Composite Metric of Glycemic Control Q-Score Is Elevated in Pediatric and Adolescent/Young Adult Hematopoietic Stem Cell Transplant Recipients

Abstract

Background:

Malglycemia in pediatric, adolescent and young adult (AYA) patients who undergo hematopoietic stem cell transplant (HSCT) is associated with increased infection and mortality rate. Continuous glucose monitoring (CGM) has been safely used in pediatric/AYA HSCT recipients, but there is a need for a composite metric that can easily be used in clinical settings to assess the glycemic control and identify high-risk patients who needs therapeutic intervention. Composite metrics derived from CGM have not been studied in pediatric/AYA HSCT patients.

Methods:

Patients aged 2–30 years old who are admitted inpatient while undergoing HSCT at Children's Hospital Colorado underwent CGM using the Abbot Freestyle Libre Pro device from up to 7 days before and 60 days after HSCT. A composite metric Q-score, comprising five primary factors of CGM profiles (central tendency, hyperglycemia, hypoglycemia, intradaily variations, and interdaily variations), was calculated for each patient for the duration of CGM wear.

Results:

Twenty-nine patients received CGM for an average of 25 days per participant. The median Q-score was 10.2 (interquartile range [IQR]: 8.3, 14.3). Sixty-nine percent of patients had Q-scores that would be categorized into the Fair or Poor category. There was no difference in the Q-score by sources of stem cell, types of primary disease, types of preparative regimen, need for PICU admission, presence of documented infections, and total parenteral nutrition use in the peri-HSCT period.

Conclusions:

Most pediatric/AYA HSCT recipients have Q-scores indicating suboptimal glycemic control in the peri-HSCT period. Future study should focus on developing screening and treatment strategies to improve malglycemia and its associated adverse clinical outcomes. This study was registered at clinicaltrials.gov (NCT03482154).

Introduction

Patients with acute illness are at increased risk of developing malglycemia, defined as hypoglycemia (blood glucose [BG] <70 mg/dL), hyperglycemia (BG ≥126 mg/dL), and/or glycemic variability (standard deviation [SD] ≥29 mg/dL).¹ Previous studies have shown associations between malglycemia and adverse clinical outcomes, including death, infections, and length of hospitalization.^2–4 Patients who undergo hematopoietic stem cell transplant (HSCT) for treatment of malignant and nonmalignant diseases are at increased risk of developing malglycemia as well. Malglycemia in adult HSCT patients is associated with increased infection, graft-versus-host disease (GVHD), organ dysfunction, delayed engraftment, and mortality rate.^1,5–16

In pediatric and adolescent and young adult (AYA) HSCT patients, malglycemia is associated with increased risk of infection, transplant-related mortality rate, and all-cause mortality rate.^17,18 Malglycemia during the peri-HSCT period is thought to be multifactorial, including underlying insulin resistance, stress hyperglycemia, decreased insulin secretion, total parenteral nutrition (TPN) use, and medication side effects from steroids and calcineurin inhibitors.^{7,8,10,16,19,20}

Continuous glucose monitoring (CGM) is a technology that automatically measures and tracks BG levels allowing for minimally invasive, real-time assessment of glycemia with values obtained up to every 5 min. Continuous glucose monitors, which are small devices inserted in the interstitial fluid space, have been shown to be safe and effective for both children and adults with type 1 diabetes mellitus (T1DM).^21,22 CGM has also been successfully used to monitor glucose levels in hospitalized patients without preexisting diabetes, including general adult inpatient ward patients, pediatric/AYA HSCT patients, solid organ transplant recipients, those with COVID-19 infection, those requiring intensive care unit (ICU) admission, as well as neonatal ICU and pediatric ICU patients.^23–32

CGM records consist of hundreds of BG measurements per patient per day and allow for visualization and analysis of daily glucose level patterns. While standard analysis metrics have been adopted for the T1DM and T2DM diabetes populations, such metrics have not been studied in other populations experiencing malglycemia in times of acute illness.^33,34 Also, these metrics, while valuable for clinical research, are difficult to incorporate into day-to-day clinical care, especially among nonendocrinology providers.

To facilitate both clinical interpretation and future studies, a single, composite metric from CGM data to assess glycemic control is needed. The Q-score is a composite metric from CGM data that is constructed from five primary factors determining a glucose profile: central glucose tendency, hyperglycemia, hypoglycemia, intradaily variations, and interdaily variations.³⁵ The Q-score is increased in individuals with poor glycemic control, and, therefore, allows for the clinician to screen for CGM profiles that require therapeutic action.³⁵ When coupled with a computerized decision support system that simulates different therapeutic options and provides the results to the treating physician, the Q-score resulted in improved short- and long-term glycemic controls in individuals with type 2 diabetes mellitus (T2DM).³⁶

In this study, we used the CGM data, obtained prospectively from a cohort of pediatric/AYA HSCT recipients, to explore the potential use of Q-score as a metric of glycemic control during the peri-HSCT period. Furthermore, we evaluated the relationship between clinical characteristics and outcomes with the Q-score to identify high-risk subgroups that could benefit from the optimization of glycemic profile by therapeutic intervention.

Methods

Participants

Patients undergoing HSCT at Children's Hospital Colorado were approached for participation in an observational prospective cohort study exploring the safety and utility of the Abbott Freestyle Libre Pro CGM from February 2017 to January 2019. The inclusion criteria were as follows: (1) age 2–30 years at the time of HSCT; (2) undergoing HSCT at Children's Hospital Colorado; and (3) willing to wear CGM for the duration of the study and willing to follow study protocols.

The exclusion criteria were as follows: (1) preexisting diagnosis of T1DM or T2DM, or insulin requirement in the 2 weeks before HSCT; (2) preexisting condition requiring use of nonphysiologic steroids within 2 weeks of HSCT; (3) severe psychiatric disease or developmental delay that may interfere with the ability to provide informed consent or wear the CGM; (4) active skin condition that would affect CGM placement; and (5) any other condition that in the opinion of the investigators impaired the person's ability to safely participate in the trial. Additional details of the protocol can be found in the safety and tolerability publication.³² This study was approved by the University of Colorado Institutional Review Board.

Continuous glucose monitoring

The Abbott Freestyle Libre Pro (Abbott Diabetes Care) was used to continuously monitor glucose. CGM placement occurred up to 7 days before HSCT, once patients met the following criteria: (1) admitted to the hospital for the HSCT; (2) completion of any planned radiotherapy; and (3) 48 h after thiotepa infusion, when relevant. CGMs were removed and subsequently replaced at the end of the food and drug administration (FDA)-approved duration of wear (14 days), radiotherapy, and for any imaging studies, such as computerized tomography and magnetic resonance imaging; CGM use ended at hospital discharge. The clinical team was blinded to the data, except for retrospective disclosure for extreme values (<50 or >300 mg/dL). All CGM data for the duration of the study were used to calculate the Q-score.

Q-score

The Q-score is a composite metric constructed using CGM parameters including mean serum glucose (MSG, in mmol/L), time outside the Q-score target range of 3.9–8.9 mmol/L (t[hyper] and t[hypo], in hours), range (in mmol/L), and mean of daily difference (MODD, in mmol/L).³⁵ The formula for the Q-score is as follows:

$$\begin{array}{cc}Q& =8+\frac{MSG-7.8}{1.7}+\frac{Range-7.5}{2.9}+\frac{t\left[hypo\right]-0.6}{1.2}\\ & +\frac{t\left[hyper\right]-6.2}{5.7}+\frac{MODD-1.8}{0.9}\end{array}$$

The developers of the Q-score defined five categories of Q-scores as follows: <4.0, very good; 4.0–5.9, good; 6.0–8.4 satisfactory; 8.5–11.9, fair; and ≥12.0, poor, with fair or poor profiles indicating insufficient glycemic controls.³⁵

Clinical characteristics and outcomes

Prospective data collection of clinical characteristics included patient characteristics (age, gender, race/ethnicity, Tanner stage, BMI percentile, and insurance type), primary disease types for indication of HSCT (malignant disorders and nonmalignant disorders), stem cell sources (allogeneic and autologous), overall survival, infectious complications, number of ICU days, and use of TPN within 100 days after HSCT. Infectious complications were defined as a positive microbiology result or diagnostic/problem list code necessitating treatment/hospitalization. Infections were further categorized by serious bacterial infection (i.e., bacteremia, pneumonia, meningitis, peritonitis), viremia/viruria, or invasive fungal infection.

Statistical analysis

The size of the sample in the current study was determined without a priori power calculation as this study used the data obtained from a previous trial.³² A Q-score was calculated for each participant using all available CGM data using R. Descriptive statistics reported are mean ± SD, median (25th, 75th percentile), or frequencies (%). Continuous variables were compared using t-tests, and chi-squared or Fisher's exact tests were used for categorical variables. All tests performed were two-sided and a P-value of <0.05 was considered statistically significant. Analyses were performed using R versions 4.1.3.

Results

Patient characteristics

Of the 66 patients meeting the study eligibility criteria, 30 patients (45.5%) consented to participation. Twenty-nine patients were included for analysis as one patient withdrew consent before wearing a CGM. Participant characteristics are detailed in Table 1. Participants had a mean age of 11.8 (SD: 6.8) years at the time of stem cell transplant (range: 7.8–24.0 years old), and 16 (55.2%) were male.

Table 1.

Participant Characteristics

Variable	Category/statistic	N (%) or Mean (SD)
Total cohort		29 (100%)
Sex	Male	16 (55.2%)
	Female	13 (44.8%)
Age at HSCT		11.8 (6.8) years
Race/ethnicity	African American	5 (17.2%)
	Asian	1 (3.4%)
	Hispanic	8 (27.6%)
	Multiple/other	2 (6.8%)
	White	13 (44.8%)
Stem cell source	Autologous	9 (31.0%)
	Allogeneic	20 (69.0%)
Indication for HSCT	Malignant disorders	19 (65.5%)
	Nonmalignant disorders	10 (34.5%)
Preparative regimen	Chemotherapy-based	25 (86.2%)
	TBI-based	4 (13.8%)
PICU admission in the first 100 days of HSCT	Yes	4 (13.8%)
	No	25 (86.2%)
Documented infections in the first 100 days of HSCT	Yes	13 (44.8%)
	No	16 (55.2%)
Received TPN in the first 100 days of HSCT	Yes	12 (41.4%)
	No	17 (58.6%)

HSCT, hematopoietic stem cell transplant; PICU, pediatric intensive care unit; SD, standard deviation; TBI, total body irradiation; TPN, total parenteral nutrition.

Clinical characteristics

A total of 29 patients wore a total of 84 CGM sensors, with a median of 2 CGM sensors per patient over a median of 25 (interquartile range [IQR]: 21.5–30.0) days. Twenty (69.0%) received allogeneic stem cell transplants. Nineteen (65.5%) patients had malignant disorders for indication of stem cell transplant. Ten patients received stem cell transplants for nonmalignant primary disorders, including hemoglobinopathy, bone marrow failure disorder, and primary immune dysfunction. No patients died within the first 100 days post-HSCT.

No patients were on insulin therapy during the time of CGM. Sixteen (55.2%) patients did not have any documented clinically significant infection in the study period, while 13 patients had one or more infectious complications. Twelve (41.4%) patients received TPN within the first 100 days after HSCT. Two (6.9%) patients were admitted to the ICU in the first 100 days of stem cell transplant.

CGM metrics

The median mean glucose level of all participants during the study period was 99.2 g/dL (IQR: 80.7–104.9). The median SD and coefficient of variation (CV) of glucose level of all participants during the study period were 17.5 g/dL (IQR: 14.6–24.5) and 0.195 (IQR: 0.158–0.277), respectively. The mean glucose, SD, and CV were not significantly different by types of primary disease, types of preparative regimen, need for PICU admission, presence of documented infections, and TPN use in the post-HSCT period.

Q-score

The Q-score in the whole study group during the peri-HSCT period ranged from 7.8 to 24.0. The median Q-score of all participants during the study period was 10.2 (IQR: 8.3–14.3). There was no significant difference in median Q-scores in the pre-HSCT period and post-HSCT period (Supplementary Table S1). The median Q-score for those who received autologous HSCT was 14.3 (IQR: 10.0–15.4), compared with the median Q-score of 9.7 (IQR: 8.3–13.2) among those who received allogeneic HSCT (P = 0.317). Using the five categories defined by the developers of the Q-score, of 29 participants, none had Q-score that corresponds to the Very good or Good categories. While 9 participants had Satisfactory Q-scores, 20 participants (69.0%) had Q-scores that corresponded to the Fair and Poor categories (Table 2).

Table 2.

Q-Score Categories

Categorization	Q-score	n (%)
Very good	<4.0	0 (0)
Good	4.0–5.9	0 (0)
Satisfactory	6.0–8.4	9 (31.0)
Fair	8.5–11.9	9 (31.0)
Poor	≥12.0	11 (37.9)

The Q-scores were not significantly different by types of primary disease, types of preparative regimen, need for PICU admission, presence of documented infections, and TPN use in the post-HSCT period (Table 3).

Table 3.

Q-Score by Subgroups

Variable	Category/statistic	Median Q-score (IQR)	P
Total cohort		10.2 (8.3, 14.3)
Stem cell source	Autologous	14.3 (10.0, 15.4)	0.317
	Allogeneic	9.7 (8.3, 13.2)
Indication for HSCT	Malignant disorders	10.0 (8.3, 14.3)	1
	Nonmalignant disorders	10.3 (8.3, 14,2)
Preparative regimen	Chemotherapy-based	10.2 (8.2, 14.5)	0.879
	TBI-based	10.2 (9.0, 11.7)
PICU admission in the first 100 days of HSCT	Yes	10.4 (9.2, 12.0)	0.927
	No	10.2 (8.3, 14.3)
Documented infections in the first 100 days of HSCT	Yes	10.4 (9.6, 14.5)	0.329
	No	9.7 (8.2, 14.3)
Received TPN in the first 100 days of HSCT	Yes	11.1 (9.3, 14.3)	0.879
	No	9.6 (8.3, 14.5)

IQR, interquartile range.

Discussion

Patients with acute illness are at increased risk of developing malglycemia, a complication associated with adverse clinical outcomes, including death, infection, and length of hospitalization.^2–4 Pediatric/AYA HSCT recipients have a high incidence of malglycemia, which is shown to be associated with worsening overall survival, transplant-related mortality rate, and infection. Sopfe et al conducted a retrospective study of 344 pediatric and AYA patients who underwent HSCT and investigated the relationship between glycemic control and HSCT outcomes. Malglycemia occurred in 43.9% of patients.^17,18 Those with a post-HSCT day 0–100 mean glucose of ≥125 mg/dL had a sevenfold (95% confidence interval [CI]: 3.84–12.99; P < 0.0001) increased risk of death compared with patients with mean glucose <100 mg/dL. For every 10 mg/dL increase in pre-HSCT glucose level, there was 1.11-fold (95% CI: 1.04–1.18; P = 0.0013) increased risk of post-HSCT infection.¹⁷

In addition to hyperglycemia, higher glycemic variability, measured by coefficients of variation, was independently associated with increased mortality rate as well.¹⁸ While these studies were limited by the retrospective design, they provided insight into the significance of the problem and the need to prospectively identify those at risk of developing malglycemia to initiate prompt treatment to mitigate the adverse outcomes.

Some components of malglycemia, such as glycemic variability or transient hyperglycemia, or hypoglycemia, can be difficult to detect without continuous monitoring of glucose levels. With CGM devices resulting in hundreds of data points for each patient, there is a need for a composite metric that allows for estimation of the quality of glycemic control and stratification of those who will benefit from therapeutic interventions. This study used the data from the only study of CGM in the pediatric/AYA HSCT recipient population and is the first of its kind investigating a composite metric of glycemic control using CGM data in this at-risk group.

The present study showed that pediatric/AYA patients undergoing HSCT have a median Q-score of 10.2. The majority of patients had Q-scores that correspond to the Fair or Poor categories, according to the Q-score categorization defined by its developers. The Fair category is described as glucose levels often outside the target range (50%–80%), with high variability, and where hypoglycemic episode events can occur. The Poor category is described as glucose levels mostly outside the target range (>80%), with very high variability, and where hypoglycemic episodes occur. Both the Fair and Poor categories indicate that there is insufficient glycemic control and require therapy optimization.³⁵

The median Q-score in the present study population also was comparable with the mean Q-scores of T2DM and T1DM patients who required insulin therapy, which are 9.2 (SD: 3.4) and 12.7 (SD: 3.2), respectively.³⁵ This high median Q-score in our population corroborates the previous study that showed a high prevalence of malglycemia in pediatric HSCT recipients and further supports the need for close monitoring and the potential need for insulin therapy to manage it.

The Q-score, a composite metric used in the present study, is constructed from CGM data, to provide global information on glucose profiles summarized into a single value. Among different composite metrics, it is unique in using all components of malglycemia, including mean glycemia, hyperglycemia, hypoglycemia, and variability both within days and between days,³⁷ making it ideal to assess the glycemic control over several days in a population that may have transient malglycemia. In addition, this metric has been categorized by diabetes specialists, and the higher Q-score is shown to be associated with the need for increased therapy complexity.³⁵

We also investigated the relationship between the Q-score and several clinical characteristics, including the source of stem cell, underlying primary diagnoses categories, preparative regimen, ICU admission, infection, and TPN use. While we did not find any statistically significant difference between the subgroups by clinical characteristics in this small sample, the median Q-scores of all subgroups were 9.65 or higher and the majority of patients were categorized into the Fair and Poor classes, which are characterized by having glucose levels outside the target range for the majority of the time with high or very high variabilities. Future larger studies should prospectively evaluate risk factors and consequences of malglycemia, as represented by the Q-score.

The present study is limited by the small number of participants. Due to its small size and uneven distribution in clinical characteristics, it is possible that we were not able to detect a difference in the Q-scores in subgroups. Furthermore, we did not investigate the relationship between the Q-scores and the use of medications that alter glucose metabolism, such as glucocorticoids. Glucocorticoids are often the first-line therapy for engraftment syndrome, GVHD, autoimmune hemolytic anemia, and other complications of HSCT. In addition, while the Q-score conveniently presents a single value comprised from several aspects of glycemic profile, this metric has been only studied in patients with diabetes mellitus. There are many other composite metrics of CGM available, such as the Blood Glucose Risk Index(BGRI), Comprehensive Glucose Pentagon (CGP), Glycemic Risk Assessment Diabetes Equation (GRADE), and Index of Glycemic Control (IGC).³⁷

This study does not demonstrate whether the Q-score is the superior or optimal metric to use in this population over other metrics available. Future larger studies should also evaluate shorter intervals of Q-scores (such as immediately surrounding infection or ICU stay) to determine more proximal glycemia associations.

Conclusions

Q-scores are a promising composite metric of CGM profiles and glycemic control in children and young adults receiving HSCTs during their inpatient admission. The majority of pediatric/AYA HSCT recipients had a Q-score that corresponds to the Fair or Poor category, comparable with Q-scores of T2DM and T1DM patients requiring insulin therapy. These findings corroborate previous findings of a high prevalence of malglycemia in pediatric/AYA HSCT recipients and indicate a potential need for proactive treatment of malglycemia. Further investigations are needed to better understand the utility of Q-score use, clinical associations with Q-score derangements, and to develop screening and treatment strategies to improve glycemic control in this at-risk population and the effect of improved glycemic control on clinical outcomes.

Disclaimer

The contents are the authors' sole responsibility and do not necessarily represent the official NIH views.

Author Disclosure Statement

G.F. has research support from Medtronic, Dexcom, Abbott, Tandem, Insulet, Lilly, and Beta Bionics; consulting, speaking, and on the board for Medtronic, Dexcom, Abbott, Tandem, Insulet, Lilly, and Beta Bionics. Other authors have no conflict of interest to report.

Authors' Contributions

S.C.: Conceptualization (equal); writing—original draft (lead); formal analysis (equal); and writing—review and editing (equal). T.V.: Methodology (lead); formal analysis (equal); and writing—review and editing (equal). L.P.: Methodology (supporting); formal analysis (equal); and writing—review and editing (equal). A.F.: Writing—review and editing (equal). J.S.: Data collection (lead) and writing—review and editing (equal). F.J.: Data collection (supporting) and writing—review and editing (equal). G.F.: Conceptualization (equal); writing—original draft (supporting); formal analysis (equal); and writing—review and editing (equal).

Funding Information

Supported by NIH/NCATS Colorado CTSI Grant Number UL1 TR002535.

Supplementary Material

Supplementary Table S1