Patients in the mucosal barrier injury–laboratory confirmed bloodstream infection (MBI-LCBI) plus other bloodstream infection (BSI) cohort are represented in both curves.
A, Occurrence of infection of interest at 30 days. B, Occurrence of infection of interest at 60 days. C, Occurrence of infection of interest at 100 days. Mucosal barrier injury–laboratory confirmed bloodstream infection (MBI-LCBI) cohort includes those with at least 1 MBI-LCBI, bloodstream infection (BSI)-other cohorts include those with at least 1 BSI that is not classified as an MBI-LCBI, MBI-LCBI and BSI-other group includes those with at least 1 MBI-LCBI and BSI-other, and the control group includes those who underwent allogeneic transplant and did not have any BSI documented in the first 100 days after transplant.
eTable 1. Variables Examined in the Cox Proportional Hazards Models for Overall Survival, Transplant Related Mortality, Chronic GVHD, and Risk Factor Analysis for Development of MBI-LCBI
eTable 2. Organisms Identified as Blood Stream Infections in the MBI-LCBI Only, BSI-Other Only, and the MBI-LCBI+BSI-Other Categories
eTable 3. Outcomes of Patients Included in the Analysis
eFigure 1. CONSORT Diagram
eFigure 2. Infection Density Examines the Number of Infections per Days at Risk During the First 100 Days
Customize your JAMA Network experience by selecting one or more topics from the list below.
Dandoy CE, Kim S, Chen M, et al. Incidence, Risk Factors, and Outcomes of Patients Who Develop Mucosal Barrier Injury–Laboratory Confirmed Bloodstream Infections in the First 100 Days After Allogeneic Hematopoietic Stem Cell Transplant. JAMA Netw Open. 2020;3(1):e1918668. doi:10.1001/jamanetworkopen.2019.18668
在一项针对 2009 年至 2016 年间接受异基因造血干细胞移植的 16,875 例儿科和成人患者的病例研究中，粘膜屏障损伤-实验室确认血液感染的累积发病率到第 100 天为 13％，感染的中位数为干细胞移植后 8 天。在发展为粘膜屏障损伤-实验室确认血液感染的患者中，总体生存率显著降低。
Patients undergoing hematopoietic stem cell transplant (HSCT) are at risk for bloodstream infection (BSI) secondary to translocation of bacteria through the injured mucosa, termed mucosal barrier injury–laboratory confirmed bloodstream infection (MBI-LCBI), in addition to BSI secondary to indwelling catheters and infection at other sites (BSI-other).
To determine the incidence, timing, risk factors, and outcomes of patients who develop MBI-LCBI in the first 100 days after HSCT.
Design, Setting, and Participants
A case-cohort retrospective analysis was performed using data from the Center for International Blood and Marrow Transplant Research database on 16 875 consecutive pediatric and adult patients receiving a first allogeneic HSCT from January 1, 2009, to December 31, 2016. Patients were classified into 4 categories: MBI-LCBI (1481 [8.8%]), MBI-LCBI and BSI-other (698 [4.1%]), BSI-other only (2928 [17.4%]), and controls with no BSI (11 768 [69.7%]). Statistical analysis was performed from April 5 to July 17, 2018.
Main Outcomes and Measures
Demographic characteristics and outcomes, including overall survival, chronic graft-vs-host disease, and transplant-related mortality (only for patients with malignant disease), were compared among groups.
Of the 16 875 patients in the study (9737 [57.7%] male; median [range] age, 47 [0.04-82] years) 13 686 (81.1%) underwent HSCT for a malignant neoplasm, and 3189 (18.9%) underwent HSCT for a nonmalignant condition. The cumulative incidence of MBI-LCBI was 13% (99% CI, 12%-13%) by day 100, and the cumulative incidence of BSI-other was 21% (99% CI, 21%-22%) by day 100. Median (range) time from transplant to first MBI-LCBI was 8 (<1 to 98) days vs 29 (<1 to 100) days for BSI-other. Multivariable analysis revealed an increased risk of MBI-LCBI with poor Karnofsky/Lansky performance status (hazard ratio [HR], 1.21 [99% CI, 1.04-1.41]), cord blood grafts (HR, 2.89 [99% CI, 1.97-4.24]), myeloablative conditioning (HR, 1.46 [99% CI, 1.19-1.78]), and posttransplant cyclophosphamide graft-vs-host disease prophylaxis (HR, 1.85 [99% CI, 1.38-2.48]). One-year mortality was significantly higher for patients with MBI-LCBI (HR, 1.81 [99% CI, 1.56-2.12]), BSI-other (HR, 1.81 [99% CI, 1.60-2.06]), and MBI-LCBI plus BSI-other (HR, 2.65 [99% CI, 2.17-3.24]) compared with controls. Infection was more commonly reported as a cause of death for patients with MBI-LCBI (139 of 740 [18.8%]), BSI (251 of 1537 [16.3%]), and MBI-LCBI plus BSI (94 of 435 [21.6%]) than for controls (566 of 4740 [11.9%]).
Conclusions and Relevance
In this cohort study, MBI-LCBI, in addition to any BSIs, were associated with significant morbidity and mortality after HSCT. Further investigation into risk reduction should be a clinical and scientific priority in this patient population.
Hematopoietic stem cell transplant (HSCT) is an effective treatment strategy for many malignant neoplasms, marrow failure syndromes, and immune deficiencies in children, adolescents, and adults.1-5 Each year, more than 50 000 HSCTs are performed worldwide. Transplant strategies and supportive care have evolved, resulting in improved overall survival (OS)6; however, patients who have undergone HSCT remain at high risk for bloodstream infections (BSIs) and associated morbidity and mortality.5,7,8
Studies have identified immunocompromised patients, including those who have undergone HSCT, who are at risk of developing BSIs once classified as central line–associated BSIs (CLABSIs) that do not result from contamination of the central venous catheter but instead from other mechanisms such as translocation of bacteria through nonintact mucosa.9,10 The Centers for Disease Control and Prevention developed a modification of the CLABSI definition, termed mucosal barrier injury–laboratory confirmed bloodstream infection (MBI-LCBI) through literature review and expert opinion.11,12 This definition was integrated into National Healthcare Safety Network methods for primary BSI surveillance to classify a subset of BSIs reported as CLABSI that are associated with mucosal barrier injury and not the presence of a central venous catheter.9 Unlike CLABSI,13-15 MBI-LCBIs are not prevented by improved central venous catheter maintenance care.9,12,16
A BSI is defined as an MBI-LCBI if it resulted from 1 or more of a group of selected organisms known to be commensals of the oral cavity or gastrointestinal tract and it occurred in a patient with specific signs or symptoms compatible with the presence of mucosal barrier injury, such as gastrointestinal graft-vs-host disease (GVHD) and/or neutropenia.9,11,12 To our knowledge, there are few data describing the incidence, risk factors, or outcomes of patients who develop an MBI-LCBI after HSCT. This study aims to determine the incidence, timing, risk factors, and outcomes of patients who develop MBI-LCBI in the first 100 days after HSCT.
We analyzed data from the Center for International Blood and Marrow Transplant Research (CIBMTR) registry to compare the outcomes of patients with BSIs. The CIBMTR comprises a voluntary working group of more than 400 transplant centers worldwide that contribute detailed data on allogeneic and autologous HSCTs. The details regarding the CIBMTR and its data collection method are in the eAppendix in the Supplement. This process occurred under the guidance of the CIBMTR via the National Marrow Donor Program Institutional Review Board. Patients provided written informed consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.17
The study population consisted of all pediatric and adult patients undergoing first allogeneic HSCT reported to the CIBMTR between January 1, 2009, and December 31, 2016, including both malignant and nonmalignant conditions (eFigure 1 in the Supplement). The study included patients receiving umbilical cord blood, bone marrow, or peripheral blood stem cell grafts. To limit center bias, patients were included only from centers in which at least 1 patient with MBI-LCBI was identified, and either 1 control patient or 1 patient with BSI secondary to indwelling catheters and infection at other sites (BSI-other) was present. In addition, we excluded patients from centers reporting no GVHD prophylaxis in more than 15% of patients because this may indicate that other data are incomplete.
Centers report infections to the CIBMTR using an organism code, a site code, and the date of the infection. There are no data provided to assess infection prophylaxis, treatment, diagnostic criteria used by the center, or infection severity. Centers are instructed to report clinically significant infections with both online and in-person education regarding appropriate reporting.18 Patients were classified into 1 of 4 groups based on BSIs during the first 100 days after HSCT. The first group was the MBI-LCBI cohort, comprising patients who developed at least 1 MBI-LCBI in the first 100 days after transplant (and no BSI-other). Patients in the MBI-LCBI cohort were classified as such if the infection met the following criteria: the organism was a commensal of the oral cavity or gastrointestinal tract, and infection occurred 14 days before or 60 days after stage 3 or 4 gastrointestinal acute GVHD diagnosis or an absolute neutrophil count of more than 500 cells/μL (to convert to 109 cells per liter, multiply by 0.001) was never achieved after HSCT or the infection occurred before or within 3 days of an absolute neutrophil count of 500 cells/μL or less at any time in the first 100 days after HSCT. The second group was the BSI-other group, comprising patients who developed at least 1 fungal or bacterial BSI by 100 days after transplant that did not meet criteria for MBI-LCBI. The third group was the MBI-LCBI and BSI group, comprising patients who developed at least 1 MBI-LCBI and at least 1 BSI-other in the first 100 days after transplant. The fourth group was the control group, comprising recipients of allogeneic HSCT who did not develop a bacterial or fungal BSI documented in the first 100 days.
We compared OS in the first year after HSCT between patients in each cohort. The cumulative incidences of MBI-LCBI and BSI-other, with death as the competing risk, were assessed in the first 100 days. We calculated infection density, determined as the number of infections per patient per 100 days, for MBI-LCBI and BSI separately. We computed the frequency of infection as a primary or secondary cause of death within the first year after HSCT as reported by the center. The cumulative incidence function (using relapse or progression as a competing risk) was used to estimate transplant-related mortality (TRM), defined as the time to death without evidence of disease relapse.19-21 Thus, only patients with malignant disease have a TRM estimate. Furthermore, for patients with malignant disease, we evaluated disease relapse using the cumulative incidence function with death in remission as the competing event.
The clinical data of patients were described, including demographic characteristics, disease and therapy characteristics, transplant complications, and outcomes. The following variables were evaluated: sex, age at transplant, diagnosis, donor relationship, HLA match, source of stem cell graft, conditioning intensity,22 and neutrophil engraftment. Currently accepted clinical criteria were used for the diagnosis of acute GVHD,23 transplant-associated thrombotic microangiopathy,24,25 and engraftment syndrome in recipients of allogeneic HSCT.26
Statistical analysis was performed from April 5 to July 17, 2018. Because MBI-LCBI is a time-dependent variable, we used a dynamic landmark study with 3 landmark time points at 30, 60, and 100 days to graphically show the probability of 1-year OS.27
Multivariable Cox proportional hazards regression analysis with an examination of the proportional hazards assumption was used to evaluate potential risk factors for MBI-LCBI and for survival. For the Cox proportional hazards regression model for survival, infections and acute GVHD were used as time-dependent variables. If the proportional hazards assumption was violated, the variable was added as a time-dependent covariate. A stepwise selection procedure with a significance level of P < .10 was used to identify the final model. Pairwise interactions and center effects were tested.28 If center effects were significant, we adjusted them in the final model. Hazard ratios (HRs) and their 99% CIs, using the Wald confidence limit in the final model, were reported. All P values were from 2-sided tests and results were deemed statistically significant at P = .01.
For the assessment of risk factors for the development of an MBI-LCBI, only the subset of patients with malignant disease was analyzed. The variables examined are shown in eTable 1 in the Supplement.
From 2009 to 2016, 22 393 pediatric and adult patients undergoing allogeneic HSCT were reported to the CIBMTR. eFigure 1 in the Supplement depicts the exclusions resulting in the final population of 16 875 patients. For the risk factor analysis for the development of MBI-LCBI, only the subset of 13 686 patients with malignant disease (1.1%) were examined owing to different clinical characteristics and preceding therapies.
Of the 16 875 patients (9737 [57.7%] male; median [range] age, 47 [0.04-82] years), 1481 (8.8%) had at least 1 MBI-LCBI, 2928 (17.4%) developed at least 1 BSI-other, 698 (4.1%) developed both an MBI-LCBI and BSI-other, 3189 (18.9%) underwent HSCT for a nonmalignant condition, and 11 768 (69.7%) did not develop a bacterial or fungal BSI in the first 100 days (control group). The demographic and transplant characteristics of the 4 cohorts of patients are shown in Table 1.
The cumulative incidence of MBI-LCBI was 13% (99% CI, 12%-13%) by day 100, whereas the probability of BSI not meeting MBI-LCBI criteria was 21% (99% CI, 21%-22%) by day 100. The median (range) time from transplant to first MBI-LCBI was 8 (<1 to 98) days, MBI-LCBI plus BSI-other was 8 (<1 to 97) days, and BSI-other was 29 (<1 to 100) days. Most cases of MBI-LCBI occurred in the first 2 weeks after HSCT, whereas the incidence of BSI-other continued to increase throughout the first 100 days after HSCT (Figure 1). Most cases of MBI-LCBI met the definition secondary to neutropenia alone (1915 of 2179 [87.9%]), with the remaining 12.1% (264 of 2179) meeting criteria owing to the presence of gastrointestinal GVHD (166 of 2179 [7.6%]) or gastrointestinal GVHD with neutropenia (98 of 2179 [4.5%]). Reported organisms and infection density, accounting for multiple infections, are shown in eTable 2 and eFigure 2 in the Supplement.
Table 2 shows the risk factors associated with MBI-LCBI. Multivariable analysis revealed an increased risk of MBI-LCBI in those with a lower Karnofsky/Lansky performance status (score <90) (HR, 1.21 [99% CI, 1.04-1.41]). In addition, myeloablative conditioning (HR, 1.46 [99% CI, 1.19-1.78]), posttransplant cyclophosphamide as GVHD prophylaxis (HR, 1.85 [99% CI, 1.38-2.48]), and receipt of cord blood (HR, 2.89 [99% CI, 1.97-4.24]) were associated with a significant increase in the risk of MBI-LCBI. Preceding GVHD was not examined because it is incorporated in the definition of MBI-LCBI. The results are adjusted for center effects.
Overall mortality was higher for patients with MBI-LCBI only (HR, 1.81 [99% CI, 1.56-2.12]), BSI only (HR, 1.81 [99% CI, 1.60-2.06]), and MBI-LCBI plus BSI-other (HR, 2.65 [99% CI, 2.17-3.24]) compared with controls (Table 3). A center effect was noted, and the results were adjusted. Figure 2 depicts the OS curves as a series of dynamic landmark analyses examining the outcome of infection by day 30, day 60, and day 100. For patients alive at day 100, the 1-year survival was inferior for patients with MBI-LCBI (n = 1146 [75.1%]; 99% CI, 71.6%-78.3%), BSI only (n = 2473 [70.8%]; 99% CI, 68.3%-73.1%), or MBI-LCBI plus BSI-other (n = 482 [66.8%]; 99% CI, 61.1%-72.2%) compared with controls (n = 10 668 [79.3%]; 99% CI, 78.2%-80.3%; P < .001). Additional factors associated with survival are shown in Table 3.
One-year TRM (nonrelapse mortality) among patients with malignant disease increased for patients with any BSI. The increased risk was similar for patients with MBI-LCBI (HR, 2.34 [99% CI, 1.95-2.80]) or BSI-other (HR, 2.12 [99% CI, 1.78-2.52]) but further worsened for patients with MBI-LCBI plus BSI-other (HR, 3.93 [99% CI, 3.10-4.97]) compared with controls. There was no association of any BSI with the development of chronic GVHD. Additional factors associated with TRM and chronic GVHD are listed in eTable 3 in the Supplement.
Infection was reported as the primary cause of death more often for patients with MBI-LCBI (139 of 740 [18.8%]), BSI only (251 of 1537 [16.3%]), and MBI-LCBI plus BSI (94 of 435 [21.6%]) than for controls (566 of 4740 [11.9%]) (P < .001). In addition, infection as an associated secondary cause of death was higher in patients with MBI-LCBI (158 of 740 [21.4%]), BSI only (343 of 1537 [22.3%]), and MBI-LCBI plus BSI (116 of 435 [26.7%]) than in with controls (739 of 4740 [15.6%]).
In this large study, we report a high incidence of MBI-LCBI in recipients of allogeneic HSCT. Moreover, MBI-LCBI, similar to BSI-other, was associated with decreased OS as well as increased TRM. Furthermore, infection was more commonly reported as the primary or secondary cause of death for patients with MBI-LCBI or BSI. These data indicate that a reduction in BSI should be a key target for quality-improvement work to reduce mortality, morbidity, and consumption of health care resources.
Multivariable analysis of risk factors identified an increased risk of MBI-LCBI in patients with poor performance status, cord blood grafts, myeloablative conditioning, and posttransplant cyclophosphamide GVHD prophylaxis. Delayed engraftment is seen with umbilical cord blood grafts, increasing the time patients are at risk for MBI-LCBI. These data support current efforts to use umbilical cord blood graft expansion to reduce the duration of neutropenia. The increased risk seen with myeloablative conditioning likely reflects greater mucosal barrier injury and provides another focus for quality-improvement efforts. The increase in MBI-LCBI in patients receiving posttransplant cyclophosphamide may be associated with increased mucositis leading to susceptibility to translocation of bacteria into the bloodstream.
Reported evidence over the last decade shows that major progress has been made in preventing CLABSIs.13,29-31 However, to our knowledge, there are few data describing the mechanisms to decrease MBI-LCBIs. One of the original incentives for defining MBI-LCBI was to separate infections that could be reduced by attention to central venous catheter care from those that could not. In support of this definition, data demonstrate no change in MBI-LCBI rates with CLABSI prevention standard compliance, while the interventions were associated with CLABSIs.9,12,16 Although MBI-LCBI may not be amenable to central venous catheter care interventions, our data show that these infections are still associated with significant patient morbidity and mortality and that these infections are prevalent in this population.32,33 Mucosal barrier injury–laboratory confirmed bloodstream infections are associated with significant health care resource use. A single-center retrospective analysis demonstrated that 40% of patients with an MBI-LCBI required central venous catheter removal, 46% of patients developed septic shock at the time of blood culture, 23% of patients were transferred to the intensive care unit within 48 hours of infection and that all-cause mortality within 10 days was 9%.34
The National Healthcare Safety Network (NHSN) created the MBI-LCBI definition in 2013 to enable surveillance staff in hospitals to identify and report BSIs in oncology patients and those undergoing HSCT that likely were the result of mucosal barrier injury and therefore not preventable through recommended central line insertion and maintenance practices. There are limitations to the National Healthcare Safety Network’s MBI-LCBI classification scheme. The National Healthcare Safety Network list is likely not inclusive of all organisms that may cause BSI, owing to translocation across compromised oral or gastrointestinal mucosa.32 To support this, Tamburini et al35 demonstrated that organisms not classically thought to originate in the gut may develop a reservoir, leading to bacterial translocation (eg, Pseudomonas aeruginosa strains in the gut microbiome of a patient undergoing HSCT and in a subsequent BSI from the same individual). In addition, an absolute neutrophil count of greater than 500 cells/μL (a key part of the definition), is not necessarily associated with an intact mucosa.
Our study has limitations inherent to the registry database. First, our classification of MBI-LCBI is limited to the organisms in the National Healthcare Safety Network and correlated with the dates of neutrophil engraftment or subsequent decrease in neutrophil count and the onset of stage 3 or stage 4 acute GVHD as reported by centers. Consequently, there may be patients in the BSI-other group that actually had MBI-LCBI and vice versa. However, given the large number of patients in this study and the rigor used in data verification for engraftment and acute GVHD by the CIBMTR, this possibility is unlikely to have a significant association with our results. Second, there are no data captured on antibiotic prophylaxis or treatment, which may have varied considerably across centers and over time. Our analysis attempted to account for these variations by limiting centers to those with at least 1 patient with MBI-LCBI, with patients in the control and/or BSI-other category as centers apply antimicrobial prophylaxis and treatment in a standard manner across patients. Third, the degree of mucosal injury is a key factor for translocation of bacteria in the bloodstream; however, the severity of mucositis is not reported. Our finding of increased risk for recipients of myeloablative preparative regimens and those receiving posttransplant cyclophosphamide supports a role for the severity of mucositis. In contrast, use of palifermin, intended to decrease mucositis,36-38 was different across the 4 cohorts, with a slightly lower frequency in the control cohort. However, the small numbers of patients receiving palifermin in our cohort limited the examination in multivariable analysis. Fourth, the true association of MBI-LCBI with chronic GVHD may be underestimated owing to the time frame of the assessments in this cohort.
Our study has several strengths, including a robust sample size from 186 centers from diverse geographical locations and reflecting current transplant practices. In addition, to our knowledge, this is the first large-scale study to evaluate MBI-LCBI. The inclusion of multiple centers provides a diverse population of all ages, stem cell sources, and transplant types and minimizes overreporting or underreporting biases inherent in single-center studies. Uniform definitions were used for data collection stipulated by the CIBMTR, and long-term follow-up is ensured.
We found that MBI-LCBI, particularly in combination with another BSI, is negatively associated with post-HSCT outcomes and presents a burden to our health care system. Reduction in MBI-LCBI will require a better understanding of its mechanisms and risk factors, and our data contribute to the knowledge needed to make important progress.
Accepted for Publication: November 10, 2019.
Published: January 8, 2020. doi:10.1001/jamanetworkopen.2019.18668
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Dandoy CE et al. JAMA Network Open.
Corresponding Author: Christopher E. Dandoy, MD, MS, Division of Bone Marrow Transplantation and Immune Deficiency, Cincinnati Children’s Hospital Medical Center, 3333 Burnet Ave, MLC 11027, Cincinnati, OH 45229 (firstname.lastname@example.org).
Author Contributions: Dr Dandoy had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Dandoy, Kim, Diaz, Dvorak, Ganguly, Hashmi, Hematti, Olsson, Rotz, Lindemans, Komanduri, Riches.
Acquisition, analysis, or interpretation of data: Dandoy, Kim, Chen, Ahn, Ardura, Brown, Chhabra, Diaz, Dvorak, Farhadfar, Flagg, Ganguly, Hale, Hematti, Martino, Nishihori, Nusrat, Olsson, Sung, Perales, Lindemans, Komanduri, Riches.
Drafting of the manuscript: Dandoy, Kim, Chen, Ahn, Hale, Riches.
Critical revision of the manuscript for important intellectual content: Dandoy, Chen, Ardura, Brown, Chhabra, Diaz, Dvorak, Farhadfar, Flagg, Ganguly, Hale, Hashmi, Hematti, Martino, Nishihori, Nusrat, Olsson, Rotz, Sung, Perales, Lindemans, Komanduri, Riches.
Statistical analysis: Dandoy, Kim, Chen, Ahn, Diaz, Ganguly, Rotz, Riches.
Administrative, technical, or material support: Dandoy, Kim, Brown, Hale.
Supervision: Dandoy, Kim, Diaz, Ganguly, Hale, Perales, Komanduri, Riches.
Conflict of Interest Disclosures: Dr Ardura reported receiving grants from the National Institutes of Health National Institute of Allergy and Infectious Diseases, the National Institutes of Health Division of Microbiology and Infectious Diseases, and Shire; and personal fees from Imedex outside the submitted work. Dr Dvorak reported receiving personal fees from Alexion Inc, Jazz Pharmaceuticals, and Jasper Pharmaceuticals outside the submitted work. Dr Ganguly reported receiving personal fees from Seattle Genetics outside the submitted work. Dr Olsson reported receiving personal fees from AstraZeneca outside the submitted work. Dr Sung reported receiving grants from Merck and Novartis and was a consultant for Celltrion outside the submitted work. Dr Perales reported receiving personal fees from AbbVie, Bellicum, Bristol-Myers Squibb, Nektar Therapeutics, Novartis, Omeros, Takeda, Molmed, Servier, Medigene, and Merck; and personal fees and other from Incyte outside the submitted work. Dr Lindemans reported being a medical consultant for Chimerix outside the submitted work. Dr Riches reported being on the review board of Gamida Cell outside the submitted work. No other disclosures were reported.