Late-onset autoimmune diabetes in 2 octogenarians: diagnostic misclassification and therapeutic heterogeneity

Abstract

Latent autoimmune diabetes in adults (LADA) may be misclassified as type 2 diabetes mellitus (T2DM) and remain unrecognized without targeted evaluation. We describe 2 octogenarians with LADA initially managed as T2DM. The first, an 84-year-old man with chronic kidney disease and normal body mass index, developed glycemic variability with hypoglycemia on multiple daily insulin injections and was intolerant of noninsulin therapies. Evaluation revealed positive pancreatic autoantibodies and preserved C-peptide of 5.68 ng/mL (SI: 1.89 nmol/L) (reference range, 1.1-5.5 ng/mL [SI: 0.37-1.83 nmol/L]). Following transition to automated insulin delivery (AID), time in range (TIR; glucose 70-180 mg/dL [SI: 3.9-10.0 mmol/L]) improved from 33% to 75% and glucose management indicator (GMI) from 8.4% (SI: 68 mmol/mol; targets individualized) to 7.0% (SI: 53 mmol/mol), within 1 month, sustained at 6 months without hypoglycemia. The second, an 80-year-old woman receiving insulin and metformin, developed hypoglycemia; evaluation showed positive pancreatic autoantibodies and preserved C-peptide (1.6 ng/mL [SI: 0.53 nmol/L]). She underwent treatment de-escalation, achieving control on metformin alone with 77% TIR and GMI of 7.0% (SI: 53 mmol/mol) at 1 month, sustained at 3 months without hypoglycemia. These cases highlight preserved endogenous insulin secretion and therapeutic heterogeneity in LADA, requiring individualized management, including AID.

Introduction

Type 2 diabetes mellitus (T2DM) is highly prevalent in older adults, and management becomes more complex with advancing age because of multimorbidity, polypharmacy, frailty, and increased susceptibility to hypoglycemia [1]. Latent autoimmune diabetes in adults (LADA) may be misclassified as T2DM, particularly when insulin resistance is presumed. Epidemiologic studies suggest that approximately 10% of adults initially diagnosed with T2DM have underlying autoimmune diabetes [2]. Accurate diagnosis requires clinical suspicion and islet autoantibody testing.

Interpretation of endogenous insulin secretion is challenging. Detectable C-peptide does not exclude autoimmune diabetes and may persist for years. Reduced renal function can influence C-peptide, complicating interpretation in chronic kidney disease (CKD) [3]. Failure to recognize LADA may lead to glycemic instability, overtreatment, or avoidable hypoglycemia.

Diabetes technology, including continuous glucose monitoring (CGM) and automated insulin delivery (AID), has transformed insulin-treated diabetes, improving glycemic outcomes compared with multiple daily insulin injections [4-6]. However, very elderly adults remain underrepresented in these studies, and chronological age is often viewed as a barrier despite preserved functional status.

We describe 2 octogenarians aged 84 and 80 years with preserved endogenous insulin secretion and positive islet autoantibodies initially managed as T2DM. These cases illustrate delayed recognition in advanced age, preserved pancreatic reserve, and therapeutic heterogeneity ranging from deintensification to initiation of AID, underscoring the need for individualized management guided by physiology and functional status rather than chronological age alone.

Case presentation

Case 1

An 84-year-old man was referred to endocrinology for uncontrolled diabetes, treated as T2DM for >20 years. Medical history included stage 3b CKD, hypertension, and hyperlipidemia on statin therapy. Body mass index (BMI) was 23.1 kg/m² (reference range, 18.5-24.9 kg/m²). He reported unintentional 40-lb weight loss over 1 year and lived independently with preserved cognition and dexterity.

He described recurrent hypoglycemia (glucose < 70 mg/dL [SI: <3.9 mmol/L]) requiring emergency department visits and marked glycemic variability. Noninsulin therapies were limited by intolerance and renal impairment. He used basal insulin intermittently (35 units/day) and had discontinued prandial insulin due to hypoglycemia fear.

Continuous glucose monitoring demonstrated glycemic instability, with time in range (TIR; glucose 70-180 mg/dL [SI: 3.9-10.0 mmol/L]) of 33% and mean glucose of 211 mg/dL (SI: 11.7 mmol/L) (Table 1; Fig. 1A). Postprandial hyperglycemia alternated with hypoglycemia. Hemoglobin A1c (HbA1c) was 7.8% (SI: 62 mmol/mol; reference range, 3.8%-5.6% [18-38 mmol/mol]), whereas CGM-derived glucose management indicator (GMI) was 8.4% (SI: 68 mmol/mol; targets individualized), indicating greater glycemic exposure than reflected by HbA1c.

Table 1.

Continuous glucose monitoring metrics before and after automated insulin delivery (case 1)

Metrics	Baseline (pre-AID)	1 month on AID	6 months on AID	Target/goalsa
Time in range, %	33%	75%	73%	>70% within 70 to 180 mg/dL (SI: 3.9-10.0 mmol/L)
Time above range, %	67%	25%	27%	<25% above 180 mg/dL (SI: >10.0 mmol/L)
Very high glucose, %	25%	2%	1%	<5% above 250 mg/dL (SI: >13.9 mmol/L)
Time below range, %	<1%	0%	<1%	<4% below 70 mg/dL (SI: <3.9 mmol/L)
Very low glucose, %	<1%	0%	<1%	<1% below 54 mg/dL (SI: <3.0 mmol/L)
Average glucose	211 mg/dL (SI: 11.7 mmol/L)	156 mg/dL (SI: 8.7 mmol/L)	155 mg/dL (SI: 8.6 mmol/L)	<154 mg/dL (SI: <8.5 mmol/L)
Standard deviation	55 mg/dL (SI: 3.1 mmol/L)	40 mg/dL (SI: 2.2 mmol/L)	41 mg/dL (SI: 2.3 mmol/L)	<50 mg/dL (SI: <2.8 mmol/L)
Coefficient of variation, %	26%	25.5%	26.6%	≤36%
Glucose management indicator, %	8.4%	7.0%	7.0%	<7% (SI: <53 mmol/mol)

All data represent 14-day CGM summaries obtained using the Dexcom system (Dexcom, Inc., San Diego, CA). Representative images are shown in Fig. 1A to 1C.

^a CGM targets/goals are based on international consensus recommendations and may be individualized according to patient age, comorbidities, and hypoglycemia risk.

Abbreviations: AID, automated insulin delivery; CGM, continuous glucose monitoring; SI, Système International.

Figure 1.

Continuous glucose monitoring profiles before and after automated insulin delivery (case 1). All panels represent 14-day reports generated by the Dexcom CGM system (Dexcom, Inc., San Diego, CA). Images represent original CGM reports; corresponding quantitative metrics are reported in Table 1. (A) Baseline CGM report prior to initiation of AID, demonstrating marked glycemic variability with 33% TIR and substantial TAR. (B) CGM report 1 month after initiation of AID, showing improvement in TIR to 75% with reduction in TAR and no hypoglycemia. (C) CGM report at 6-month follow-up during AID, demonstrating sustained glycemic stability with 73% TIR and minimal hypoglycemia exposure. Abbreviations: AID, automated insulin delivery; CGM, continuous glucose monitoring; TIR, time in range (glucose 70-180 mg/dL [SI: 3.9-10.0 mmol/L]); TAR, time above range (glucose > 180 mg/dL [SI: >10.0 mmol/L]). Hypoglycemia defined as glucose < 70 mg/dL (SI: <3.9 mmol/L).

Given the normal BMI, longstanding insulin requirement, and glycemic instability disproportionate to insulin resistance, pancreatic autoantibody testing was performed.

Case 2

An 80-year-old woman with a 3-year history of presumed T2DM was referred to endocrinology for recurrent hypoglycemia on 6 units of premixed insulin and metformin. Body mass index was 26.4 kg/m². Medical history included hyperlipidemia on statin therapy and hypothyroidism secondary to Hashimoto thyroiditis. She lived independently, was cognitively intact, and was nonadherent to levothyroxine therapy.

She reported symptomatic hypoglycemia with glucose levels as low as 42 mg/dL (SI: 2.3 mmol/L), below the CGM target range of 70 to 180 mg/dL (SI: 3.9-10.0 mmol/L). Continuous glucose monitoring demonstrated 89% TIR with a mean glucose of 127 mg/dL (SI: 7.0 mmol/L) (Table 2; Fig. 2A). Hemoglobin A1c was 5.8% (SI: 40 mmol/mol), whereas CGM-derived GMI was 6.4% (SI: 46 mmol/mol), suggesting that recurrent hypoglycemia may have contributed to a lower HbA1c than reflected by the GMI.

Table 2.

Continuous glucose monitoring metrics before and after insulin discontinuation (case 2)

Metrics	On insulin (baseline)	1 month postinsulin discontinuation	3 months postinsulin discontinuation	Target/goalsa
Time in range, %	90%	77%	62%	>70% within 70 to 180 mg/dL (SI: 3.9-10.0 mmol/L)
Time above range, %	10%	23%	38%	<25% above 180 mg/dL (SI: >10.0 mmol/L)
Very high glucose, %	0%	6%	8%	<5% above 250 mg/dL (SI: >13.9 mmol/L)
Time below range, %	2%	0%	0%	<4% below 70 mg/dL (SI: <3.9 mmol/L)
Very low glucose, %	<1%	0%	0%	<1% below 54 mg/dL (SI: <3.0 mmol/L)
Average glucose	127 mg/dL (SI: 7.0 mmol/L)	153 mg/dL (SI: 8.5 mmol/L)	178 mg/dL (SI: 9.9 mmol/L)	<154 mg/dL (SI: <8.5 mmol/L)
Standard deviation	37 mg/dL (SI: 2.1 mmol/L)	48 mg/dL (SI: 2.7 mmol/L)	44 mg/dL (SI: 2.4 mmol/L)	<50 mg/dL (SI: <2.8 mmol/L)
Coefficient of variation, %	29%	31.2%	25%	≤36%
Glucose management indicator, %	6.4%	7.0%	7.6%	<7% (SI: <53 mmol/mol)

All data represent 14-day CGM summaries obtained using the Dexcom system (Dexcom, Inc., San Diego, CA). Representative images are shown in Fig. 2A to 2C.

^a CGM targets/goals are based on international consensus recommendations and may be individualized according to patient age, comorbidities, and hypoglycemia risk.

Abbreviations: CGM, continuous glucose monitoring; SI, Système International.

Figure 2.

Continuous glucose monitoring profiles before and after insulin discontinuation (case 2). All panels represent 14-day reports generated by the Dexcom CGM system (Dexcom, Inc., San Diego, CA). Images represent original CGM reports; corresponding quantitative metrics are reported in Table 2. (A) CGM report while receiving insulin therapy, demonstrating tight glycemic control with high TIR and low mean glucose. (B and C) CGM reports at 1 month and 3 months, respectively, after insulin discontinuation and metformin dose reduction, demonstrating resolution of hypoglycemia with acceptable TIR, considered a safer glycemic target for her age. Abbreviations: CGM, continuous glucose monitoring; TIR, time in range (glucose 70-180 mg/dL [SI: 3.9-10.0 mmol/L]). Hypoglycemia defined as glucose < 70 mg/dL (SI: <3.9 mmol/L).

Recurrent hypoglycemia on minimal insulin therapy, near-normal BMI, and coexisting autoimmune thyroiditis raised suspicion for LADA, prompting pancreatic autoantibody testing.

Diagnostic assessment

Case 1

Laboratory evaluation revealed strongly positive glutamic acid decarboxylase 65 (GAD65) antibodies (>250 IU/mL; reference range, <5 IU/mL), positive islet cell IgG antibodies, and elevated insulin autoantibodies (3.7 U/mL; reference range, <0.4 U/mL). Insulinoma-associated antigen-2 (IA-2) and zinc transporter 8 (ZnT8) antibodies were negative.

C-peptide was 5.68 ng/mL (SI: 1.89 nmol/L) (reference range, 1.1-5.5 ng/mL [SI: 0.37-1.83 nmol/L]), consistent with preserved endogenous insulin secretion, although stage 3b CKD may increase circulating C-peptide concentrations because of reduced renal clearance.

Serum creatinine was 1.61 to 1.63 mg/dL (SI: 142-144 µmol/L) (reference range, 0.70-1.20 mg/dL [SI: 62-106 µmol/L]), with an estimated glomerular filtration rate (eGFR) of 42 mL/min/1.73 m² (reference range, ≥60 mL/min/1.73 m²). Urine protein-to-creatinine ratio was 4.571 mg/mg creatinine (reference range, 0.025-0.148 mg/mg creatinine). Hemoglobin was 11.2 g/dL (SI: 112 g/L) (reference range, 13.2-17.1 g/dL [SI: 132-171 g/L]).

Multiple islet autoantibodies confirmed LADA. Despite preserved pancreatic reserve, these findings explained progressive insulin dependence and glycemic variability.

Case 2

Autoimmune evaluation demonstrated strongly positive GAD65 antibodies (>250 IU/mL) and markedly elevated IA-2 antibodies (110.7 U/mL; reference, <5.4 U/mL). Insulin autoantibodies were elevated (1.9 U/mL), whereas islet cell IgG and ZnT8 antibodies were negative.

C-peptide was 1.6 ng/mL (SI: 0.53 nmol/L), indicating preserved but limited insulin secretion.

Thyroid-stimulating hormone (TSH) was 10.5 mIU/L (reference range, 0.40-4.50 mIU/L) due to levothyroxine nonadherence.

Serum creatinine was 0.70 mg/dL (SI: 62 µmol/L) (reference range, 0.57-1.00 mg/dL [SI: 50-88 µmol/L]), with an eGFR of 88 mL/min/1.73 m². Urine microalbumin was <12 mg/L (reference, 0-30 mg/L). Hemoglobin was 12.9 g/dL (SI: 129 g/L) (reference range, 11.1-15.9 g/dL [SI: 111-159 g/L]).

Positive islet autoantibodies confirmed LADA.

Treatment

Case 1

Given recurrent hypoglycemia, glycemic variability, unintentional 40-lb weight loss, and intolerance to noninsulin agents, these therapies were discontinued, and low-dose basal–bolus insulin was restarted. Despite insulin dose optimization, glycemic instability persisted. Because of ongoing glycemic variability, increased hypoglycemia risk with CKD, and preserved cognition and dexterity, a tubeless CGM-integrated AID system was initiated.

Case 2

Given recurrent hypoglycemia on low-dose premixed insulin and an HbA1c of 5.8% (SI: 40 mmol/mol), insulin was discontinued, and metformin was reduced from 1000 mg twice daily to 500 mg twice daily. Thyroid-stimulating hormone improved to 3.89 mIU/L after resumption of levothyroxine therapy.

These cases highlight LADA heterogeneity across the spectrum between type 1 diabetes mellitus (T1DM) and T2DM, with glycemic patterns and insulin requirements influenced by residual β-cell reserve and aging-related immune changes.

Outcome and follow-up

Case 1

Following initiation of AID, CGM demonstrated improvement in TIR from 33% to 75% within 1 month and GMI from 8.4% (SI: 68 mmol/mol) to 7.0% (SI: 53 mmol/mol), with no readings of <70 mg/dL (SI: <3.9 mmol/L) (Table 1; Fig. 1B). Total daily insulin requirements decreased from approximately 30 to 50 units before AID to an average of 22 units/day. Glycemic variability decreased, and automated mode was utilized 100% of the time.

At 6-month follow-up, CGM demonstrated 73% TIR and GMI of 7.0% (SI: 53 mmol/mol) (Table 1; Fig. 1C). Hemoglobin A1c decreased from 7.8% (SI: 62 mmol/mol) before AID to 7.2% (SI: 55 mmol/mol) at 6 months.

At the most recent 8-month visit, these findings were sustained; he remained independent, continued successful AID use, reported improved treatment satisfaction, and had no hypoglycemia.

Case 2

Reduced-dose metformin without insulin resulted in resolution of hypoglycemia. At 1-month follow-up, CGM demonstrated 77% TIR with no readings of <70 mg/dL (SI: <3.9 mmol/L) and GMI of 7.0% (SI: 53 mmol/mol) (Table 2; Fig. 2B). At the most recent 3-month visit, CGM demonstrated 62% TIR, again with no readings of <70 mg/dL (SI: <3.9 mmol/L) and GMI of 7.6% (SI: 60 mmol/mol) (Table 2; Fig. 2C). Hemoglobin A1c was 6.7% (SI: 50 mmol/mol) at 3 months, compared with 5.8% (SI: 40 mmol/mol) during insulin therapy at the initial visit, reflecting intentional relaxation of glycemic targets to minimize hypoglycemia risk.

Glycemic stability was maintained without insulin, representing a safer glycemic target for an older adult with prior hypoglycemia.

Discussion

Latent autoimmune diabetes in adults may remain misclassified as T2DM, particularly in older adults where autoimmune testing is not routinely pursued. Accurate classification requires islet autoantibody testing and C-peptide interpretation when clinical features are atypical. Latent autoimmune diabetes in adults is characterized by pancreatic autoimmunity with variable preservation of endogenous insulin secretion and may present well beyond midlife.

Latent autoimmune diabetes in adults represents a continuum between T1DM and T2DM with slower immune-mediated β-cell destruction. Despite measurable C-peptide levels, ongoing autoimmune β-cell dysfunction leads to progressive glycemic instability and eventual insulin dependence. Preserved C-peptide levels do not exclude significant insulin deficiency.

Compared with T2DM, patients with LADA exhibit higher mean glucose, lower TIR, and greater glycemic variability, reflected by higher standard deviation, mean amplitude of glycemic excursions, and coefficient of variation on CGM [7]. This variability correlates with declining C-peptide and progression from partial to absolute insulin deficiency. Studies demonstrate an inverse relationship between C-peptide and glycemic variability, with LADA occupying an intermediate position between T1DM and T2DM, supporting a spectrum spanning T1DM, high- and low-titer LADA, and T2DM rather than discrete categories [8].

Latent autoimmune diabetes in adults is biologically heterogeneous: Higher GAD antibody titers (≥180 U/mL) resemble T1DM, whereas lower titers (<180 U/mL) resemble T2DM [9]. Additional heterogeneity may reflect aging-related immunologic changes, including immunosenescence, contributing to delayed or atypical presentations.

Latent autoimmune diabetes in adults should be considered in individuals with atypical features for T2DM, particularly lean phenotype, progressive insulin requirement, coexisting autoimmune disease, unexplained glycemic variability, or recurrent hypoglycemia on modest insulin doses. Autoimmune clustering is a recognized feature of LADA [10], and LADA accounts for approximately 2% to 12% of adult-onset diabetes, although prevalence varies by population [11].

Management should be guided by residual β-cell function. A C-peptide–based framework classifies values <0.3 nmol/L (SI: <0.9 ng/mL) as severe insulin deficiency requiring insulin therapy, 0.3 to 0.7 nmol/L (SI: 0.9-2.1 ng/mL) as a “gray area” where a modified T2DM algorithm with consideration of early insulin is appropriate, and >0.7 nmol/L (SI: >2.1 ng/mL) as permitting initial management similar to T2DM. Given the progressive nature of LADA, periodic reassessment of C-peptide is recommended to guide treatment adjustments [11].

Despite preserved C-peptide in both cases, therapeutic responses differed, underscoring LADA heterogeneity. Continuous glucose monitoring and AID may improve glycemic stability and reduce hypoglycemia, particularly in complex cases and with CKD. In case 1, AID improved control without hypoglycemia, demonstrating feasibility in selected octogenarians. In contrast, case 2 achieved stable glycemia after deintensification to noninsulin therapy. These contrasting responses support individualized management based on pancreatic reserve, comorbidities, and glycemic variability rather than age alone.

This study is limited by small sample size, short follow-up, and nonfasting C-peptide levels without paired glucose measurements. Nevertheless, these cases show that LADA can present late in life and may require management ranging from advanced technology to cautious deintensification. Recognition of this phenotype may enable safer individualized therapy and reduce preventable hypoglycemia.

Learning points

• Latent autoimmune diabetes in adults may present late in life and remain misclassified as T2DM; pancreatic autoantibody testing should be considered in individuals with atypical features, unexplained glycemic variability, or recurrent hypoglycemia.

• Preserved C-peptide does not exclude autoimmune diabetes; LADA is a clinical spectrum requiring individualized management.

• Latent autoimmune diabetes in adults demonstrates therapeutic heterogeneity; management may range from insulin deintensification to advanced insulin delivery, guided by pancreatic reserve, comorbidities, and hypoglycemia risk.

• Continuous glucose monitoring-integrated AID can safely improve glycemic stability in selected older adults, with candidacy determined by cognitive and functional status.

Contributors

S.S. was involved in patient diagnosis and management. S.S. and N.S. collected clinical data, drafted and revised the manuscript, and approved the final version for submission. All authors agree to be accountable for all aspects of the work.

Funding

No public or commercial funding.

Disclosures

The authors acknowledge the use of an AI-based language model (ChatGPT) to assist with language editing and to improve the clarity of the manuscript. All clinical content, interpretations, and conclusions were reviewed and approved by the authors.

Informed patient consent for publication

Signed informed consent obtained directly from the 2 patients.

Data availability

Original data generated and analyzed during this study are included in this published article.