,35.
The vacuum therapy, reported in some cases to promote faster healing of the wound bed36, has been also applied in 5 patients, and in 3 cases prevented the occurrence of deep scars. The treatment with hyperbaric chamber has been also attempted in 4 cases, with successful result in 2.
In those patients with extensive tissue loss, different techniques for wound healing, such as negative-pressure therapy, hyperbaric chamber and dermal substitutes implant, might have a role.
The mortality rate for EG reported in the literature ranges from 8-15% in non-bacteremic infections to 20-77% in bacteremic EG5-6. In our experience we documented lower fatality rate, probably due to the prompt start of targeted therapy in almost all patients. However in some cases the delay in the delivery of proper anticancer treatment probably hampered the control of the underlying disease.