Insulin therapy in acute decompensation of holocarboxylase synthetase deficiency with hyperglycemia and ketoacidosis

Abstract

An 11-month-old girl with severe acidosis, lethargy and vomiting, was diagnosed with holocarboxylase synthetase deficiency. She received biotin and was stable until age 8 years when vomiting, severe acidosis, hypoglycemia, and hyperammonemia developed. Management with intravenous glucose aiming to stimulate anabolism led to hyperglycemic ketoacidosis. Insulin therapy rapidly corrected biochemical parameters, and clinical status improved. We propose that secondary Krebs cycle disturbances affecting pancreatic beta cells impaired glucose-stimulated insulin secretion, resulting in insulinopenia.

Graphical abstract

Highlights

• •Holocarboxylase synthetase deficiency (HLCSD) is a treatable metabolic disease.
• •Hyperglycemic ketoacidosis is a rare but recurrent presentation of HLCSD.
• •In an HLCSD patient with hyperglycemia and ketoacidosis, insulin therapy (0.01 U/kg/h) was associated with rapid recovery.
• •In HLCSD patients with hyperglycemia, beta cell dysfunction should be considered, and insulin levels should be measured.

1. Introduction

Holocarboxylase synthetase (HLCS, EC 6.3.4.10) deficiency (HLCSD, MIM #253270), first described by Roth et al. [1], is a rare autosomal recessive inborn error of metabolism (IEM) affecting the covalent binding of biotin to the four apocarboxylases (i.e. pyruvate (PC, EC 6.4.1.1), propionyl-CoA (PCC, EC 6.4.1.3), acetyl-CoA (EC 6.4.1.2), and 3-methylcrotonyl-CoA (EC 6.4.1.4) carboxylases) leading to multiple carboxylase deficiency (MCD) [2]. More than half of HLCSD patients present within hours to days after birth, with lethargy, seizures, vomiting, and tachypnea secondary to severe metabolic acidosis with an elevated anion gap, suggestive of organic acidurias [3]. A late-onset form of the disease, manifesting up to 8 years of age by an acute decompensation following catabolic stress, has also been reported [[4], [5], [6]]. Some symptomatically-treated patients who do not receive biotin treatment may additionally develop psychomotor retardation, hair loss, and skin lesions [3,7]. Biochemical diagnosis is made based on a urine organic acid profile compatible with MCD, showing elevated 3-hydroxyisovalerate, 3-hydroxypropionate, propionylglycine, tiglylglycine, 3-methylcrotonylglycine, methylcitrate, lactate and pyruvate. HLCS gene sequencing allows diagnosis confirmation and genetic counselling [8]. When started before irreversible neurological damage has occurred, treatment with oral biotin at pharmacological doses prevents further decompensation in most cases [9,10].

Hyperglycemic ketoacidosis, mimicking diabetic ketoacidosis, is a recognized presentation of several organic acidurias, including HLCSD [[11], [12], [13]]. Recurrent episodes of pancreatitis have been proposed as an explanation for a small fraction of the episodes of hyperglycemic ketoacidosis that occur in classic organic acidemias [14,15]. Pancreatitis has not previously been reported in HLCSD patients. The precise mechanisms underlying hyperglycemic ketoacidosis in HLCSD are yet to be discovered.

We report on a previously well-controlled late-onset HLCSD patient presenting hyperglycemic ketoacidosis during her first acute decompensation since her initial diagnostic episode.

2. Case report

The diagnosis of HLCSD was made in an 11-month-old girl presenting lethargy, tachypnea, and signs of dehydration following 2 episodes of vomiting. She was born of non-consanguineous parents after an uneventful pregnancy. The biochemical features are presented in Table 1. Serum biotinidase activity was normal (158 nmol/s/L, normal (N) > 69). Molecular testing revealed compound heterozygosity for HLCS (NM_001242784) variants c.848delG, p.(Ser283Thrfs*3) (inherited from her mother) and c.2126C>T, p. (Pro709Leu) (not inherited from her mother, father unavailable for testing). She was treated with biotin 2.5 mg/kg/d and carnitine 50 mg/kg/d and developed normally with no acute metabolic decompensation until the episode described here. She attends regular classes at an age-appropriate level.

Table 1.

Biochemical parameters measured at pertinent timepoints.

	Normal values	Units	Diagnosis								Admission	Insulin start	Insulin stop	Discharge	Last follow-up
			11 m	2 y 4 m	3 y 6 m	4 y 5 m	5 y 4 m	6 y 4 m	7 y 2 m	8 y 3 m	8 y 5 m	+ 10 h	+ 30 h	+ 59 h	8 y 8 m
Blood sample
pH	7.34–7.44		6.95	7.43	7.39	7.40	7.40	7.39	7.39	7.36	6.90	6.98	7.32	7.43	7.39
pCO2	37–47	mmHg	9.4	40.9	39.4	42.3	39.8	42.2	42.6	45.6	21.3	22.5	37.4	40.5	42
HCO3−	22–28	mM	2	26.5	23.4	25.5	24.1	25	24.9	24.9	4	5.1	18.8	26.6	24.7
Lactate	0.6–3.2	mM	8.1	1.9	1	1.8	1.9	1.4	1.6	3.7	17	10.4	1.7	1.6	2.7
Glucose	4.1–5.9	mM	5.8	4	5.9	4.7	6.2	4.4	4.4	4.3	2.1	17	6.6	5.5	4.8
3-OH-butyrate	<0.6	mM	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	n.a.	2	1.9	0.2	0.2	n.a.
Anion gap (with K+)	12–20	mEq/L	30	10.5	8.1	12.6	12.9	13.2	14.5	13.3	42.8	21.2	7.6	9.5	14.2
Ammonia	5–55	μM	161	n.a.	42	28	29	34	23	24	119	114	36	37	24
Propionylcarnitine (C3)	0.12–0.69	μM	7.96	0.69	n.a.	0.73	0.51	0.42	0.45	5.1	16	n.a.	n.a.	n.a.	5.42
Tiglylcarnitine (C5:1)	0.01–0.03	μM	0.06	0.01	n.a.	0	0.01	0.01	0.01	0.05	0.23	n.a.	n.a.	n.a.	0.02
3-OH-isovalerylcarnitine (C5OH)	0.03–0.11	μM	2.7	0.01	n.a.	0.08	0.02	0.02	0.02	0.85	3.24	n.a.	n.a.	n.a.	0.83
Urine sample
Lactate	25–134	mmol/mol Cr	42.771	31	12	n.a.	25	13	66	580	10,557	n.a.	n.a.	n.a.	42
Pyruvate	4–17	mmol/mol Cr	693	11	3	n.a.	3	4	3	93	1246	n.a.	n.a.	n.a.	19
Acetoacetate	<6	mmol/mol Cr	50,401	0	0	n.a.	3	0	1	1302	21,069	n.a.	n.a.	n.a.	2
3-OH-butyrate	<11	mmol/mol Cr	108,750	8	2	n.a.	8	7	3	165	5698	n.a.	n.a.	n.a.	5
Methylcitrate	1–5	mmol/mol Cr	222	0	8	n.a.	8	1	2	6	16	n.a.	n.a.	n.a.	7
3-OH-isovalerate	10–50	mmol/mol Cr	11,498	16	6	n.a.	8	0	17	740	3512	n.a.	n.a.	n.a.	376
3-OH-propionate	1–36	mmol/mol Cr	2636	19	6	n.a.	5	4	8	100	189	n.a.	n.a.	n.a.	50
3-methylcrotonylglycine	<0.9	mmol/mol Cr	1035	0	0	n.a.	0	0	0	43	6224	n.a.	n.a.	n.a.	0.3
Propionylglycine	<1	mmol/mol Cr	765	0	0	n.a.	0	0	0	9	1069	n.a.	n.a.	n.a.	0
Tiglylglycine	<1	mmol/mol Cr	3113	2	0	n.a.	0	0	0	73	192	n.a.	n.a.	n.a.	0

After diagnosis and treatment with oral biotin, the biochemical parameters nearly normalized until 2 months before the admission for severe decompensation. Retrospectively, her metabolic control was suboptimal at the outpatient visit (8 y 3 m) a few weeks before the episode, presumably reflecting a lack of compliance with biotin treatment.

Abnormal values are shown in bold. +10 h, +30 h and + 59 h refer to the number of hours since hospital admission. m: months, y: years, h: hours, n.a.: not available.

At age 8, she presented with stupor, feeding intolerance, and signs of dehydration. For 12 h before admission, she vomited all medications and had abdominal pain, headache, tachycardia, and tachypnea. She was normotensive and afebrile, with no diarrhea. No lack of compliance with treatment preceding the acute presentation was disclosed. Point-of-care testing revealed severe lactic and ketoacidosis with increased anion gap (Table 1, Fig. 1). Marked neutrophilic leukocytosis associated with mild inflammation (leucocytes 29.4 × 10^9/L (N 4.5–13.5), neutrophils 15.6 × 10^9/L (N 1.5–7.3), C-reactive protein 6.1 mg/L (N0−1)) initially suggested a bacterial infection. Urine and blood cultures were sterile, and the nasopharyngeal swab polymerase chain reaction detected non-SARS-coronavirus-2. Plasma pancreatic and liver enzyme levels were normal, as was an abdominal ultrasound. In the pediatric intensive care unit, she was managed with fluid replacement, dextrose 10% with normal saline (D10%-NS) (maximum 6.9 mg glucose/kg/min), oral biotin 3 mg/kg/d, l-carnitine 100 mg/kg/d IV q3h and ondansetron 0.1 mg/kg q8h. Progressive hyperglycemia (peak 21.8 mM after 4 h of D10%-NS) was treated with insulin for 20 h (maximum 0.08 U/kg/h). The metabolic profile improved under insulin therapy, with normalization of anion gap (2 h post-insulin initiation), lactate (7 h), 3-hydroxybutyrate (7 h), ammonia (13h), acidosis (28 h) and bicarbonate (28 h). Bicarbonate therapy was not employed to manage the patient given the demonstrated rapid response to insulin.

Fig. 1.

Course of metabolic acidosis and blood glucose in relation to insulin infusion rate. pH normalization over time compared with the dose of insulin therapy (A) and glucose infusion rate (B). A significant correlation was detected between the speed of acidosis correction (ΔpH/h) and insulin infusion rate (C) but not glucose infusion rate (D). Lactate (E), 3-OH-butyrate (F), glycemia (G), and ammonia (H) correction over time (normal value range in grey) with dose of insulin therapy as comparator.

HLCSD causes secondary PC deficiency. The lactic and ketoacidoses of PC deficiency are typically suppressed by the administration of glucose, which removes the need for gluconeogenesis, thus reducing the metabolic flux directed towards PC. In this patient, we hypothesize that insulinopenia prevented cell entry and utilization of glucose despite hyperglycemia, prolonging demand for PC-mediated gluconeogenesis, reducing lactate removal, and enhancing fatty acid oxidation and ketogenesis, which resolved rapidly with insulin therapy (Fig. 2).

Fig. 2.

Insulin therapy inhibited pyruvate carboxylase (PC)-mediated gluconeogenesis (GNG) presumably triggered by insulinopenia secondary to holocarboxylase synthetase deficiency (HLCSD). Treatment with a high glucose infusion rate (up to 6.9 mg glucose/kg/min, in green) led to hyperglycemia (peak 21.8 mM). We hypothesize that this hyperglycemia resulted from pancreatic beta cell insufficiency with impaired glucose-stimulated insulin secretion (GSIS, dashed red arrow) and non-suppression of GNG (dashed red line) despite hyperglycemia. Impaired GSIS might be secondary to PC and propionyl-CoA carboxylase deficiency in beta cells (see Discussion and Fig. 3). Insulin therapy (in orange) allowed for normal suppression of lipolysis (triglyceride (TG) hydrolysis with release of fatty acids (FA)), of ketogenesis and of GNG, correcting the hyperlactatemia and ketoacidosis. KG: ketogenesis; LDH: lactate dehydrogenase; PDH: pyruvate dehydrogenase; βOX: beta-oxidation.

3. Discussion

We report on a previously well-controlled HLCSD patient who developed acute, severe lactic and ketoacidosis following a brief vomiting episode. The biochemical profile was compatible with untreated MCD; the family denied reduced adherence to the longstanding prescribed biotin therapy. Two months prior to admission moderate increases of several biomarkers of MCD were noted at an outpatient visit when the child was clinically well (Table 1). This suggests that some unidentified factor may have modulated the oral biotin dose efficacy or perhaps the presence of undeclared non-adhesion to biotin supplementation. Management with fluid replacement and intravenous glucose alone led to slow (<0.01 pH unit/h) and partial correction of the acidosis and hyperglycemia. The administration of insulin resulted in rapid correction of the acidosis (up to 0.04 pH unit/h) and of other biochemical markers (Fig. 1). There was a significant correlation between the rate of resolution of the acidosis and the rate of insulin infusion (Fig. 1B) but no such correlation was detected with glucose infusion rate (Fig. 1D).

Two HLCSD patients have previously been reported to be compound heterozygotes for c.2126C>T, p.(Pro709Leu) and another variant in HLCS [16,17]. One was diagnosed at 3 months of age (no phenotype available), with the second HLCS variant being c.1544G>A, p.(Ser515Arg) (no data available on residual enzyme activity) [16]. The second patient was developing in accordance with his age at 4 years [17]. He carried the second HLCS variant c.1533dupT, p.(Val512CysfsTer65), which is predicted to disrupt the enzyme's catalytic site (residues 463 to 652), leading to absent residual enzymatic activity. All 3 patients who are genetic compounds for c.2126C>T have had late clinical onset ([16,17] and this report). Intriguingly, two have frameshift variants on the opposite HLCS allele ([17] and this report). Together, these observations suggest that c.2126C>T, p.(Pro709Leu) may have residual enzyme activity that confers a late-onset biotin-responsive phenotype.

Normally, glucose-stimulated insulin secretion (GSIS) by pancreatic beta cells precisely regulates blood insulin levels. In beta cells, glucose degradation via glycolysis and the Krebs cycle increases the intracellular ATP/ADP ratio, triggering cell depolarization and insulin release [18]. Consideration of the location and properties of the carboxylases that are deficient in MCD, particularly PC and PCC, suggests that they will exert synergistic effects in the decompensations of HLCSD. PC has been shown to play a major role in GSIS as suggested by (1) the presence of a glucose-responsive element in the PC gene promoter [19], (2) the rapid increase in PC protein level in pancreatic islets in response to glucose [20], (3) the decreased GSIS following PC suppression [21], and (4) its increase following PC overexpression [22]. Although the precise mechanisms are not documented, PC function is necessary for insulin secretion stimulated by glucose and other secretagogues [21]. In addition, reduced anaplerosis (low oxaloacetate synthesis) secondary to PC deficiency (PCD) is predicted to deplete the Krebs cycle in beta cells, impeding GSIS (Fig. 3). Some reported PCD patients have mimicked diabetic ketoacidosis at their initial presentation, and have been treated with insulin (dose not reported) [23,24]. These observations support the hypothesis that insulinopenia might be clinically relevant in PCD patients in acute decompensation, and by extension, in symptomatic HLCSD patients with secondary PCD.

Fig. 3.

Holocarboxylase synthase deficiency (HLCSD) may impact the Krebs cycle homeostasis at three crucial points. We postulate a triple insult (in orange) to the Krebs cycle in HLCSD driven by deficiencies of pyruvate carboxylase (PC) and propionyl-CoA carboxylase (PCC). This could impair glucose-induced insulin secretion in pancreatic beta cells, leading to insulinopenia during acute HLCSD decompensation. Deficiencies of PC and PCC would decrease the supply of oxaloacetate and succinyl-CoA, respectively (reduced anaplerosis), and accumulated propionyl-CoA would capture oxaloacetate as methylcitrate, which is lost by excretion (cataplerosis). During hypoglycemia, in gluconeogenic cells like hepatocytes, oxaloacetate would also be required for gluconeogenesis, producing an additional cataplerotic drain.

Some patients with primary PCC deficiency (PCCD), have brittle glucose homeostasis, developing hyperglycemia or hypoglycemia depending upon glucose intake, for which the underlying mechanisms are still poorly understood [13]. In PCCD, we hypothesize that reduced anaplerosis of the Krebs cycle due to low succinyl-CoA synthesis and cataplerosis of oxaloacetate moieties, through methylcitrate synthesis and excretion, will impair the Krebs cycle (Fig. 3 and S1 Fig.). In pancreatic beta cells, this could reduce GSIS, leading to insulinopenia and hyperglycemia. Lehnert et al. [12] were the first to report a newborn diagnosed with PCCD presenting severe hyperglycemia. Since then, a few PCCD patients have been reported with diabetic ketoacidosis-like presentation that responded to low insulin doses (0.02–0.1 U/kg/h), suggesting insulinopenia [25,26]. Acute management of PCCD patients comprises intravenous glucose, to which insulin (0.01–0.02 U/kg/h) has been added to enhance anabolism while maintaining normoglycemia [13,27]. Other PCCD patients with hyperglycemia have required higher doses of insulin (0.3–1.4 U/kg/h), compatible with insulin resistance [[28], [29], [30]]. Our patient (max insulin doses 0.08 U/kg/h) received low-dose insulin therapy, supporting the hypothesis of insulinopenia following Krebs cycle insult in pancreatic beta cells of PCCD and, by extension, in HLCSD patients.

Glutamate dehydrogenase (GDH) catalyzes a reversible reaction, converting L-glutamate into alpha-ketoglutarate, which provides an additional anaplerotic source, independent of PC and PCC. GDH activity drives insulin release in pancreatic beta cells, justifying its tight regulation. ADP and leucine are potent allosteric activators, whereas palmitoyl-CoA, ATP, and GTP inhibit the GDH activity [31]. We postulate that in HLCSD, the Krebs cycle is impaired in pancreatic beta cells, leading to a low ATP/ADP ratio and insulinopenia during acute decompensations. GDH should theoretically be activated by the low ATP/ADP ratio and play an anaplerotic role in the Krebs cycle during these acute HLCSD decompensations. Alpha-ketoglutarate urine levels measured during the two decompensations of our patient (at diagnosis and during the present episode) are lower than during routine outpatient visits (Fig. S1), and thus, at least at the whole body level, it does provides no evidence that GDH compensates adequately during the acute crises of HLCSD patients. In addition, intra-mitochondrial palmitoyl-CoA accumulation secondary to Krebs cycle impairment (secondary fatty acid oxidation deficiency) could inhibit GDH and impair its anaplerotic effect.

Hyperglycemic ketoacidosis is a recurrent complication of HLCSD. In addition to our patient, three other late-onset HLCSD patients presented with hyperglycemia (15.2 to 32.9 mM) and ketoacidosis, initially suggesting diabetes type 1 [32,33]. One other patient received insulin (dose not reported). We postulate that insulinopenia secondary to altered GSIS following Krebs cycle disturbances in beta cells might be the underlying cause. Measurement of plasma insulin level in HLCSD patients at times of abnormal glucose homeostasis has not been reported but could help confirm (or dismiss) this hypothesis.

A limitation of the study is that it describes only one patient. However, taken together with the available literature, which shows that hyperglycemia with ketoacidosis has occurred repeatedly in late-onset HLCSD, it suggests a plausible and testable hypothesis. Insulin level was not quantified in this patient and insulin secretion has not been studied in detail in HLCSD, PCD, or PCCD patients during decompensation. Discussion of the underlying pathophysiology is therefore speculative but it identifies a potentially important and treatable mechanism and invites specific recommendations for future cases: in HLCSD patient in acute decompensation, we propose close surveillance of blood glucose levels. If hyperglycaemia develops, plasma insulin and glucose levels should be documented and low dose insulin infusion should be considered.

4. Conclusion

A HLCSD patient presenting severe lactic and ketoacidosis successfully managed with insulin therapy led us to hypothesize insulinopenia as a possible complication of HLCSD decompensation. We propose starting insulin therapy at a low dose (0.01 U/kg/h) to promote anabolism during HLCSD decompensation. Further research is required to decipher the pathophysiology underlying the brittle glucose metabolism in HLCSD and organic acidurias.

Funding

TD was supported by the SofinaBoël Fund for Education and Talent, managed by the King Baudouin Foundation, and Wallonie-Bruxelles International. The funding sources had no role in the study design, in the collection, analysis, and interpretation of data, in the report's writing, and in the decision to submit the article for publication.

CRediT authorship contribution statement

Tanguy Demaret: Writing – review & editing, Writing – original draft, Funding acquisition, Formal analysis, Data curation, Conceptualization. Jean-Sébastien Joyal: Writing – review & editing, Supervision, Methodology. Aspasia Karalis: Writing – review & editing, Supervision, Methodology. Fabienne Parente: Writing – review & editing, Supervision, Formal analysis. Marie-Ange Delrue: Writing – review & editing, Supervision, Methodology. Grant A. Mitchell: Writing – review & editing, Writing – original draft, Supervision, Methodology, Formal analysis, Conceptualization.

Declaration of competing interest

None.

Appendix A. Supplementary data

Supplementary material - Comparison of Krebs cycle intermediates (citrate, alpha-ketoglutarate, succinate, fumarate and malate) urine levels during acute episodes and routine follow-up.

Data availability

Data will be made available on request.