Subgroups of adult-onset diabetes: a prospective follow-up study of progression of insulin resistance and deficiency and association with liver steatosis and fibrosis

This prospective study demonstrates that while diabetes subgroups based on insulin deficiency or resistance become less distinct over time, liver fibrosis remains strongly associated with insulin resistance independent of BMI, suggesting that subgroup classification is most effective near diagnosis and that insulin resistance, rather than just obesity, is a key driver of liver disease risk.

The Gist

The Big Picture: Diabetes Isn't Just One Thing

Imagine Adult-Onset Diabetes not as a single, uniform monster, but as a "Swiss Army Knife" with many different tools. For a long time, doctors treated all adult diabetes the same way. But this study suggests that the "tools" (the underlying causes) are actually very different for different people.

The researchers took a group of people with diabetes and sorted them into five different teams based on their biology:

1. SAID: The "Autoimmune Team" (The body is attacking itself).
2. SIDD: The "Empty Tank Team" (The body isn't making enough insulin).
3. SIRD: The "Clogged Pipes Team" (The body makes insulin, but the cells are resistant to it).
4. MOD: The "Heavy Load Team" (Linked strongly to obesity).
5. MARD: The "Age-Related Team" (Linked strongly to getting older).

The Experiment: A Time-Travel Study

The researchers didn't just look at these people once; they followed them for about 4 to 5 years. They wanted to see two things:

1. Do the teams stay the same? If you are on the "Clogged Pipes Team" today, will you still be there in five years?
2. What happens to the liver? They specifically looked at Liver Fat (Steatosis) and Liver Scarring (Fibrosis), which are like the "rust" and "cracks" that form in the body's chemical processing plant.

Key Finding #1: The Teams Shuffle (Stability)

The Analogy: Imagine a high school where students are sorted into clubs based on their grades and hobbies.

• The Stable Clubs: The "Age-Related" and "Obesity" clubs were very stable. If you joined them, you likely stayed there (76–91% of people).
• The Shuffling Clubs: The "Empty Tank" and "Clogged Pipes" clubs were chaotic. Only about 1 in 5 people stayed in their original group. Most of them moved to the "Age-Related" club.

Why?
Think of it like fixing a leaky roof.

• The "Empty Tank" group started with very high sugar because their insulin tank was empty. Once they started treatment (filling the tank), their sugar levels dropped, and they started looking more like the "Age-Related" group.
• The "Clogged Pipes" group started with high resistance. Even though they didn't lose much weight, their treatment helped clear the pipes a bit, changing their numbers enough to move them to a different category.

The Lesson: You have to sort people into these teams early, right after diagnosis. If you wait too long, the treatment changes their biology so much that the original "team" label no longer fits.

Key Finding #2: The Liver Connection

The researchers wanted to know: Who is most likely to get liver scarring (fibrosis)?

The Old Way of Thinking: "If you are heavy (high BMI) and have a fatty liver, you will get liver scarring."
The New Discovery: "It's not just about the weight; it's about the resistance."

The Analogy: Imagine the liver is a busy warehouse.

• BMI (Weight) is like the number of trucks arriving at the warehouse. More trucks = more traffic.
• Insulin Resistance is like the warehouse workers refusing to unload the trucks. They are stuck, blocking the doors.

The study found that even if the number of trucks (BMI) was the same, the warehouses with the uncooperative workers (Insulin Resistance) were the ones getting the most damage (scarring/fibrosis).

Specifically, the "Clogged Pipes" (SIRD) group had the highest rates of liver scarring (42%), followed closely by the "Heavy Load" (MOD) group. But here is the kicker: Even after accounting for weight, the level of insulin resistance was the strongest predictor of liver damage.

Key Finding #3: The "Hidden" Resistance

The study used some fancy tests (like a "glucagon-insulin stress test") to see how the body actually reacted, rather than just guessing from a blood test.

• They found that the "Clogged Pipes" group had terrible resistance in their liver and fat tissue, but their muscles were actually doing okay.
• This is important because it means the "Clogged Pipes" group needs specific treatments that target the liver and fat, not just general weight loss.

The Takeaway for You

1. Timing is everything: If you are diagnosed with diabetes, get classified early. The "team" you are on today might change in a few years as treatment works.
2. Weight isn't the whole story: You can be a normal weight but still have severe insulin resistance that damages your liver. Doctors need to check for "resistance," not just "weight."
3. Liver Health: If you have diabetes, especially if you are in the "Clogged Pipes" group, your liver is at high risk for scarring. This isn't just about fatty liver; it's about the body's inability to process sugar properly, which acts like a corrosive acid on the liver.

In short: Diabetes is a shape-shifter. To protect your liver and heart, we need to understand which shape it is taking right now, not just treat everyone the same way.

Technical Summary

1. Problem Statement

Adult-onset diabetes is a heterogeneous disease with varying pathophysiological mechanisms (e.g., insulin deficiency vs. insulin resistance) that translate into different risks for comorbidities. Previous research (Andersson et al., 2018) identified five distinct subgroups based on clinical variables near diagnosis:

1. SAID: Severe Autoimmune Diabetes (GADA-positive).
2. SIDD: Severe Insulin-Deficient Diabetes (GADA-negative).
3. SIRD: Severe Insulin-Resistant Diabetes.
4. MOD: Moderate Obesity-Related Diabetes.
5. MARD: Mild Age-Related Diabetes.

While these subgroups show different risks for complications (e.g., SIRD has higher risks for kidney disease and liver issues), critical gaps remain:

• Stability: It is unclear if these subgroups remain stable over time or if treatment and disease progression cause patients to "migrate" between clusters.
• Pathophysiology: The specific tissue-level mechanisms (liver, muscle, adipose) driving insulin resistance in these subgroups, particularly regarding Steatotic Liver Disease (SLD) and Liver Fibrosis (LF), are not fully understood.
• Clinical Utility: It is unknown if standard fasting measures (like HOMA2-IR) accurately reflect stimulated insulin sensitivity or if they predict liver fibrosis independently of BMI.

2. Methodology

This was a prospective, regional follow-up study conducted in the Vaasa Hospital District, Finland, utilizing the DIREVA diabetes registry.

• Study Population:
  • Group A (n=547): Participants with questionnaire data and new fasting laboratory samples. Follow-up occurred 2–10 years after initial registration (median 4.6 years).
  • Group B (n=194): A subset of Group A who underwent a comprehensive study visit including an intravenous glucagon-insulin tolerance test (GITT) and Vibration-Controlled Transient Elastography (VCTE/FibroScan).
• Subgroup Assignment: Clustering was performed at baseline (near diagnosis) using the ANDIS cohort algorithm based on: Age, BMI, HbA1c, HOMA2-B (insulin secretion), and HOMA2-IR (insulin resistance).
• Key Measurements:
  • Metabolic: Fasting glucose, C-peptide, insulin, lipids, and GADA/IA-2A antibodies.
  • Dynamic Testing (Group B):
    • Glucagon Stimulation: Measured $\Delta$C-peptide (0–6 min) to assess beta-cell function.
    • Insulin Tolerance: Measured the glucose disappearance rate ($K_{ITT}$) to assess peripheral (muscle) insulin sensitivity.
    • Adipo-IR: Calculated as Fasting Free Fatty Acids (FFA) $\times$ Insulin to assess adipose tissue insulin resistance.
  • Liver Imaging (Group B):
    • CAP (Controlled Attenuation Parameter): Quantitative measure of liver steatosis.
    • LSM (Liver Stiffness Measurement): Quantitative measure of liver fibrosis (cut-off >8 kPa defined fibrosis).
• Statistical Analysis: Used Kruskal-Wallis for group comparisons, Wilcoxon signed-rank for longitudinal changes, and multivariable logistic regression to assess associations between insulin resistance markers and liver fibrosis, adjusting for BMI and CAP.

3. Key Contributions

• Longitudinal Stability Analysis: First prospective study to track the stability of diabetes subgroups over a 4–10 year period, revealing that severe phenotypes (SIDD/SIRD) are highly unstable compared to mild phenotypes (MOD/MARD).
• Tissue-Specific Insulin Resistance: Differentiated between liver, muscle, and adipose tissue insulin resistance using both fasting indices (HOMA2-IR, Adipo-IR) and dynamic tests ($K_{ITT}$), showing that subgroups differ primarily in hepatic resistance rather than muscle resistance.
• Liver Fibrosis Predictors: Established that insulin resistance in multiple tissues is an independent predictor of liver fibrosis, even after adjusting for BMI and liver fat content.

4. Key Results

A. Subgroup Stability and Progression

• Attenuation of Differences: The distinct metabolic differences between subgroups diminished over time.
• Cluster Stability:
  • Stable: MOD (76%) and MARD (91%) retained their original cluster assignment.
  • Unstable: SIDD (only 13% remained) and SIRD (only 21% remained). Most participants from SIDD and SIRD were reassigned to the MARD cluster.
• Drivers of Change:
  • SIDD: Marked decrease in HbA1c and increase in HOMA2-B (insulin secretion) likely due to treatment initiation, making them resemble MARD.
  • SIRD: HOMA2-IR decreased significantly despite stable BMI, suggesting treatment effects or natural history changes.

B. Insulin Secretion and Sensitivity (Group B)

• Secretion: Glucagon-stimulated C-peptide correlated moderately with fasting HOMA2-B. SAID and SIDD had the lowest secretion; SIRD had the highest.
• Sensitivity:
  • HOMA2-IR vs. $K_{ITT}$: There was a weak correlation ($R=-0.33$) between fasting HOMA2-IR and the dynamic $K_{ITT}$ (muscle sensitivity).
  • Tissue Specificity: While SIRD had the highest HOMA2-IR (liver resistance), their $K_{ITT}$ (muscle sensitivity) was not significantly different from MOD/MARD (except SAID). This suggests SIRD is characterized primarily by hepatic insulin resistance, not muscle resistance.
  • Adipose Resistance: Adipo-IR was highest in SIRD and lowest in SAID.

C. Liver Steatosis (SLD) and Fibrosis (LF)

• Prevalence: SLD and LF were most prevalent in SIRD (70% SLD, 42% LF) and MOD (60% SLD, 41% LF).
• BMI Confounding: Differences in SLD between groups were largely explained by BMI.
• Independent Association with Fibrosis:
  • In a multivariable model adjusting for BMI and CAP, Liver Fibrosis was significantly associated with:
    • HOMA2-IR: OR 3.90 (95% CI 2.05–7.42).
    • Adipo-IR: OR 1.14 (95% CI 1.06–1.22).
    • $K_{ITT}$ (Muscle Sensitivity): OR 0.33 (95% CI 0.17–0.64) (lower sensitivity = higher risk).
  • Predictive Value: Adding HOMA2-IR to a model containing BMI or CAP significantly improved the detection of liver fibrosis (AUC increased from ~0.73 to ~0.84).

5. Significance and Clinical Implications

• Timing of Clustering: Diabetes subtyping should be performed near diagnosis. The distinct phenotypic features of severe subgroups (SIDD/SIRD) attenuate over time due to treatment and disease progression, leading to misclassification if done later.
• Risk Stratification for Liver Disease:
  • Relying solely on BMI or liver fat (steatosis) is insufficient to predict liver fibrosis risk.
  • Insulin resistance (specifically hepatic and adipose) is a critical, independent driver of liver fibrosis in diabetes.
  • Clinicians should evaluate the degree of insulin resistance, not just BMI, to identify patients at high risk for liver fibrosis, particularly in the SIRD phenotype.
• Pathophysiological Insight: The study clarifies that the "Severe Insulin-Resistant Diabetes" (SIRD) phenotype is defined by profound hepatic and adipose tissue resistance, while muscle sensitivity remains relatively preserved compared to other groups. This distinction may inform targeted therapies (e.g., agents specifically targeting hepatic insulin resistance).

Limitations: The study is a preprint (not yet peer-reviewed), has a relatively small sample size for the SIDD group (n=16), and uses VCTE (FibroScan) rather than liver biopsy for fibrosis assessment. Follow-up time (median 4.6 years) may be short for the full development of advanced fibrosis.