ICU Hospitalization
Intensive care hospitalization in the first 100 days was necessary for 75 (22%) patients, with a median of 29 days (IQR 11-54) between HSCT and ICU hospitalization. For allogeneic HSCT patients, there was a 1.65 (95% CI 1.15-2.35; p=0.006) fold increased risk of ICU hospitalization for every doubling of pre-HSCT glucose CV. There was no association among autologous HSCT patients. Because 53% of patients requiring ICU hospitalization did so before day 30, associations between day 0-30 variability and time-to-ICU could not be evaluated.
Discussion
This retrospective cohort study demonstrates that GV is common in pediatric patients undergoing HSCT, and that it is associated with increased risk of infection and critical illness. These associations are consistent with research describing malglycemia as a whole, both in adult and pediatric HSCT patients.1-3 Associations between GV and mortality were less clear.
Roughly one third of patients both pre- and post-HSCT had glucose CVs that were in the range of adults who are either obese or have diabetes, with a striking 8-10% of pediatric patients having glucose CVs above the proposed cutoff for poor control in patients with type 1 diabetes.5 This suggests that the physiologic stress or iatrogenic exposures prior to and during HSCT, such as pre-HSCT asparaginase, radiation and preparative chemotherapy, or post-HSCT intravenous fluids, TPN, immunosuppressive agents, and steroids, appears to lead to abnormal insulin secretion or sensitivity, resulting in pediatric HSCT patients experiencing a pathologic level of GV.
In the pre-HSCT period, non-malignant diagnosis and prior treatment with asparaginase were associated with increased glucose CV. Asparaginase has known associations with pancreatitis and hyperglycemia, so this finding is not surprising.11-13 However, the reason for an association between non-malignant HSCT indications and higher glucose CV is unclear, as no patients in this group had disorders with known associations with glucose intolerance, and no patients had existing diabetes mellitus.14 It is possible that patients with malignant underlying diagnoses were more tolerant of the stress associated with HSCT preparation due to their prior treatment with chemotherapy, while for nonmalignant HSCT patients, HSCT preparation may have been more disruptive. While our prior research evaluating risk factors for pre-HSCT malglycemia identified underlying malignant diagnosis and adolescent age as risk factors, these were not associated with pre-HSCT glucose CV; thus, these findings were likely driven by hyperglycemia and not variability.3 Additionally, several other hypothesized risk factors, such as being overweight/obese, having prior steroid exposure, or having prior insulin needs, were not significant risk factors for increased pre-HSCT glucose CV. This may suggest that stress and exposures associated with HSCT result in these patients having similar glucose tolerance to an obese adult, regardless of BMI. The number of obese/overweight patients was small, so further studies to confirm this relationship are needed.
Consistent with prior literature on malglycemia or hyperglycemia, post-HSCT glucose CV was increased for patients who underwent allogeneic HSCT or who were treated with post-HSCT steroids or TPN.1,3,15-19 Prior insulin needs and any pre-HSCT radiation to the pancreas were associated with post-HSCT but not pre-HSCT increased glucose CV, suggesting that either the stress/exposures of undergoing a conditioning regimen levels out these factors pre-HSCT, or that post-HSCT, patients are more sensitive to their prior exposures.
The high prevalence of elevated glucose CV in this study, and risk factors for it, are notable because increased glucose CV was associated with adverse HSCT outcomes. Specifically, there was nearly a 5-fold increased risk of infection with doubling of pre-HSCT glucose CV, after adjusting for HSCT type. As can be visualized in Figure 1, median glucose CVs spanned a wide range, so differences in glucose CV of this magnitude (for example from 15% to 30%) are not an uncommon occurrence. Fortunately, there was not a similar association for patients who underwent autologous HSCT, which may be related to lower risk of early death.
While glucose CV was associated with viremia among allogeneic HSCT patients, our analysis did not demonstrate significant associations with SBI or fungal infection. This could be due to the confounding factor of the potential for multiple infections of different subtypes and inability to explore time-to-infection subtype, or due to the low number of fungal infections in our cohort. While potential underlying pathophysiology is detailed more below, future studies are needed to explore the mechanisms underlying the association between GV and viremia, including an evaluation of additional factors such as conditioning regimens (e.g. T-cell depletion), and lymphocyte function and microenvironment. As may be expected given the associations between GV and risk of infection, there was also an association between glucose CV and ICU hospitalization, but only among allogeneic HSCT patients. While we did not evaluate for an association between GV and relapse based on lack of association between malglycemia and relapse in our prior study, future studies may evaluate this specifically, and should also consider the association between GV and rarer events, such as graft failure, sinusoidal obstructive syndrome, and thrombotic microangiopathy. This study was not able to evaluate the latter two of these events not only because of anticipated low numbers of events, but because diagnosis and treatment of these events changed drastically over the course of this observation period. Of note, in addition to adjusting for treatment with steroids, HSCT type and other potential confounders when able, for all analyses, glucose values after the time of the event were censored. By doing so, any GV preceded the associated event. demonstrating temporality, which is critical in considering potential causation.
While the underlying pathophysiology of the association between hyperglycemia and adverse outcomes is more clearly hypothesized to be related to known immunologic effects such as increased inflammation and impaired leukocyte function, the possible pathophysiologic relationship between GV and outcomes is less well described.1,15,20-23 However, several studies have demonstrated associations between wide glucose fluctuations and both oxidative stress and endothelial cell apoptosis, which may lead to microvascular damage, inflammatory milieu, and disruption of normal immune function.24-26. In addition to these potential mechanisms, hyperglycemia is known to 1) impair neutrophil activity including chemotaxis, formation of reactive oxygen species, and phagocytosis; 2) impair lymphocytes by increasing apoptosis, suppressing proliferation and attenuating antibody production and function; and 3) increase pro-inflammatory state cytokines such as IL-6 and TNF-α.1,15,20,21,27-29 These changes, if true of variability as well, may explain the association between GV and viral infections, particularly through lymphocyte dysfunction. Further, if GV potentiates post-HSCT inflammatory milieu, increased risk for HSCT complications, more generally, may be expected.
Understanding the degree and mechanism by which GV contributes to HSCT complications and, potentially, mortality, and how these might be modifiable with improved glycemic control, is critical. If these are strengthened with future prospective studies, patients may benefit from supportive care interventions (e.g. enteral feeding instead of TPN, selection of immunosuppressive agents, insulin therapy) aimed at reducing GV. It is important to note, that multiple pediatric and adult studies evaluating the benefit of improved glucose control in non-HSCT patients in the intensive care setting were disappointing, often due to high risk of hypoglycemia.30-34 However, diabetes technology has progressed since these studies, and safe, intensive glucose control is increasingly feasible. Further, HSCT patients, who are likely to be at risk for longer than typical ICU patients, may find more benefit.
Even if there is no causal link, GV may serve as a useful biomarker for risk of post-HSCT complications such as infection or critical illness; predicting abnormal glucose control based on pre-HSCT characteristics may be useful for tools such as the pre-transplant comorbidity prognostic index.35 However, prior to assessing the utility of adjusting such tools or performing glucose interventions in patients, additional prospective work is needed.
This study has several limitations that must be considered when discussing implications and future studies. As with any retrospective study, causality cannot be determined. Prospective studies are necessary to confirm our results, as well as re-evaluate for associations between GV and mortality. Further, outcomes such as invasive fungal infection and GVHD relied on correct international classification of diseases (ICD) diagnostic coding and manual review and therefore have potential for misclassification bias. While this is unlikely to bias the data as classification challenges would be true across all patients, it may have resulted in decreased statistical power due to underreporting. Most limiting was our previously mentioned inability to evaluate for associations between post-HSCT GV and outcomes that tended to occur early in the post-HSCT period, such as infection and ICU hospitalization. As such, it is possible that this study missed a true underlying association between post-HSCT GV and increased infection or ICU stays. Additionally, while this study aims to highlight correlations between GV and adverse events, any biological causes are hypothetical and require prospective human and animal model-based studies. To this end, future work will use continuous glucose monitoring to prospectively describe GV, its risk factors, and associated outcomes. While this study demonstrated relationships between stress GV immediately preceding and following HSCT and outcomes; future work should evaluate whether GV that predates HSCT-related stress is also associated with adverse outcomes.
In summary, GV is a common occurrence in the pediatric HSCT population and is associated with increased risk of infection and critical illness in HSCT patients. Because GV is subtle and may be more easily overlooked than hyperglycemia or hypoglycemia by care teams, special attention to understanding its incidence, risk factors, and implications, is necessary.
Conflicts of Interest
G.P.F. has served as a consultant for Abbott Diabetes Care, an advisory board member for Dexcom, and conducts research sponsored by Medtronic, Dexcom, Bigfoot, Tandem, Insulet, and NovoNordisk. MRV has served as a consultant for Fate Therapeutics and B-mogen. All other authors declare no competing financial interests.
Acknowledgements
This work was supported by the National Institutes of Health (grant NIDDK 2K12DK094712-06). Contents are the authors’ sole responsibility and do not necessarily represent official NIH views.
Data Sharing
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Correspondence: Jenna Sopfe, Center for Cancer and Blood Disorders, Department of Pediatrics, University of Colorado School of Medicine, 13123 E. 16th Ave, B115, Aurora, Colorado, 80045; email: jenna.sopfe@cuanschutz.edu