Diabetic ketoacidosis due to a sensor defect of FreeStyle Libre: A case report

ABSTRACT

The FreeStyle Libre Flash Glucose Monitoring System allows users to obtain sensor glucose values by scanning with the reader or their mobile phone. We report a case of a 59‐year‐old man with type 1 diabetes mellitus who developed diabetic ketoacidosis due to a sensor defect. After replacing the sensor with a new one, the glucose value shown in the device was much lower than usual, which made him consider that he was hypoglycemic. Accordingly, he reduced his insulin dose and eventually developed diabetic ketoacidosis. He was unaware of the discrepancy due to the lack of self‐monitoring of his blood glucose, although he was educated to do. In sum, glucose monitoring with the FreeStyle Libre is helpful; however, it is necessary to remind the patient that a sensor defect leading to a severe complication frequently happens.

The patient was unaware of the discrepancy between the sensor glucose value and blood glucose, judged it hypoglycemia, and gradually reduced his insulin dose to diabetic ketoacidosis.

INTRODUCTION

The FreeStyle Libre Flash Glucose Monitoring System (Abbott Diabetes Care, Alameda, CA, USA), recently coined as intermittently scanned continuous glucose monitoring (isCGM), is a device from which users can obtain sensor glucose values by scanning the sensor with the reader or their mobile phone. The sensor worn on the arm can be measured for 14 days and is less expensive than the real‐time CGM. Libre is factory‐calibrated, using wired enzyme technology, and finger prick glucose calibration is not supposed to be required during use ¹ . However, we experienced a patient with type 1 diabetes, admitted to the hospital in an emergency with diabetic ketoacidosis (DKA) caused by a sensor defect of FreeStyle Libre, and report this case here.

CASE REPORT

A 59‐year‐old male had been suffering from fulminant type 1 diabetes for about 25 years and visiting our outpatient clinic routinely for 7 years. Currently there was no retinopathy or stage 1 nephropathy, but neuropathy was present. He had been treated with intensive insulin therapy and started using FreeStyle Libre 4 years ago. His HbA1c had been always from 8.5 to 10% before and after coming to our hospital.

The patient was diagnosed with fulminant type 1 diabetes from his onset pattern at the age of 34 years old. His C peptide level had been below 0.02 (ng/mL) regardless of blood glucose levels, which is considered a complete depletion of insulin. When he was transferred to our hospital at the age of 52, he refused to perform self‐monitoring of blood glucose (SMBG), thus, had been hospitalized for education twice. However, he had not used SMBG, and seemed to have mainly adhered to the insulin units set for him at that time. At the age of 55 years old, the FreeStyle Libre had been initiated because he agreed to use it.

He injected a self‐adjusted dose of insulin at least four times a day, according to his sensor glucose value and food intake at the time. Most recently, he had been using rapid‐acting insulin Aspart, about 22 units at breakfast, about 8 units at lunch, about 8 units at dinner, and long‐acting insulin Glargine, about 12 units before sleep. Six days prior to the visit to the hospital, he had replaced the FreeStyle Libre sensor, where glucose value had been much lower than usual (Figure 1). Since he had not been self‐monitoring his blood glucose, he was unaware of the discrepancy between the sensor glucose value and blood glucose, judged it hypoglycemia, and gradually reduced his insulin dose (Figure 2). He reduced Aspart to 6 units at each meal time and discontinued Glargine to cope with the low glucose shown in the sensor 5 days before his regular visit to the hospital. As he observed no improvement of the hypoglycemic trend in the sensor for 3 days, he completely quit insulin use 2 days before the visit. The day before the visit, he felt fatigued and lost his appetite, and ate noodles and half of an apple for dinner and injected 6 units of Aspart. On the day of the visit to the hospital, he drank apple juice in the morning and injected 4 units of Aspart.

Figure 1.

The FreeStyle Libre data: 6 days prior to the visit to the hospital, he replaced the FreeStyle Libre sensor, where glucose value had been much lower than usual.

Figure 2.

Clinical course of the patient: The patient was unaware of the discrepancy between the sensor glucose value and blood glucose, judged it hypoglycemia, and gradually reduced his insulin dose. He reduced Aspart to 6 units at each meal time and discontinued Glargine to cope with the low glucose shown in the sensor 5 days before his regular visit to the hospital. As he observed no improvement of the hypoglycemic trend in the sensor for 3 days, he completely quit insulin use 2 days before the visit. The day before the visit, he felt fatigued and lost his appetite, and ate noodles and half of an apple for dinner and injected 6 units of Aspart. On the day of the visit to the hospital, he drank apple juice in the morning and injected 4 units of Aspart.

At his visit, he had no fever with clear consciousness; however, he was suffering from severe fatigue and loss of appetite. Laboratory data showed hyperglycemia (592 mg/dL), high‐anion gap metabolic acidosis (pH 7.245), and ketonuria, with elevated serum levels of β‐hydroxybutyrate (5,782 μmol/L; reference range 0–85 μmol/L) (Table 1). He was 171 cm tall, weighed 79.2 kg at the time of admission. Based on the diagnosis of DKA, he was admitted to the hospital and the standard management for DKA was initiated with intravenous fluid replacement and insulin infusion.

Table 1.

Laboratory findings (at the outpatient clinic on the day of admission)

Variable	Result	Reference range
WBC (/μL)	12,300	3,900–9,700
RBC (×104/μL)	473	430–567
Hb (g/dL)	15	13.4–17.1
Plt (×104/μL)	21.8	15.3–34.6
TP (g/dL)	7.5	6.5–8.5
BUN (mg/dL)	20	9.0–21
Cre (mg/dL)	0.89	0.6–1.0
UA (mg/dL)	8.2	3.5–6.9
Na (mmol/L)	132	133–145
Cl (mmol/L)	102	96–107
K (mmol/L)	5.1	3.5–5.0
Ca (mg/dL)	9.2	8.8–10.6
Pi (mg/dL)	3.3	2.4–4.5
AST (U/L)	17	5–37
ALT (U/L)	26	6–43
γ‐GTP (U/L)	35	0–75
TG (mg/dL)	395	30–149
HDL‐C (mg/dL)	75	40–70
LDL‐C (mg/dL)	84	70–139
Glu (mg/dL)	592	65–109
HbA1c (%)	8.5	4.6–6.2
Acetoacetic acid (μmol/L)	1,378	0–55
β‐Hydroxybutyric acid (μmol/L)	5,782	0–85
Total ketone body (μmol/L)	7,160	0–130
C‐peptide (ng/mL)	<0.02	1.1–3.3
TSH (mIU/L)	0.46	0.61–4.23
FT4 (ng/dL)	1.1	1–1.7
FT3 (pg/mL)	1.5	2.4–4.5
Anti‐GADAb (U/mL)	<5.0	0–4.9
Anti‐IA‐2 Ab (U/mL)	<0.6	0–0.5
Anti‐insulinAb (U/mL)	<0.4	0–0.39
Anti‐TPOAb (IU/mL)	<9	0–15
Anti‐TgAb (IU/mL)	14	0–27
TRAb (3rd) (IU/mL)	1.1	0–1.9
Blood gas
pH	7.245	7.35–7.45
pCO2 (mmHg)	27.2	35–45
${\mathrm{HCO}}_3^{-}$ (mmol/L)	11.5	22–26
Lactate (mmol/L)	2.6	0.5–1.4
Urinary test
pH	5.0	5.0–8.0
Specific gravity	1.036	1.011–1.03
Sugar	(4+)
Ketone	(3+)

ALT, alanine aminotransferase; Anti‐GADAb, anti‐glutamic acid decarboxylase antibody; Anti‐IA‐2Ab, anti‐insulinoma‐associated protein‐2; Anti‐TgAb, anti‐thyroglobulin antibody; Anti‐TPOAb, anti‐thyroid peroxidase antibody; AST, aspartate aminotransferase; CPR, c‐peptide immunoreactivity; FT3, free thyroxine 3; FT4, free thyroxine4; Glu, glucose; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; TG, triglyceride; TRAb; Anti‐thyrotropin receptor antibody; TSH, thyroid‐stimulating hormone; γ‐GTP, γ ‐glutamyl transpeptidase.

After admission, the sensor glucose level of his FreeStyle Libre which he had been wearing, was 40 mg/dL, while his blood glucose level obtained with SMBG was 417 mg/dL. No apparent corruption, such as sensor filament bending, was identified under careful observation of the sensor he had applied (Figure 3). The standard therapy improved his fatigue in 20 h, and he started to eat meals as usual on day 2. He was discharged on day 5 of the admission following his request to go home as soon as possible. Since the patient was almost stable on the original insulin units, rapid‐acting insulin Aspart, 22 units at breakfast, 8 units at lunch, 8 units at dinner, and long‐acting insulin Glargine, 12 units before sleep with a full intake of 2,200 kcal of regular food, we decided to have the patient go home with those units and to make adjustments on an outpatient basis.

Figure 3.

The sensor filament; after admission, the sensor glucose level of his FreeStyle Libre which he had been wearing, was 40 mg/dL, while his blood glucose level obtained with SMBG was 417 mg/dL. No apparent corruption, such as sensor filament bending, was identified under the careful observation of the sensor he had applied.

DISCUSSION

The Freestyle Libre is currently available in Japan as an isCGM for type 1 and type 2 diabetes when insulin is used at least once daily. Several studies have also noted its potential to improve glycemic control and to reduce the frequency of hypoglycemia ¹ , ² , ³ ; however, as the FreeStyle Libre indirectly estimates the real blood glucose value, its impairment could lead to severe metabolic failure. Though no cases of DKA due to the incorrect use of the FreeStyle Libre have yet been reported, it is controversial whether Freestyle Libre decreases the occurrence of DKA ⁴ , ⁵ .

The causes of false glucose value indication with the FreeStyle Libre are generally due to the sensor filament being installed improperly, not sticking in, and being disconnected in the process, or the sensor itself being defective. We requested Abbott Japan to analyze the sensor to determine the cause, but the cause of the error was unknown (Figure 3). It is thought that the device itself malfunctioned due to a problem in the manufacturing process of FreeStyle Libre. Dexcom G6 is now available in Japan as a CGM that does not require regular blood glucose monitoring and is as inexpensive as the FreeStyle Libre. The same problem may occur with either device.

Inconvenience and pain are recognized reasons for nonadherence to SMBG ⁶ , just as it was for this patient. Generally, the frequency of SMBG decreases as the patients become accustomed to the FreeStyle Libre. Moreover, this patient had not been self‐monitoring his blood glucose since he started insulin therapy. After his hospitalization, we again instructed him on the proper use of the FreeStyle Libre ⁷ , how to read data ⁸ , ⁹ , ¹⁰ , and the necessity of SMBG and its technique. He did not seem to have DKA except at the time of onset, and had never had severe hypoglycemia. The medical record showed that he rarely had hypoglycemia, and he did not seem to recognize the typical symptoms of hypoglycemia. We believe that it was useful that the patient actually experienced the dissociation of isCGM sensor values from glucose values. There is also the possibility of trusting the FreeStyle Libre data too much without questioning the accuracy of the measurements, putting the patient at risk. This patient had been using the FreeStyle Libre for 4 years without any problems and did not doubt the accuracy of the unusual waveform in the sensor.

In summary, this is the first report to our knowledge of a patient with DKA caused by the failure of FreeStyle Libre. We should remember that sensor glucose values may deviate significantly from actual blood glucose values, leading to a serious condition. We also need to create an environment in which we provide specific education to the patients and train staff.

DISCLOSURE

Junko Sato has received honoraria for lectures for Abbott. Hirotaka Watada has received honoraria for lectures for Bayer Pharma Japan, Teijin Pharma Ltd, MSD, Sanofi‐Aventis K.K., Novo Nordisk, Nippon Boehringer Ingelheim, Eli Lilly, Sumitomo Pharma, Mitsubishi Tanabe Pharma, Daiichi Sankyo Company, Ltd, Abbott, Kowa Co., Ltd, Taisho Pharmaceutical, Astellas Pharma, Kissei Pharmaceutical Co., Ltd, AstraZeneca K.K., Ono Pharmaceutical Co. Ltd, Sanwa Kagaku, Takeda Pharmaceuticals, and research activities for Takeda Pharmaceuticals, Nippon Boehringer Ingelheim, Kissei Pharmaceutical, Novo Nordisk, Mitsubishi Tanabe Pharma, Lifescan Japan, Kyowa Kirin, Sumitomo Pharma, Eli Lilly, Teijin Pharma, Taisho Pharmaceutical, Abbott, Ono Pharmaceutical Co. Ltd, Soiken Inc., Sanwa Kagaku, and Kowa. The other authors declare no conflict of interest. Hirotaka Watada is an Editorial Board member of Journal of Diabetes Investigation and a co‐author of this article. To minimize bias, they were excluded from all editorial decision‐making related to the acceptance of this article for publication.

Approval of the research protocol: N/A.

Informed consent: Written informed consent was obtained from the patient.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.