Society for Endocrinology
European Society of Endocrinology
Digital vs conventional glycemic monitoring in rare endocrine cancers: comparison of effectiveness during chemotherapy
Lukas van Baal ®, Harald Lahner, Jasna Pavlovic, Lars C Moeller, Nicole Unger, Dagmar Führer-Sakel and Annie Mathew®
Department of Endocrinology, Diabetes and Metabolism, European Neuroendocrine Tumour Society (ENETS) Center of Excellence and Member of ENDO-ERN and EURACAN, University Hospital Essen, University Duisburg-Essen, Essen, Germany
Correspondence should be addressed to L van Baal: lukas.van-Baal@uk-essen.de
Abstract
Aims: Chemotherapy regimens can induce severe hyperglycemia, which may be underestimated using conventional point-of-care blood glucose (POC-G) measurement techniques. Real-time continuous glucose monitoring (rtCGM) systems may offer a more accurate assessment of glucose metabolism. In this study, we compared blood glucose monitoring using POC-G and rtCGM in patients with rare endocrine cancers caused by chemotherapy and steroid medication.
Methods: In this single-center observational study, we analyzed data from 76 hospitalized patients with pancreatic neuroendocrine tumors (n = 48) or adrenocortical carcinoma (n = 28) undergoing chemotherapy. Patients were monitored using either POC-G (n = 38) or rtCGM (n = 38). Glycemic metrics included time in range (TIR), prevalence of steroid-induced hyperglycemia (SIH), and HbA1c.
Results: Using POC-G, TIR was 23.6 + 0.9 h/day in patients without diabetes (NoD) and 20.0 ± 4.2 h/day in patients with diabetes undergoing chemotherapy (mean: five cycles). However, when rtCGM was used, drastic changes in TIR were documented under the same regimen. The mean TIR decreased from 21.7 h/day in patients with NoD to 14.6 h/day in patients with diabetes (P < 0.01). Similarly, the overall incidence of SIH was 30% using conventional POC monitoring, but this figure rose to 79% using rtCGM. During rtCGM use, HbA1c decreased by 0.3% over the course of the chemotherapy cycles, whereas during POC-G use, HbA1c increased by 0.2% (P < 0.01).
Conclusion: We demonstrated a previously underestimated frequency of hyperglycemia and SIH in patients undergoing chemotherapy by using rtCGM. The use of rtCGM enabled more detailed recognition of dysglycemia and may improve glucose metabolism during and after chemotherapy regimens.
Keywords: adrenocortical carcinoma (ACC); chemotherapy; continuous glucose monitoring (CGM); pancreatic neuroendocrine tumor (pNET)
Introduction
Approximately 20% of cancer patients present with diabetes as a relevant comorbidity (1). When undergoing chemotherapy, the risk of glycemic
decompensation increases significantly under steroids, frequently used to minimize adverse effects (1, 2, 3, 4, 5) Severe hyperglycemia during cancer treatment has been
CC
BY
NC ND
associated with poorer tumor response, a higher incidence of drug toxicity and an increased all-cause mortality (6, 7). In particular, dexamethasone is associated with severe hyperglycemia because it enhances insulin resistance in the liver, muscle, and adipose tissue as well as increasing hepatic gluconeogenesis (8). Therefore, the American Diabetes Association (ADA) recommends screening for existing diabetes before administering high-dose glucocorticoids and using insulin to control blood glucose levels (9). However, also patients without pre-existing diabetes may develop steroid-induced hyperglycemia (10).
Glucose monitoring during chemotherapy is typically performed using point-of-care capillary blood glucose (POC-G) measurements (11). However, the precision of the POC-G relies to a great extent on the number of measurements taken per day. This has significant clinical implications, as studies have demonstrated that at least one-third of postprandial hyperglycemic events may go undetected (12, 13). In contrast, real-time continuous glucose monitoring systems (rtCGM) provide a more comprehensive view of glucose dynamics, which may facilitate management and improve patient outcomes (14). However, inaccuracies must also be considered in relation to rtCGM, particularly since rtCGM measures glucose levels in a different body compartment. Consequently, changes in blood glucose levels will not be reflected in the interstitial fluid until at least 15 min have passed (15). Furthermore, there is a possibility of interference with medications that are frequently used in routine clinical practice (16).
To evaluate the differences in glucose monitoring using POC-G versus rtCGM during chemotherapy, we studied an in-house cohort of patients with two rare endocrine malignancies: pancreatic neuroendocrine tumor (pNET) and adrenocortical carcinoma (ACC). These patients are at high risk of developing decompensated glucose metabolism since chemotherapy regimens for these cancers include the prodiabetic drug streptozotocin, as well as high doses of dexamethasone (4, 17, 18).
Materials and methods
Patient cohort
This observational study was conducted from January 1, 2017, to January 10, 2025, at a European Neuroendocrine Tumor Society (ENETS) Center of Excellence. We included 76 consecutive, steroid-naive inpatients with rare endocrine cancers who received systemic streptozotocin-based chemotherapy and antiemetic treatment with high-dose dexamethasone (4.0-8.0 mg/day intravenously). A multidisciplinary tumor board made the indication for chemotherapy, which was performed according to established protocols. These protocols included streptozotocin
435 mg/m2 and 5-fluorouracil 396 mg/m2 over five days for patients with pNET, and cisplatin 40 mg/m2, doxorubicin 40 mg/m2, and etoposide 95 mg/m2 over four days for patients with ACC (17, 18). Patients with pNET received 8 mg of dexamethasone on days 1-5 and patients with ACC on days 1-3 for antiemetic prophylaxis. Patients were discharged after the last day of therapy based on their clinical state and logistical options. None of the patients had to remain in hospital longer due to acute side effects of the therapy or reduction of their general condition. Briefly, chemotherapy was administered for three cycles, followed by reevaluation and tumor board decision to continue therapy based on RECIST 1.1 criteria showing either stable disease or partial response and clinical benefit (19). In patients diagnosed with pNET, the interval between two cycles was six weeks, while in patients diagnosed with ACC, the interval was four weeks. The mean number of chemotherapy courses was five in both the pNET and ACC cohorts.
Patients who had received glucocorticoid therapy prior to or at admission (e.g., for pain control) those on somatostatin analogues or those with hormone- secreting tumors (e.g., insulin- or cortisol-producing tumors) were excluded from the analysis.
Assessment of dysglycemia
On admission for each chemotherapy cycle, all patients were systematically screened for dysglycemia by hemoglobin A1c (HbA1c) and random plasma glucose measurement. Diabetes was defined as HbA1c ≥ 6.5% and/or random glucose ≥ two times ≥11.1 mmol/L (200 mg/dL). Prediabetes (PreD) was defined as HbA1c 5.7-6.4%. Patients with random plasma glucose <11.1 mmol/L (200 mg/dL) and HbA1c < 5.7% were classified as normoglycemic (NoD).
From January 1, 2017, to March 31, 2022, all patients received usual diabetes care (UDC). From April 1, 2022, to January 10, 2025, digitalized diabetes management (smart diabetes care (SDC)) was performed. This digitalized in-hospital diabetes management service is performed at our department since 4th September 2020 as a quality improvement project. In accordance with a decision by the executive board of the clinic and the clinic’s management, the concept was subsequently extended to patients receiving steroid therapy from the second quarter of 2022 onwards. The entire nursing team underwent personal training in the utilization of the digital devices by a member of the diabetes team. Furthermore, all instructions were available for review on a tablet located on the ward and were freely accessible to all members of the nursing team. Furthermore, all elements of the SDC were prepared and introduced in a stepwise fashion in a wash-in phase (July 1, 2020-September 3, 2020). This approach enabled the team to address any queries or concerns
with regard to the diabetes team in a timely and effective manner.
During the UDC period, nurses were instructed to measure point-of-care glucose (POC-G) in all patients undergoing chemotherapy four times a day (before meals and once at night) using a StatStrip glucose meter (Nova Biomedical, Germany; coefficient for the slope of a laboratory glucose measurement system (Yellow Spring Instrument) vs StatStrip glucose meter: 1.023 (r = 0.989) with all StatStrip glucose meter readings within the A zone of Clarke error grid (CEG) analysis (15)). Each glucose value measured was documented in the patient’s electronic health record (EHR). The attending physicians routinely reviewed these values during each visit and made appropriate adjustments if necessary.
In the SDC period, all patients received an rtCGM (= FreeStyle Libre 3, Abbott, Germany), which continuously measured subcutaneous interstitial fluid glucose concentrations and a smartphone (iPhone SE, Apple, USA) receiving data from the rtCGM via an application (app; FreeStyle LibreLink). Glucose data transfer was performed in real time. The data were shared with the ward’s nursing team via app-to-app (LibreLinkUp) and cloud-based with the diabetes team (LibreView). Glucose measured by rtCGM (CGM-G) was documented at least four times per day in the patient’s EHR. To ensure quality control, a POC-G measurement with a maximum time lag of less than two minutes to CGM-G measurement was performed with the StatStrip glucose meter at least once a day and in defined situations, e.g., if doubts about CGM-G. In case of discrepancy between CGM-G and POC-G values, POC-G was used to guide further treatment. In this setting, the attending physicians also routinely reviewed the glucose values documented in the patient’s EHR during each visit and made appropriate adjustments if necessary. Furthermore, the option to access the ambulatory glucose profile reports for individual patients via the cloud was available. The rtCGM sensors were applied on the day of admission by a member of the diabetes team and remained in place until discharge. Following discharge from hospital, no further glucose monitoring was conducted.
At the time of the aforementioned change in diabetes management, no patients were undergoing chemotherapy. Consequently, we present a single enrollment and no patient examined received both diabetes management protocols.
The following parameters were used to describe glucose homeostasis in patients during chemotherapy in the SDC group: time in range (TIR, glucose 3.9-10 mmol/L (70-180 mg/dL), target >70%), time above range 1 (TAR1, glucose 10-13.9 mmol/L (180-250 mg/dL), target: <5%), time above range 2 (TAR2, glucose >13.9 mmol/L (>250 mg/dL), target <5%), time below range 1 (TBR1, glucose 3.8-3.0 mmol/L (69-54 mg/dL),
target <4%), time below range 2 (TBR2, glucose mmol/L < 3.0 mmol/L (<54 mg/dL), target <1%), and mean glucose (target <8.5 mmol/L (<154 mg/dL)) (20).
In the UDC group, available POC-G values were extrapolated, as if the calculated percentage of measurements within a time range equaled the actual time spent in that range.
Furthermore, patients in both groups were checked for the presence of steroid-induced hyperglycemia (SIH). SIH was defined as the presence of ≥ two POC-G and/or CGM-G ≥11.1 mmol/L (200 mg/dL) during chemotherapy. No or only one POC-G and/or CGM-G ≥11.1 mmol/L (200 mg/dL) was regarded as the absence of SIH.
Based on HbA1c and random plasma glucose at each admission, as well as POC- and CGM-G values during hospitalization, antidiabetic treatment was adjusted or initiated according to the American Diabetes Association guidelines (9).
Clinical and anthropometric data were extracted from the EHR. EHRs of study participants were also checked for substances that have been reported to interfere with the glucose measurement of the rtCGM (acetaminophen >4 g/d, acetylsalicylic acid >100 mg/d, ascorbic acid >500 mg/dL or a tetracycline) (16).
Informed consent was obtained from all individual participants included in the study. This study was performed in accordance with the principles of the Declaration of Helsinki. The study was approved by the responsible ethics committee (20-9333-BO).
Laboratory analysis
HbA1c was determined in EDTA blood samples using the Tosoh Automated Glycohemoglobin Analyzer HLC-723G8 (Tosoh Bioscience, Japan) by cation exchange chromatography using a three-step salt gradient with a measuring range of 2.4% (3 mmol/mol) to 22.3% (220 mmol/mol). HbA1c measurement in the Clinical Chemistry Department is accredited according to DIN EN ISO 15189:2014 and fulfills the standards of the ‘National Glycohemoglobin Standardization Program (NGSP; DCCT-aligned)‘.
Glucose was determined on NaF plasma blood samples with the Atellica CH GluH_3 (Siemens Healthcare Diagnostics Inc., Germany) by photometry. Glucose is phosphorylated with ATP by hexokinase. The resulting glucose-6-phosphate is oxidized by a glucose-6-phosphate dehydrogenase, which reduces NAD to NADH. Photometry is performed at 340 nm with deduction of the extinction of the buff.
Statistical analysis
Data were analyzed using GraphPad Prism (GraphPad Software Inc., USA) and SPSS 27.0 (IBM Corporation, USA)
| Variable | Total | SDC (n = 38) | UDC (n = 38) | Test statistics |
|---|---|---|---|---|
| Sex (f:m) (%) | 57.9:42.1 | 60.5:39.5 | 65.8:34.2:0 | X2 = 0.2 |
| P = 0.63 | ||||
| Age (years) | 58.1 ± 11.0 | 60.5 ± 10.9 | 55.7 ± 11.0 | F = 2.5 |
| P = 0.06 | ||||
| BMI c1 (kg/m2) | 26.8 ± 4.0 | 26.3 ± 4.8 | 27.2 ± 3.2 | F = 1.0 |
| P = 0.34 | ||||
| BMI overall (kg/m2) | 26.3 ± 5.0 | 26.1 ± 4.5 | 26.4 ± 5.5 | F = 1.2 |
| P = 0.80 | ||||
| pNET (%) | 63.2 | 65.8 | 60.5 | X2 = 0.2 |
| P = 0.63 | ||||
| ACC (%) | 36.8 | 34.2 | 39.5 | X2 = 0.2 |
| P = 0.63 | ||||
| Dexamethasone dose (mg/d) | 7.2 ± 1.9 | 7.5 ± 2.8 | 7.0 ± 1.4 | F = 1.0 |
| P = 0.34 | ||||
| Length of hospitalization (days) | 4.7 ± 0.7 | 4.7 ± 0.6 | 4.6 ± 0.7 | F = 0.1 |
| P = 0.99 | ||||
| Observation period | 26.1 ± 0.2 | 26.1 ± 0.1 | 26.1 ± 0.2 | F = 0.1 |
| P = 0.99 | ||||
| HbA1c c1 (%) | 6.3 ± 0.9 | 6.4 ± 1.0 | 6.1 ± 0.7 | F = 0.2 |
| P = 0.13 | ||||
| HbA1c overall (%) | 6.2 ± 0.8 | 6.2 ± 0.9 | 6.2 ± 0.7 | F = 0.1 |
| P = 0.99 | ||||
| NoD:PreD:D (%) | 32.9:31.6:35.5 | 23.7:31.6:44.7 | 42.1:31.6:26.3 | X2 = 0.4 |
| P = 0.10 | ||||
| Metformin prior to CTX (%) | 18.4 | 21.1 | 15.8 | X2 = 2.8 |
| P = 0.55 | ||||
| Insulin therapy prior to CTX (%) | 10.5 | 13.2 | 7.9 | X2 = 0.6 |
| P = 0.46 | ||||
| Newly initiated metformin therapy (%) | 19.7 | 26.3 | 13.2 | X2 = 2.0 |
| P = 0.15 | ||||
| Newly initiated insulin therapy (%) | 3.9 | 5.3 | 2.6 | X2 = 0.4 |
| P = 0.55 | ||||
| Hemoglobin (g/dL) | 11.3 ± 2.9 | 11.7 ± 2.7 | 10.8 ± 3.1 | F = 0.7 |
| P = 0.18 | ||||
| GFR (mL/min/1.73 m2) | 65.4 ± 17.2 | 66.9 ± 17.1 | 63.9 ± 17.3 | F = 4.0 |
| P = 0.45 | ||||
| CKD (%) | 22.4 | 21.1 | 23.7 | X2 = 0.1 |
| P = 0.79 | ||||
| Total bilirubin (mg/dL) | 0.7 ± 0.6 | 0.7 ± 0.6 | 0.6 ± 0.5 | F = 0.1 |
| P = 0.43 |
Values are given as mean ± standard deviation or percentage affected, and P-values are provided for comparison of SDC vs UDC. SDC, smart diabetes care; UDC, usual diabetes care; f, female; m, male; d, diverse; BMI, body mass index; NoD, no diabetes; PreD, prediabetes; D, diabetes; pNET, pancreatic neuroendocrine tumor; ACC, adrenocortical carcinoma; mg, milligram; d, day; g, gram; dL, deciliter; GFR, glomerular filtration rate; mL, milliliter; min, minute; m2, square meters; CKD, chronic kidney disease.
software. Results are shown as mean ± standard deviation or as an absolute number and percentage affected as indicated. A value of P < 0.05 was considered statistically significant. Laboratory values below and above the detection limit were set to the lower or higher detection limit, respectively. For categorical data, patients with and without diabetes as well as patients with SDC and UDC were compared using the chi-squared test. For continuous variables, a univariate ANOVA, followed by Bonferroni-corrected post hoc tests, was computed.
Analyses were performed both unadjusted and adjusted for age, sex, BMI, streptozotocin, and daily dexamethasone dose (as analysis of covariance, ANCOVA, with age, BMI, and daily dexamethasone dose as covariates, and sex and streptozotocin as between- subject factors). Correlation analysis between the change in HbA1c and glucometrics provided by the rtCGM in the SDC group were computed as Pearson’s r. The resulting number of pairs was sufficient to depict at least medium effect sizes (f = 0.32) with good statistical power (0.8) for all outcomes examined.
| Variable | NoD (n = 25) | PreD (n = 24) | D (n = 27) | Test statistics |
|---|---|---|---|---|
| Sex (f:m) | 64.0:36.0 | 43.5:56.5 | 62.9:37.1 | X2 = 0.3 |
| P = 0.57 | ||||
| Age (years) | 60.9 ± 13.1 | 65.0 ± 10.8 | 61.0 ± 9.3 | F = 0.8 |
| P = 0.38 | ||||
| BMI (kg/m2) | 25.9 ± 4.0 | 26.6 ± 6.1 | 26.1 ± 3.1 | F = 0.1 |
| P = 0.97 | ||||
| pNET (%) | 72.0 | 41.7 | 66.7 | X2 = 0.1 |
| P = 0.95 | ||||
| ACC (%) | 28.0 | 58.3 | 33.3 | X2 = 1.1 |
| P = 0.29 | ||||
| Dexamethasone dose (mg/d) | 7.3 ± 1.6 | 7.5 ± 1.4 | 7.8 ± 4.1 | F = 0.1 |
| P = 0.71 | ||||
| HbA1c (%) | 5.4 ± 0.5b,c | 6.1 ± 0.3a,c | 7.1 ± 1.2a,b | F = 15.8 |
| P < 0.01 | ||||
| Hemoglobin (g/dL) | 11.8 ± 2.7 | 12.3 ± 2.2 | 11.7 ± 2.3 | F = 0.9 |
| P = 0.42 | ||||
| GFR (mL/min/1.73 m2) | 67.1 ± 12.7 | 71.5 ± 17.8 | 67.1 ± 16.2 | F = 0.1 |
| P = 0.73 | ||||
| CKD (%) | 18.2 | 12.5 | 22.2 | X2 = 0.1 |
| P = 0.87 | ||||
| Total bilirubin (mg/dL) | 0.8 ± 0.8 | 0.7 ± 0.3 | 0.6 ± 0.4 | F = 0.8 |
| P = 0.38 |
Values are given as mean ± standard deviation or percentage affected; P values are given as a result of non-parametric Kruskal-Wallis tests on the three subgroups, the letters (a = NoD; b = PreD, c = D) indicate significant differences in Bonferroni-corrected post hoc Man-Whitney U-tests, computed in case of significant results from Kruskal-Wallis tests. NoD, no diabetes; PreD, prediabetes; D, diabetes; f, female; m, male; d, diverse; BMI, body mass index; pNET, pancreatic neuroendocrine tumor; ACC, adrenocortical carcinoma; mg, milligram; d, day; g, gram; dL, deciliter; GFR, glomerular filtration rate; mL, milliliter; min, minute; m2, square meter; CKD, chronic kidney disease; ASA, acetylsalicylic acid.
Results
Patient characteristics
A total of 76 patients (56.7% female, mean age 56.2 + 12.1 years) with rare endocrine cancers receiving chemotherapy were included in the study. Of these patients, 48 presented with a pNET (67.7%) and 28 with an ACC (32.3%). Mean BMI was 26.6 ± 3.9 kg/m2, and mean HbA1c was 6.2 + 1.0%. The mean length of hospitalization was 4.7 + 0.7 days with a mean observation period of 26 weeks for each patient. These parameters did not differ between the UDC and SDC group. Baseline characteristics are shown in Table 1 and Table 2.
Prior to chemotherapy, diabetes was present in 19 (25.0%), PreD in 15 (19.7%), and NoD in 42 (55.3%) patients. During chemotherapy, nine (11.8%) patients developed PreD and eight (10.5%) patients developed diabetes based on HbA1c and random plasma glucose. The proportion of patients with NoD, PreD, and diabetes did not differ significantly between the SDC and UDC group (Table 1).
Regarding quality control in the SDC group, a mean absolute relative difference (MARD) of 10.8% was observed between CGM- and POC-G. In no patient was there any doubt about any CGM G value, meaning that a POC counter measurement would have been necessary.
Alteration of glucose homeostasis in patients with and without diabetes
In both, the UDC and SDC group TIR significantly decreased from patients with NoD to patients with diabetes. In the UDC group, TIR decreased from 23.6 ± 0.9 to 20.0 ± 4.2 h/day (P = 0.01) and in the SDC group from 21.7 ± 1.2 to 14.6 + 1.7 h/day (P < 0.01) (Fig. 1). Furthermore, in the SDC group, TIR was significantly higher in patients with PreD compared to patients with diabetes (P < 0.001; Fig. 1). Conversely, time above range increased from patients with NoD to patients with diabetes. In detail, in the UDC group, TAR2 increased from to 0.0 ± 0.0 h/day to 1.5 + 2.1 h/day (P < 0.05) and in the SDC group from 0.2 + 0.2 to 2.2 ± 2.3 h/day (P < 0.001). Additionally, in the SDC group, an increased TAR2 from patients with PreD to diabetes, and an increased TAR1 from patients with NoD to PreD to diabetes could be demonstrated (Fig. 1). Regarding patients with a BMI < 25.0 kg/m2 and BMI ≥25.0 kg/m2 overall no significant differences in glucometrics could be revealed. In contrast to the SDC group, a trend toward a higher TIR and lower TAR1 was observed in the UDC group for patients with a BMI of <25.0 kg/m2, though this did not reach statistical significance.
The incidence of SIH increased from 8.3% in patients with NoD to 26.7% in patients with PreD and 54.5% in patients with diabetes (P = 0.01 each) in the UDC group.
TIR *** , TAR1 *** , TAR2*
TIR*, TAR2*
TIR *** , TAR1*, TAR2*
TIR *** , TAR1 ***
TIR *** , TAR1 ***
25
TBR2
20
TBR1
TIR
Time spent in h
15
TAR1
TAR2
10
5
0
SDC UDC
SDC UDC
SDC UDC
No Diabetes
Prediabetes
Diabetes
In the SDC group, SIH was documented in 63.6% of patients with NoD, 75.0% of patients with PreD, and 81.8% of patients with diabetes. The incidence of SIH did not differ regarding the cancer type or steroid dose and neither between patients with a BMI < 25.0 kg/m2 and ≥25.0 kg/m2.
With regard to the aforementioned analyses, no statistically significant influence of sex could be demonstrated.
Alteration of glucose homeostasis in SDC vs UDC
New-onset dysglycemia (PreD or diabetes) occurred in five (13.2%; two PreD; three diabetes) patients in the SDC and in 13 patients (34.2%; eight PreD; five diabetes) in the UDC group (P < 0.05). A diabetological intervention was performed in 55.6% (n = 20) of patients with dysglycemia in the SDC and in 27.8% (n = 10) of patients with dysglycemia in the UDC group (P = 0.02). In 26.3% (n = 10) of patients in the SDC group and 13.2% (n = 5) in the UDC group, treatment with metformin was initiated, and in 10.5% (n = 4) and 5.3% (n = 2), pre-existing insulin therapy was adjusted. New insulin therapy was initiated in 5.3% (n = 2) of patients in the SDC
group and in 2.6% (n = 1) in the UDC group. In addition, 13.2% (n = 5) and 7.9% (n = 3) of patients received detailed nutritional counseling (Table 1).
Overall, mean TIR was 22.1 + 3.1 h/day in the UDC group, which was significantly higher than the mean TIR of 18.2 ± 5.4 h/day in the SDC group (P < 0.01). Similar results were demonstrated, when comparing the subgroups of patients with NoD and diabetes (Fig. 1). Additionally, the mean TAR1 was significantly lower in the total UDC group with 1.2 + 2.3 h/day compared to 4.5 + 4.1 h/day in the SDC group (P < 0.01). This was also evident in the NoD and diabetes subgroups, which had significantly lower TAR1 values in the UDC group (Fig. 1). Furthermore, in the SDC group, a stepwise and significant decrease in TIR and a corresponding increase in TAR1 and TAR2 from patients to NoD to patients with PreD to patients with diabetes could be demonstrated. The latter was not evident in the UDC group (Fig. 1).
A subgroup analysis of glucometrics between days with and without steroids in patients with ACC showed the following: In the SDC group, mean TIR was significantly lower with 17.8 + 3.3 h/day, and vice versa mean TAR1 (5.2 ± 3.8 h/day) was significantly higher on steroid than on non-steroid days (TIR: 21.1 + 4.4; P = 0.04/TAR1: 2.1 ± 4.1; P = 0.04). However, in the UDC group, no significant difference regarding glucometrics on steroid vs non-steroid days could be revealed.
Regarding SIH, the incidence was significantly higher in the SDC group (79.0%) compared to the UDC group (30.0%; P < 0.01). Furthermore, the incidence of SIH significantly increased from patients with NoD to patients with PreD to patients with diabetes in the UDC group, while the incidence of SIH did not differ significantly between subgroups in the SDC group.
Subgroup analysis of hyperglycemic patterns in the SDC group revealed the following: Patients with SIH had a significantly lower TIR of 17.0 ± 5.0 h/day compared to a TIR of 22.8 + 0.8 h/day in patients with SIH (P < 0.01). Vice versa in patients with SIH, TAR1 was 5.5 ± 3.0 h/day and significantly higher compared to a TAR1 of 0.5 + 0.5 h/day in patients without SIH (P < 0.01). Furthermore, TBR1 and TBR2 were significantly lower in patients with SIH compared to without SIH (TBR1: 0.1 ± 0.1 vs 0.5 ± 0.5 h/day; P < 0.01/TBR2: 0.0 ± 0.0 vs 0.2± 0.2 h/day; P < 0.01).
With regard to the aforementioned analyses, again no statistically significant influence of sex could be demonstrated.
Absolute HbA1c-values at initiation of chemotherapy, during each cycle, and at the end of chemotherapy did not differ significantly between the SDC and UDC groups (Table 1). However, the longitudinal change in HbA1c differed significantly between the SDC and the UDC group with an HbA1c reduction of 0.3% in the SDC and an HbA1c increase of 0.2% in the UDC group (P < 0.01) (Fig. 2). Furthermore, after each cycle of chemotherapy,
A
B
2
0.15
*
POC only
Change in HbA1c (%)
0.10-
rtCGM
Change in HbA1c (%)
1
0.05
0.00
0
-0.05
-0.10-
-1
-0.15
2
3
4
5
6
-2
SDC
UDC
a continuous increase of HbA1c was observed in the UDC group, while in the SDC group, HbA1c decreased after each cycle, with the greatest decrease occurring between the fifth and sixth cycles. Regarding the individual HbA1c changes between the cycles, there was also a significant difference between the UDC and SDC groups between the fifth and sixth cycles (P = 0.03) (Fig. 2).
Regarding correlation analyses, no significant correlation could be revealed between changes in HbA1c and rtCGM-derived glucometrics or changes in BMI.
Discussion
In this observational single-center study, we compared digital and conventional diabetes monitoring methods in patients with rare endocrine cancers over two time periods. We found that digitalized glucose monitoring by rtCGM provides significantly more detailed insights into glycemic trends during chemotherapy and improves glucose metabolism throughout the course of chemotherapy. Specifically, we found that patients with diabetes spent an average of 6.5 more hours per day in hyperglycemia than patients without diabetes. Furthermore, we demonstrated that HbA1c improved by 0.3% in patients with digitalized glucose monitoring, while HbA1c impaired by 0.2% in patients with conventional glucose monitoring.
Historically, data analyzing blood glucose and chemotherapies were obtained through POC measurement. However, recent studies have demonstrated that glucose assessment using rtCGM provides more precise insights into glucose metabolism in various situations (21). Regarding patients receiving chemotherapy only, limited data are available on rtCGM use and mainly focuses on patients with pre-existing diabetes. In our study, TIR was significantly lower and TAR was significantly higher in patients with diabetes than in those without diabetes. This is in line with the study of Ulene et al. and Legris et al. investigating rtCGM use albeit in different oncological settings of patients receiving chemotherapy (22, 23).
Furthermore, older studies measuring POC-G demonstrated frequent increases with blood glucose values > 300 mg/dL in patients with and without diabetes receiving chemotherapy (24, 25). Moreover, the incidence of SIH ranges between 9.4 and 94.0%, depending on the steroid dose used and the clinical situation studied, e.g., the combined effect of glucocorticoid exposure and the physiological stress of chemotherapeutic agents (26, 27, 28). However, POC-G likely underestimates the true extent of hyperglycemia. Accordingly, our data analysis shows that under usual diabetes care, TIR was significantly higher and TAR significantly lower both in patients with and without diabetes. The probability of this occurrence appears highly unlikely, since chemotherapy protocols administered to the patients in both groups were identical, and likewise the proportion of patients with PreD and diabetes did not vary. It can therefore be assumed that conventional POC-G monitoring misses at least three hours of hyperglycemia per day. Considering that in clinical practice, only four POC-G measurements are usually taken per day, it becomes clear that conventional glucose monitoring may not be sufficient to estimate the degree of hyperglycemia induced by steroid exposition. Furthermore, rtCGM use provided a more detailed overview of the impact of pre-existing dysglycemia, as we demonstrated that glucose metabolism significantly worsened from patients with NoD to PreD to diabetes. However, this was not evident under conventional diabetes care. The absence of evidence indicating a substantial impairment in glucose metabolism on steroid compared to non-steroid days in the UDC group is consistent with these findings. Furthermore, the incidence of SIH was significantly lower under conventional diabetes care. These findings are in line with previous studies that have reported a tendency for POC-G to underestimate the incidence of SIH compared to studies that use rtCGM (22, 29). It is assumed that using an intermittent measurement approach with pre- and/or post-meal POC-G rather than rtCGM overlooks especially mild forms of SIH throughout the day. It is also noteworthy that the SDC group demonstrated a significant association between SIH and lower TIR,
higher TAR, and a reduced probability of hypoglycemia. This suggests that the presence of SIH, even without analyzing the AGP report, is a robust indication of glucose metabolism that requires optimization during ongoing chemotherapy.
Furthermore, we demonstrated a better long-term glucose control with reduced HbA1c levels during chemotherapy with a digitalized diabetes care. In contrast, an increase in HbA1c was observed in under usual diabetes care. It could be argued that an increase in Hba1c of 0.2% or a decrease of 0.3% may not be of clinical significance. However, it is noteworthy that a direct comparison of the two approaches results in a discrepancy of 0.5%. Furthermore, if we also consider that the risk of major cardiovascular events decreases with the onset of an HbA1c decline and even a 5% increase in TIR has been associated with improved outcomes in inpatient, the clinical relevance of the different approaches becomes clear (30, 31, 32). In addition, higher doses of steroids than those employed in the present protocol are frequently administered during chemotherapy, particularly in cases of hematological neoplasms (33). Consequently, it can be hypothesized that the long-term glycemic control will exhibit a greater divergence when comparing the two approaches in an appropriate setting. In this context, we would like to revisit our surprising correlation analysis results: we were unable to detect any correlation between HbA1c and TIR or TAR. It is important to note that the CGM measurements were only taken during the period of inpatient treatment. Given that this was also the period of high-dose steroid administration, the lack of correlation once again highlights the immense influence of steroid administration during chemotherapy on acute glucose metabolism. Conversely, CGM measurement, in contrast to POC measurement, facilitates precise identification of this influence. Consequently, this enables timely antidiabetic intervention. It can be hypothesized that this intervention enhances glucose metabolism during the period between chemotherapy cycles, thereby contributing to the observed decline in HbA1c levels. Accordingly, significantly more diabetological interventions were performed under digitalized diabetes management. In this regard, previous studies have demonstrated that the greater the extent to which diabetes-specific data are stored digitally for the treating physician, the more likely it is that therapy will be intensified, which ultimately results in better glycemic control (34, 35). In summary, one could assume that the demonstrated misinterpretation of glycemic control may with POC measurement prevent these patients from receiving an early and dedicated diabetes management. Therefore, our observations can serve as a foundation to further pursue in prospective, randomized studies with continuous CGM measurements until chemotherapy is completed to clarify whether greater insights into the glucose course would result in improved long-term outcomes. In particular, since impaired glucose control
during chemotherapy increases the risk of neutropenia, infections, and mortality (36, 37).
Regarding the hypoglycemic range, TBR was strongly limited to six minutes per day in our study population. In contrast, Legris et al. revealed a considerably higher TBR of three hours per day. One possible explanation is that 62.4% of their study population received antidiabetic treatment that carries a risk of hypoglycemia, whereas only 16.7% of patients in our study population received such treatment (23). Notably, no glucose values < 70 mg/dL were observed under conventional diabetes care. This is only partially comprehensible since inappetence, nausea, and vomiting, which are side effects of chemotherapy, are associated with an increased incidence of hypoglycemic events, independent from an existing insulin therapy (38). Furthermore, studies have demonstrated that hypoglycemia detection, particularly in patients with impaired hypoglycemia awareness, is inadequate using POC-G measurements (13, 39).
Noteworthy and contrary to the findings of previous studies, no correlation was observed between BMI and glucose derangement in the digitalized cohort (40). This phenomenon may be attributed to a greater volume of glucose data in comparison to other studies that employed only capillary glucose values.
Limitations and strengths
The most notable bias of our study is that the presented glucometrics in the UDC group were extrapolated, limiting the extern validity of our results. However, calculating the exact time in range, using POC measurement is not feasible in everyday clinical practice. Since data available on the use of CGM during chemotherapy are sparse, direct comparisons with POC measurement are particularly lacking, and rtCGM describes the influence of high-dose steroids on glucose metabolism during chemotherapy in more detail than conventional monitoring would have allowed, we decided to use this study approach despite its weaknesses. The rarity of endocrine carcinoma and consequently the small cohort size may limit the comparability of our results to those of patients with common cancers. However, it should be noted that, given an incidence of 0.48/100,000 and 1-2/1,000,000 of the cancers examined, our cohort size is relatively large (41, 42). Furthermore, as a single-center study, we can guarantee that all patients have been treated consecutive to the same therapeutic measures. This, in turn, significantly increases the validity of the study. Moreover, the approval of streptozotocin for the treatment of NETs did not include a phase 3 study. However, the diabetogenic potential of streptozotocin is well documented (43). With this study, we can therefore present the first real-world clinical data on the
influence of streptozotocin on glucose metabolism. Moreover, given that this is an observational study conducted as part of the daily routine at our clinic, the measurement of background steroid or ACTH levels was not a standard procedure. However, it is important to note that the individual stress response, particularly in the context of severe underlying disease, has the potential to result in additional increases in steroid levels (44). Consequently, these factors have the potential to induce hyperglycemia, thereby impacting the study’s outcomes (45).
A considerable strength of our study is that we performed glucose monitoring with rtCGM and made a direct comparison with POC-G measurements in the same setting. Therefore, we were able to confirm the hypothesis that POC-G provides limited information about daily glucose levels and is insufficient for estimating the degree of glucose metabolic dysregulation during chemotherapy. In addition to improving glucose data quality, rtCGM reduces nursing staff workload by eliminating routine fingerstick testing and manual documentation, and it may support more efficient interdisciplinary care. Although a formal cost analysis was not conducted, integrating rtCGM may also reduce costs through workflow optimization.
Conclusion
This study reveals that the burden of impaired glucose control and the incidence of SIH are underestimated in patients with rare endocrine cancers and diabetes undergoing chemotherapy. Furthermore, we demonstrated that digitalized diabetes monitoring could improve glucose metabolism during chemotherapy regimens. Our study underscores the need for intensified glucose monitoring in patients undergoing chemotherapy and indicates that rtCGM is a feasible prospect for future cancer care. Future research should explore the predictive value of CGM- derived metrics, such as glycemic variability, TIR, and mean glucose, in relation to chemotherapy response rates, infection risk, and length of hospital stay. Additionally, expanding rtCGM use to the post- discharge outpatient phase could help identify late- onset SIH and facilitate continuity of care for patients on prolonged steroid therapy.
Declaration of interest
LVB, HL, JP, LM, NU, DF, and AM declare no conflicts of interest.
Funding
LvB and AM were supported by the DFG-funded Clinician Scientist Programme UMEA (FU 356/12-1). The authors declare that no funds, grants, or other support were received during the preparation of the manuscript.
Author contribution statement
LVB and AM are responsible for the study design. LVB, HL, JP, and AM performed the patient recruitment and data acquisition. LVB wrote the first draft of the manuscript and performed the data analysis. HL, LM, NU, DF, and AM revised the drafts and participated in the writing process by making comments and suggestions and by approving the manuscript. All authors have read and agreed to the content of the manuscript.
Data availability
Individual participant data from this study will be available in an anonymized form from the publication date of this manuscript for the consecutive 24 months, on a collaborative basis for individual participant data meta-analyses. Proposals should be directed to Lukas van Baal.
Ethics statement
Informed consent was obtained from all individual participants included in the study. This study was performed in accordance with the principles of the Declaration of Helsinki. The study was approved by the Ethics Committee of the University Hospital Essen (20-9333-BO).
Acknowledgment
We would like to thank the diabetes team (Sylvia Widera and Michaela Raue) and the nursing staff of the Department of Endocrinology, Diabetology and Metabolism, University Hospital Essen, for their daily excellent assistance and care of our patients. Furthermore, we would like to thank the Department of Central Information Technology for their excellent assistance and providing the digital infrastructure.
References
1 Joharatnam-Hogan N, Chambers P, Dhatariya K, et al. A guideline for the outpatient management of glycaemic control in people with cancer. Diabet Med 2022 39 e14636. (https://doi.org/10.1111/dme.14636)
2 Harris D, Barts A, Connors J, et al. Glucocorticoid-induced hyperglycemia is prevalent and unpredictable for patients undergoing cancer therapy: an observational cohort study. Curr Oncol 2013 20 e532-e538. (https://doi.org/10.3747/co.20.1499)
3 Hwangbo Y & Lee EK. Acute hyperglycemia associated with anti-cancer medication. Endocrinol Metab 2017 32 23-29. (https://doi.org/10.3803/enm.2017.32.1.23)
4 Boulanger J, Boursiquot JN, Cournoyer G, et al. Management of hypersensitivity to platinum- and taxane-based chemotherapy: cepo review and clinical recommendations. Curr Oncol 2014 21 e630-e641. (https://doi.org/10.3747/co.21.1966)
5 Kos-Kudła B, Castaño JP, Denecke T, et al. European Neuroendocrine Tumour Society (ENETS) 2023 guidance paper for nonfunctioning pancreatic neuroendocrine tumours. J Neuroendocrino/ 2023 35 e13343. (https://doi.org/10.1111/jne.13343)
6 Barone BB, Yeh H-C, Snyder CF, et al. Long-term all-cause mortality in cancer patients with preexisting diabetes mellitus: a systematic review and meta-analysis. JAMA 2008 300 2754-2764. (https://doi.org/10.1001/jama.2008.824)
7 Liu X, Ji J, Sundquist K, et al. The impact of type 2 diabetes mellitus on cancer-specific survival. Cancer 2012 118 1353-1361. (https://doi.org/10.1002/cncr.26420)
8 Pichardo-Lowden AR, Fan CY & Gabbay RA. Management of hyperglycemia in the non-intensive care patient: featuring subcutaneous insulin protocols. Endocr Pract 2011 17 249-260. (https://doi.org/10.4158/ep10220.ra)
9 American Diabetes Association Professional Practice Committee. 16. Diabetes care in the hospital: standards of care in diabetes - 2025. Diabetes Care 2024 48 (Supp. Supplement_1) S321-S334. (https://doi.org/10.2337/dc25-S016)
10 Hershey DS, Bryant AL, Olausson J, et al. Hyperglycemic-inducing neoadjuvant agents used in treatment of solid tumors: a review of the literature. Oncol Nurs Forum 2014 41 E343-E354. (https://doi.org/10.1188/14.onf.e343-e354)
11 American Diabetes Association Professional Practice Committee. 16. Diabetes care in the hospital: standards of care in diabetes - 2024. Diabetes Care 2023 47 (Supp. Supplement_1) S295-S306. (https://doi.org/10.2337/dc24-S016)
12 Clement S, Braithwaite SS, Magee MF, et al. Management of diabetes and hyperglycemia in hospitals. Diabetes Care 2004 27 553-591. (https://doi.org/10.2337/diacare.27.2.553)
13 Burt MG, Roberts GW, Aguilar-Loza NR, et al. Brief report: comparison of continuous glucose monitoring and finger-prick blood glucose levels in hospitalized patients administered basal-bolus insulin. Diabetes Technol Ther 2013 15 241-245. (https://doi.org/10.1089/dia.2012.0282)
14 Clubbs Coldron B, Coates V, Khamis A, et al. Use of continuous glucose monitoring in non-ICU hospital settings for people with diabetes: a scoping review of emerging benefits and issues. J Diabetes Sci Technol 2023 17 467-473. (https://doi.org/10.1177/19322968211053652)
15 Zourek M, Jankovec Z & Hykova P. Glycemia in blood, brain and subcutaneous tissue measured by a continuous glucose monitoring system. Crit Care 2011 15 (Supplement 1) P404. (https://doi.org/10.1186/cc9824)
16 Heinemann L. Interferences with CGM systems: practical relevance? J Diabetes Sci Technol 2021 16 271-274. (https://doi.org/10.1177/19322968211065065)
17 Fassnacht M, Dekkers OM, Else T, et al. European Society of Endocrinology Clinical Practice Guidelines on the management of adrenocortical carcinoma in adults, in collaboration with the European Network for The Study of Adrenal Tumors. Eur J Endocrinol 2018 179 G1-G46. (https://doi.org/10.1530/eje-18-0608)
18 Falconi M, Eriksson B, Kaltsas G, et al. ENETS consensus guidelines update for the management of patients with functional pancreatic neuroendocrine tumors and non-functional pancreatic neuroendocrine tumors. Neuroendocrinology 2016 103 153-171. (https://doi.org/10.1159/000443171)
19 Schwartz LH, Litière S, de Vries E, et al. RECIST 1.1-update and clarification: from the RECIST committee. Eur J Cancer 2016 62 132-137. (https://doi.org/10.1016/j.ejca.2016.03.081)
20 Danne T, Nimri R, Battelino T, et al. International consensus on use of continuous glucose monitoring. Diabetes Care 2017 40 1631-1640. (https://doi.org/10.2337/dc17-1600)
21 Gómez AM, Umpierrez GE, Muñoz OM, et al. Continuous glucose monitoring versus capillary point-of-care testing for inpatient glycemic control in type 2 diabetes patients hospitalized in the general ward and treated with a basal bolus insulin regimen. J Diabetes Sci Technol 2015 10 325-329. (https://doi.org/10.1177/1932296815602905)
22 Ulene S, Wang S, Cook J, et al. Continuous glucose monitoring and rates of hyperglycemia during chemotherapy for early-stage breast
cancer. J Clin Oncol 2024 42 (16_Supplement) e24134. (https://doi.org/10.1200/jco.2024.42.16_suppl.e24134)
23 Legris P, Bouillet B, Paris J, et al. Glycemic control in people with diabetes treated with cancer chemotherapy: contribution of continuous glucose monitoring. Acta Diabeto/ 2023 60 545-552. (https://doi.org/10.1007/s00592-023-02032-z)
24 Terao N & Suzuki K. Glycemic excursion, adverse drug reactions, and self-management in diabetes patients undergoing chemotherapy: a literature review. Asia Pac J Oncol Nurs 2021 8 610-622. (https://doi.org/10.4103/apjon.apjon-2131)
25 Lyall MJ, Thethy I, Steven L, et al. Diurnal profile of interstitial glucose following dexamethasone prophylaxis for chemotherapy treatment of gynaecological cancer. Diabet Med 2018 35 1508-1514. (https://doi.org/10.1111/dme.13770)
26 Liu XX, Zhu XM, Miao Q, et al. Hyperglycemia induced by glucocorticoids in nondiabetic patients: a meta-analysis. Ann Nutr Metab 2014 65 324-332. (https://doi.org/10.1159/000365892)
27 Tamez-Pérez HE, Quintanilla-Flores DL, Rodríguez-Gutiérrez R, et al. Steroid hyperglycemia: prevalence, early detection and therapeutic recommendations: a narrative review. World J Diabetes 2015 6 1073-1081. (https://doi.org/10.4239/wjd.v6.i8.1073T)
28 Healy SJ, Nagaraja HN, Alwan D, et al. Prevalence, predictors, and outcomes of steroid-induced hyperglycemia in hospitalized patients with hematologic malignancies. Endocrine 2017 56 90-97. (https://doi.org/10.1007/s12020-016-1220-2)
29 Kleinhans M, Albrecht LJ, Benson S, et al. Continuous glucose monitoring of steroid-induced hyperglycemia in patients with dermatologic diseases. J Diabetes Sci Technol 2024 18 904-910. (https://doi.org/10.1177/19322968221147937)
30 UK Prospective Diabetes Study (UKPDS) Group. Intensive blood- glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet 1998 352 837-853. (https://doi.org/10.1016/S0140-6736(98)07019-6)
31 Kyi M, Colman PG, Wraight PR, et al. Early intervention for diabetes in medical and surgical inpatients decreases hyperglycemia and hospital-acquired infections: a cluster randomized trial. Diabetes Care 2019 42 832-840. (https://doi.org/10.2337/dc18-2342)
32 Battelino T, Danne T, Bergenstal RM, et al. Clinical targets for continuous glucose monitoring data interpretation: recommendations from the international consensus on time in range. Diabetes Care 2019 42 1593-1603. (https://doi.org/10.2337/dci19-0028)
33 Tilly H, Lepage E, Coiffier B, et al. Intensive conventional chemotherapy (ACVBP regimen) compared with standard CHOP for poor-prognosis aggressive non-hodgkin lymphoma. Blood 2003 102 4284-4289. (https://doi.org/10.1182/blood-2003-02-0542)
34 Reed M, Huang J, Graetz I, et al. Outpatient electronic health records and the clinical care and outcomes of patients with diabetes mellitus. Ann Intern Med 2012 157 482-489. (https://doi.org/10.7326/0003-4819-157-7-201210020-00004)
35 Cebul RD, Love TE, Jain AK, et al. Electronic health records and quality of diabetes care. N Engl J Med 2011 365 825-833. (https://doi.org/10.1056/nejmsa1102519)
36 Meyerhardt JA, Catalano PJ, Haller DG, et al. Impact of diabetes mellitus on outcomes in patients with colon cancer. J Clin Oncol 2003 21 433-440. (https://doi.org/10.1200/jco.2003.07.125)
37 Saydah SH, Loria CM, Eberhardt MS, et al. Abnormal glucose tolerance and the risk of cancer death in the United States. Am J Epidemiol 2003 157 1092-1100. (https://doi.org/10.1093/aje/kwg100)
38 Psarakis HM. Clinical challenges in caring for patients with diabetes and cancer. Diabetes Spectr 2006 19 157-162. (https://doi.org/10.2337/diaspect.19.3.157)
39 Cryer PE. Mechanisms of hypoglycemia-associated autonomic failure in diabetes. N Engl J Med 2013 369 362-372. (https://doi.org/10.1056/nejmra1215228)
40 German CA, Laughey B, Bertoni AG, et al. Associations between BMI, waist circumference, central obesity and outcomes in type II diabetes mellitus: the ACCORD trial. J Diabetes Complications 2020 34 107499. (https://doi.org/10.1016/j.jdiacomp.2019.107499)
41 Dasari A, Shen C, Halperin D, et al. Trends in the incidence, prevalence, and survival outcomes in patients with neuroendocrine tumors in the
United States. JAMA Oncol 2017 3 1335-1342. (https://doi.org/10.1001/jamaoncol.2017.0589)
42 Bilimoria KY, Shen WT, Elaraj D, et al. Adrenocortical carcinoma in the United States: treatment utilization and prognostic factors. Cancer 2008 113 3130-3136. (https://doi.org/10.1002/cncr.23886)
43 Junod A, Lambert AE, Orci L, et al. Studies of the diabetogenic action of streptozotocin. Proc Soc Exp Biol Med 1967 126 201-205. (https://doi.org/10.3181/00379727-126-32401)
44 Sapolsky RM. Individual differences and the stress response. Semin Neurosci 1994 6 261-269. (https://doi.org/10.1006/smns.1994.1033)
45 Fadini GP. Perturbation of glucose homeostasis during acute illness: stress hyperglycemia and relative hypoglycemia. Diabetes Care 2022 45 769-771. (https://doi.org/10.2337/dci21-0069)