Type 1 Diabetes Mellitus with Diabetic Ketoacidosis Presenting with Transient Severe Fatty Liver and Subsequent Liver Dysfunction

Abstract

Patients often present with severe fatty liver (FL) due to insulin deficiency at the onset of diabetic ketoacidosis (DKA). On the other hand, glycogenic hepatopathy (GH) is a possible cause of liver dysfunction in patients with DKA. We herein report a case of type 1 diabetes mellitus with severe FL at the onset of DKA, who demonstrated subsequent marked liver dysfunction after achieving an improvement of FL. As liver dysfunction persisted even after the FL improved, GH was suspected to be the cause of liver dysfunction. FL and GH have different prognoses and should therefore be differentiated using imaging studies and biopsies.

Introduction

Insulin is responsible for inhibiting the breakdown of adipocytes. In situations where insulin is deficient, free fatty acids increase as a result of accelerated triglyceride (TG) breakdown in peripheral adipose tissue, and a decrease in very low-density lipoprotein (VLDL) secretion from the liver occurs, thus leading to TG accumulation in the liver (1-5). Other causes of liver dysfunction other than fatty liver (FL) during diabetic ketoacidosis (DKA) treatment include glycogenic hepatopathy (GH) due to the rapid accumulation of glycogen in the liver (6,7). We herein report a case of type 1 diabetes mellitus (T1DM) with transient severe FL at the onset of DKA who demonstrated subsequent prominent liver dysfunction after achieving an improvement in the FL.

Case Report

A 28-year-old woman was diagnosed with T1DM at 22 years of age, after which she started intensive insulin therapy. After starting treatment, her HbA1c level was maintained at approximately 7%. She had no medical history other than T1DM and never presented with liver dysfunction. Two days before admission, she developed nausea and vomiting, was unable to eat, and had interrupted insulin treatment. Two days later, she was found to have an impaired consciousness and was admitted to the hospital. The patient was unconscious, had tachycardia and tachypnea (Table 1), and was immediately placed on a ventilator. Laboratory data (Table 2) showed metabolic acidosis (pH 7.214, HCO₃^- 8.5 mEq/L) associated with a high level of glucose (750 mg/dL) and a high level of ketone body (12,719 μmol/L), compatible with DKA. Abdominal computed tomography (CT) revealed a severe FL (Fig. 1). She was treated with a continuous intravenous injection of regular insulin, along with a massive infusion of saline. Four days after the successful treatment of DKA, she was able to ingest food under treatment with multiple injections of insulin aspart and glargine. Eleven days after hospitalization, her transaminase levels showed sudden elevation [Aspartate transaminase (AST)/Alanine transaminase (ALT) 563/523 U/L] with no laboratory abnormalities in the coagulation system or any obvious symptoms. A CT scan was performed again on day 13 to identify the cause of the liver dysfunction. The FL at the time of admission improved. The longitudinal meridian of the liver was 15.4 cm, with slight hepatomegaly. Additional laboratory tests revealed no abnormal findings in viral hepatitis serology, anti-mitochondrial antibody, anti-smooth muscle antibody, and antinuclear antibody. Thereafter, her AST and ALT levels showed a gradual downward trend, and the patient was discharged on day 16. An improvement in liver function was confirmed 8 days after discharge (Fig. 2).

Table 1.

Vital Signs and Physical Exam of Patient on Admission.

Vital signs
Glasgow Coma Scale	10 (E2V3M5)
Body temperture	37.1	°C
Blood pressure	103/60	mmHg
Pulse rate	124	bpm, regular rhythm
Respiratory rate	34	/min
SpO2	100	% (O2 10 L/min)
Physical exam
Height	160	cm
Weight	50.7	kg
Head
Palpebral conjunctiva	Not anemic
Bulbar conjunctiva	Not icteric
Neck
No thyromegaly, no lymphadenopathy
Chest
Heart sound	S1→S2→S3(-)S4(-), no murmur
Respiratory sound	Clear to auscultation bilaterally, no rales
Abdomen
Bowel sound	Normal
Soft & flat, no tenderness, no mass, no abdominal bruit

Table 2.

Laboratory Data of Patient.

Arterial blood gas (artificial respiration)			Blood chemistry
pH	7.214		Total protein	7.1	g/dL
PO2	637	mmHg	Albmin	4.4	g/dL
PCO2	21.8	mmHg	Aspartate transaminase (AST)	39	U/L
HCO3-	8.5	mmol/L	Alanine transaminase (ALT)	27	U/L
Anion gap	20.5	mEq/L	γ-glutamyl transpeptidase	12	U/L
urinary qualitative (day 2)			Total bilirubin	1	mg/dL
Urine glucose	-		Urea nitrogen	54	mg/dL
Urine protein	+/-		Creatinine	1.74	mg/dL
Urine ketone body	2+		Sodium	132	mEq/L
Complete blood count			potassium	6.1	mEq/L
White blood cell	32,310	/μL	Chlorine	91	mEq/L
Red blood cell	405	×104/μL	Creatine kinase	285	U/L
Hemoglobin	12.2	g/dL	Hemoglobin A1c (NGSP)	7.4	%
Hematocrit	38.9	%	Blood glucose	750	mg/dL
Platelet	25.4	×104/μL	Total cholesterol	102	mg/dL
Neutrophil	86.2	%	Triglyceride	64	mg/dL
Lymphocyte	6.9	%	β-Hydroxybutyric acid	12,719	μmol/L
Monocyte	6.6	%	C-peptide	0.07	ng/mL
Eosinophil	0.1	%	Immunological test (day 14)
Basophil	0.2	%	Anti-mitochondrial antibody	Negative
			Anti-smooth muscle antibody	Negative
			Anti-nuclear antibody	Negative

Figure 1.

(A) Computed tomography scans on admission showing severe FL. (B) Computed tomography scans on day 13 showing the improvement of FL.

Figure 2.

Changes in Blood Glucose, β-Hydroxybutyric acid, AST, and ALT. A sharp rise in AST and ALT was observed on the 11th day. A CT scan was performed again on the 13th day to search for the cause, but the FL findings seen on admission had improved (Fig. 1B). AST and ALT showed a decreasing trend, so the patient was discharged on the 16th day. Eight days after discharge (on the 24 day), an improvement to within reference values was observed.

Discussion

Takaike et al. (8) reported that 22.6% of patients with fulminant T1DM and 24.4% of patients with acute-onset T1DM were diagnosed with FL during treatment for diabetic ketosis or DKA, and the prevalence of FL was similar in both groups.

TG accumulation in the liver is defined by the balance between dietary or adipose tissue-derived free fatty acid (FFA) influx via the portal vein, TG synthesis in the liver (de novo lipogenesis), oxidation of FFA in the liver, and the synthesis, secretion, and degradation of VLDL. Each of these processes involves insulin (9). Insulin inhibits adipocyte degradation, and a lack of insulin accelerates free fatty acid influx into the liver (10), thus resulting in de novo TG synthesis in the liver (1). TG synthesized in the liver is associated with apoB proteins and released from the liver as VLDL (2). However, rapid insulin depletion causes a rapid and excessive influx of FFA and glucose into the liver, thus resulting in a relative deficiency of the apoB protein and endoplasmic reticulum stress. High levels of endoplasmic reticulum stress result in the degradation of apoB protein and weakening of microsomal triglyceride transfer protein activity, leading to decreased VLDL secretion from the liver (3). When insulin action is attenuated, TG accumulates in the liver through these mechanisms, thus leading to the development of a FL.

In this case, the patient had a severe FL on admission, which may have been caused by the interruption of insulin injections, which accelerated the accumulation of TG in the liver. Even though insulin therapy was successfully performed, liver enzymes showed a rapid increase on day 11. Considering that the FL had already improved on imaging, we speculated that another factor was associated with the liver dysfunction on day 11.

GH, characterized by elevated liver enzymes and hepatomegaly, is caused by the excessive accumulation of glycogen in hepatocytes due to hyperglycemia and rapid insulin influx, which is sometimes found in T1DM (4). In this case, since an elevation in the liver enzyme level was observed after the improvement of FL following insulin therapy, GH was considered to be the cause of the increased transaminase levels.

As the blood glucose level reaches a certain threshold, glucose is rapidly phosphorylated to glucose-6-phosphate by the hepatic hexokinase isoform glucokinase, and then glucose-6-phosphate is converted to glycogen by glycogen synthase. Thus, a hyperglycemic state with poorly controlled T1DM results in excessive glycogen synthesis. In addition, patients with poorly controlled T1DM often develop hypoglycemia by taking excess insulin and treat this state by taking glucose, which is considered to be a mechanism of overglycogenosis (5).

Regarding the prognosis, GH is quite different from nonalcoholic fatty liver disease (NAFLD) or FL. Unlike most cases of NAFLD, GH can be reversed after an improvement in glycemic control and will not recur if adequate glycemic control is maintained (6,7). In contrast, NAFLD leads to nonalcoholic steatohepatitis or progresses to advanced liver cirrhosis or hepatocellular carcinoma (11). Therefore, making an appropriate diagnosis is important. However, FL and GH are often difficult to differentiate. Ultrasonography, CT and magnetic resonance imaging (MRI) have been used to differentiate between these two different disease states (12).

Ultrasonography: Both diseases show hyperechogenicity compared with the right kidney. Hepatomegaly was also found in both diseases but more frequently in GH.

CT: NAFLD is hypodense compared to the spleen. GH is hyperdense compared to the spleen, but it may not always be hypodense when FL coexists.

MRI: NAFLD shows differences in the intensities of T1 weighted gradient-dual-echo MRI images with in-phase and opposed-phase conditions, indicative of steatosis. In the GH group, this difference does not occur.

In addition to imaging analyses, a liver biopsy is the gold standard for differentiating between these two diseases. A histological examination of a liver biopsy specimen from a patient with NAFLD revealed macrovesicular steatosis, mild lobular and portal inflammation, and varying degrees of fibrosis. On the other hand, a histological examination of hepatocytes from patients with GH normally does not reveal any significant portal inflammation, steatosis, or significant fibrosis, and the characteristic histologic changes seen in GH include abundant cytoplasmic glycogen deposits within hepatocytes on periodic acid-Schiff. Hepatocytes become “ghost cells” during the diastase digestion of glycogen. Similar findings have been observed in glycogen storage diseases, which should be clinically ruled out (13).

We reviewed 16 patients with T1DM who developed GH during DKA treatment (Table 3); 15 of the 16 cases had poorly controlled T1DM (14-28) and another had fulminant T1DM (29). All 12 patients who underwent abdominal ultrasonography had hepatic hyperechogenicity at GH onset. An abdominal CT scan was performed in 9 cases, all of which showed hepatomegaly, with a low density in 3 cases, high density in 2 cases, and normal density in 1 case. An MRI scan was performed in 4 cases, with no difference in the intensities of the T1 weighted gradient-dual-echo MRI images and dual-echo MRI images with in-phase and opposed-phase MRI. An elevation of the transaminase levels was reported in 13 cases, 8 of which had a delayed transaminase elevation after treatment with DKA. In cases of fulminant T1DM, imaging studies were performed on admission and at the onset of liver dysfunction. In that case, the CT images on admission showed marked hypodensity in the liver, consistent with the images of FL, but the MRI images performed at the onset of liver dysfunction were incongruent with fat accumulation in the liver and showed findings similar to those in our case.

Table 3.

Type 1 Diabetes Mellitus Patients with Glycogenic Hepatopathy.

Age/Sex	Type	Diabetic control	Ultrasound		CT		MRI (Difference in the intensities of T1 weighted gradient-dual-echo images with in-phase and opposed-phase)	Day of admission with elevation in transaminase	Lactate elevation	Ref.
			Hepatomegly	Echogenicity	Hepatomegaly	Dense
15 F	1	Poor	+	Hyper				Day 1		(14)
19 F	1	HbA1c 12.1%			+				+	(15)
19 F	1	HbA1c 12.9%	+	Hyper				Day 1		(16)
15 M	1	HbA1c 12.4%	+	Normal				Day 4		(17)
19 F	1	HbA1c 11.3%	+	Hyper					+	(18)
23 M	1	Poor	+	Normal	+	Hyper	No difference	Day 2		(19)
20 F	1	HbA1c 13.5%	+	Hyper	+	Low			+	(20)
18 M	1	Poor	+	Hyper				Day 2		(21)
19 M	1	HbA1c 14.0%	+	Hyper	+	Low		Day 1		(22)
23 F	1	HbA1c 12.3%			+	Normal	No difference	Day 2		(23)
16 F	1	HbA1c 11.5%	+	Hyper	+	Hyper		Day 1		(24)
19 F	1	HbA1c 14.6%					No difference	Day 1		(25)
17 F	1	HbA1c 14.0%	+	Hyper	+			Day 2		(26)
33 F	1	Poor		Hyper	+			Day 1		(27)
16 F	1	HbA1c 12%	+					Day 1	+	(28)
21 M	Fulminant type 1	HbA1c 6.2%		Hyper	+	Low	Subtle difference	Day 12		(29)

In cases of DKA with poorly controlled T1DM, GH develops within a couple of days of DKA treatment. In contrast, in the case of fulminant T1DM, GH developed 12 days after the DKA treatment. Patients with poor diabetes control tended to have earlier increases in the transaminase levels after DKA treatment than those whose blood glucose levels were under control. We hypothesized that glycogen accumulation originally occurred in poorly controlled cases, so it did not take a long time for glycogen accumulation to elevate the transaminase levels. Because glycemic control in this case prior to the onset of DKA was not poor, we speculated that it took time for the transaminase levels to increase in this case.

Most of the GH cases reviewed were young-onset cases. In previous studies, Bak et al. reported that sex differences were not found in the onset of GH (30). In terms of age, no previous studies have examined the relation between age and onset of GH. However, in elderly individuals, GH may sometimes be misdiagnosed as NAFLD.

Our study focused on GH that occurred during DKA treatment. GH has been reported not only in T1DM with long-term poor control but also in type 2 diabetics who received high-dose insulin in a suicide attempt followed by 3 days of high-calorie infusions, infants with dumping syndrome, and non-diabetic infants receiving high-dose glucocorticoids (29). We speculate that exposure to insulin, enormous amounts of glucose, and fluctuations in their levels are risk factors for GH development.

In the present case, a CT scan performed on the 13th day of the disease showed that the FL had improved, and hepatomegaly was observed. Therefore, GH may have been the cause of tardive liver dysfunction in this case.

One limitation associated with this case is that we did not evaluate fat deposition in the liver using the dual-echo technique on MRIT1-weighted images and glycogen accumulation by liver biopsy.

Although there have been many reports of GH in patients with poorly controlled T1DM, this is the first case report of GH in a patient who controlled T1DM properly and developed GH because of a lack of treatment for a short period of time. The mode of GH onset in this case was similar to that in previous cases of fulminant T1DM, and we hypothesized that glycemic control prior to the onset of DKA was associated with the mode of GH onset. Despite appropriate control of T1DM, GH may develop approximately 10 days after DKA treatment in patients with acute insulin deficiency. When patients present with delayed liver dysfunction after DKA treatment, we recommend considering GH as the possible cause of liver dysfunction.

The authors state that they have no Conflict of Interest (COI).