High intensity intermittent exercise improves cardiac structure and function and reduces liver fat in patients with type 2 diabetes: a randomised controlled trial

The researchers tested whether high intensity intermittent training (HIIT) could improve heart structure and function in people with type 2 diabetes who had no diagnosed heart disease. They also measured effects on liver fat, visceral fat, blood sugar control, and overall fitness.

Intervention: HIIT performed on a stationary cycle ergometer, three sessions per week for 12 weeks. Each session consisted of:

• 5-minute warm-up at low intensity
• 5 cycles of: 3 minutes at 90% of peak heart rate (very hard effort), followed by 2 minutes of active recovery at 50% of peak heart rate (easy pedalling)
• 5-minute cool-down
• Total session time: approximately 35 minutes, with only 15 minutes of actual high-intensity work

Comparator: Standard care — participants were instructed to continue their usual lifestyle and medical management without any structured exercise program.

Primary outcome: Left ventricular structure (wall mass) and systolic function (stroke volume — the amount of blood pumped per heartbeat)

Secondary outcomes: Diastolic function (early filling rates), cardiac torsion (twisting motion of the heart), liver fat percentage, visceral fat, HbA1c, fasting glucose, 2-hour glucose during an oral glucose tolerance test, and peak oxygen consumption (VO₂peak)

• Sample size: 28 patients with type 2 diabetes were randomised (14 per group); 23 completed the study (12 HIIT, 11 control)
• Population: Adults with type 2 diabetes, stable on diet and/or metformin for at least 6 months
• Age: Mean 59–61 years (range not reported, but standard deviation ±9 years suggests approximately 50–70 years)
• Sex: 18 men, 5 women (uneven split — 10 men/2 women in HIIT, 8 men/3 women in control)
• BMI: Mean 31–32 kg/m² (classified as obese)
• Time since diagnosis: Mean 4–5 years
• Baseline fitness: Very low — mean VO₂peak of 20–22 ml/kg/min (typical for sedentary adults with type 2 diabetes)
• Medications: Approximately half on metformin, half on statins, some on blood pressure medications
• Setting: Newcastle upon Tyne, UK; recruited from newspaper ads and diabetes community groups between September 2012 and September 2013
• Exclusions: Anyone with diagnosed heart disease, regular exercise (≥60 min/week of moderate-vigorous activity), taking beta-blockers, or with contraindications to exercise stress testing

Cardiac structure and function: 3.0 Tesla MRI (Philips Achieva) — a high-field-strength scanner that provides detailed images of the heart. Key measurements included:

• Left ventricular wall mass (grams) — measured from manual tracing of endocardial and epicardial borders on short-axis images
• Stroke volume (ml) — calculated from end-diastolic and end-systolic volumes
• Early diastolic filling rate (ml/s) — how fast the heart relaxes and fills with blood
• Ejection fraction (%) — percentage of blood pumped out with each beat
• Cardiac output (l/min) — total blood pumped per minute
• Eccentricity ratio — a measure of whether the heart is thickening abnormally (concentric remodelling)

Cardiac tagging: A specialised MRI technique that applies a grid pattern to the heart muscle and tracks how it deforms during contraction. This measured:

• Peak torsion (degrees) — the twisting motion of the heart during contraction
• Strain — how much the heart muscle stretches and shortens

Cardiac energy metabolism: ³¹P-magnetic resonance spectroscopy — measured the ratio of phosphocreatine (PCr) to ATP, which reflects the energy status of heart muscle cells

Liver fat: ¹H-magnetic resonance spectroscopy — a precise, non-invasive method that measures the fat fraction within liver tissue as a percentage of total signal

Visceral fat: Three-point Dixon MRI sequence at the L4-L5 spinal level — measures the deep abdominal fat surrounding organs

Glycaemic control:

• HbA1c (%) — average blood sugar over ~3 months
• Fasting glucose (mmol/l)
• 2-hour glucose during oral glucose tolerance test (OGTT) — measures how well the body handles a sugar load
• Area under the glucose curve (AUGC) — total glucose exposure during the OGTT

Fitness: Peak oxygen consumption (VO₂peak) measured during a maximal exercise test on a semirecumbent cycle ergometer, with resistance increasing by 1 W every 6 seconds until exhaustion

Blood markers: Liver enzymes (ALT, AST, ALP), lipids (total cholesterol, HDL, LDL, triglycerides), and blood pressure

MRI analysis was performed by a single observer blinded to group allocation — meaning the person measuring the heart images did not know which group each participant was in.

Study design: Randomised controlled trial (RCT) — the gold standard for testing whether an intervention causes an effect.

Randomisation: Simple random allocation sequence generated by an online tool (www.randomization.com). Concealed envelopes with consecutive numbers were locked in a drawer and withdrawn in numerical order by the main author. This means the researchers could not predict or influence which group a participant would be assigned to.

Blinding:

• The MRI analyst was blinded to group allocation — this prevents bias in measuring outcomes
• Participants could not be blinded because they knew whether they were exercising or not (no sham exercise control)
• The exercise trainers and study coordinators were not blinded

Duration: 12 weeks of intervention, with measurements taken at baseline and after 12 weeks. Glucose control was measured 48–72 hours after the final exercise session to avoid capturing the acute (immediate) effects of exercise on blood sugar.

Exercise adherence: Not explicitly reported in the abstract, but the full text notes that participants completed the prescribed sessions (no adverse events were recorded).

Statistical approach: Intention-to-treat analysis — meaning all participants who started were included in the final analysis, even if they didn't complete the study. This is the conservative approach that preserves the benefits of randomisation. Between-group comparisons were made (HIIT vs control), with p-values reported for key outcomes.

What this design can prove:

• Causality — because of randomisation, any differences between groups at 12 weeks can be attributed to the HIIT intervention (assuming no major confounds)
• The effect of HIIT specifically, compared to no exercise

What this design cannot prove:

• Whether HIIT is better than moderate continuous exercise (no comparison group)
• Whether effects persist beyond 12 weeks (no follow-up)
• Whether effects apply to people with diagnosed heart disease (excluded)
• Whether effects apply to people already exercising regularly (excluded)
• Whether the results are driven by the small number of women (only 5 out of 23 completers)

Major methodological weaknesses:

• Very small sample size — 23 completers means the study is underpowered for detecting small-to-moderate effects; some true effects may have been missed
• No blinding of participants — the placebo effect (expecting to improve because you're exercising) could influence some outcomes, though objective MRI measures are less susceptible
• No active control group — we don't know if the same time spent doing moderate exercise would produce similar or different results
• Uneven sex distribution — results may not generalise well to women with type 2 diabetes
• High dropout rate — 5 out of 28 (18%) did not complete; while intention-to-treat analysis helps, dropouts could bias results
• Short duration — 12 weeks is long enough to see changes but not to know if they are sustained
• Single-centre study — results may not generalise to other populations or settings

Primary outcomes — Cardiac structure and function:

• Left ventricular wall mass increased in the HIIT group (104±17 g to 116±20 g) compared to control (107±25 g to 105±25 g), p<0.05. This represents a ~12% increase in heart muscle mass — but importantly, this was physiological (healthy) hypertrophy, not pathological (disease-related) thickening, as it was accompanied by improved function.

• Stroke volume increased in the HIIT group (76±16 ml to 87±19 ml) compared to control (79±14 ml to 75±15 ml), p<0.01. This is a ~14% increase in the amount of blood pumped per heartbeat.

• Early diastolic filling rate increased in the HIIT group (241±84 ml/s to 299±89 ml/s) compared to control (250±44 ml/s to 251±47 ml/s), p<0.05. This is a ~24% improvement in how fast the heart relaxes and fills with blood.

• Peak torsion decreased in the HIIT group (8.1±1.8° to 6.9±1.6°) compared to control (7.1±2.2° to 7.6±1.9°), p<0.05. This reduction in twisting motion is actually a positive sign — it indicates the heart is no longer overcompensating for stiffness.

• Ejection fraction did not change significantly (was normal at baseline in both groups).

• Cardiac energy metabolism (PCr/ATP ratio) did not change significantly.

Secondary outcomes — Metabolic and body composition:

• Liver fat decreased by 39% relative reduction in the HIIT group (p<0.05). Absolute values not reported in the abstract, but baseline was ~7% in both groups.

• HbA1c decreased in the HIIT group (7.1±1.0% [54.5 mmol/mol] to 6.8±0.9% [51.3 mmol/mol]) compared to control (7.2±0.5% [54.9 mmol/mol] to 7.4±0.7% [57.0 mmol/mol]), p<0.05. This is a 0.3 percentage point reduction in HbA1c.

• Changes in liver fat correlated with changes in HbA1c (r=0.70, p<0.000) and with changes in 2-hour glucose (r=0.57, p<0.004). This means people who lost more liver fat also had greater improvements in blood sugar control.

• VO₂peak increased in the HIIT group (21.8±5.4 to 25.6±6.0 ml/kg/min) compared to control (20.3±6.1 to 20.0±5.9 ml/kg/min), p<0.001. This is a ~17% improvement in cardiorespiratory fitness.

• Fasting glucose, 2-hour glucose, and area under the glucose curve did not change significantly between groups (though the correlation with liver fat suggests individual variation).

• Visceral fat did not change significantly.

• Body weight did not change significantly (both groups remained ~90 kg).

• Blood pressure did not change significantly.

• Liver enzymes (ALT, AST, ALP) did not change significantly.

Heart function: The 14% increase in stroke volume means that after 12 weeks of HIIT, each heartbeat pumped about 11 ml more blood. For context, a typical resting stroke volume is 60–80 ml, so this is a meaningful improvement — roughly equivalent to what you'd expect from several months of endurance training. The 24% improvement in diastolic filling rate means the heart relaxed and refilled about 58 ml/s faster — this is a large effect that moves many participants from "impaired relaxation" toward normal function.

Liver fat: The 39% relative reduction in liver fat is striking. If someone started with 7% liver fat (mild-to-moderate fatty liver), a 39% reduction would bring them to about 4.3% — below the typical threshold for fatty liver disease (5%). This is the largest reduction in liver fat ever recorded from an exercise intervention at the time of publication. For comparison, weight loss of 5-10% of body weight typically reduces liver fat by 20-50%, but participants in this study did not lose weight — meaning HIIT reduced liver fat independently of weight loss.

Blood sugar control: The 0.3 percentage point reduction in HbA1c (from 7.1% to 6.8%) is modest but clinically meaningful. For someone with type 2 diabetes, each 1% reduction in HbA1c is associated with a 14% reduction in heart attack risk and a 37% reduction in microvascular complications. A 0.3% reduction is roughly one-third of that benefit — not trivial, but not a cure.

Fitness: The 17% improvement in VO₂peak (from 21.8 to 25.6 ml/kg/min) is substantial for a 12-week program in previously sedentary older adults. This moves someone from "poor" to "fair" cardiorespiratory fitness for their age group.

What the authors acknowledge:

• Small sample size limits generalisability
• No long-term follow-up to assess sustainability
• No comparison with moderate continuous exercise
• Uneven sex distribution (mostly men)
• Participants were relatively well-controlled (mean HbA1c 7.1%) — results may differ for those with poorer control

What a critical reader would note:

• No blinding of participants or trainers — while objective MRI measures are robust, the lack of blinding could influence effort during exercise tests or adherence to other lifestyle behaviours
• No control for attention/time — the control group received no equivalent time commitment or attention from researchers, so some benefits could be from the structured routine itself rather than HIIT specifically
• High dropout (18%) — though comparable to other exercise studies, dropouts could bias results if they differed systematically from completers
• Industry funding — funded by the Medical Research Council (public, non-industry), so no conflict of interest concerns
• Single-centre, single-country — results may not generalise to different healthcare systems, ethnicities, or dietary patterns
• Short duration — 12 weeks is too short to assess whether cardiac improvements translate to reduced heart attacks or mortality
• No dietary control — participants were not instructed to change their diet, but dietary intake was not rigorously monitored; undetected dietary changes could confound results
• Only 5 women — sex differences in cardiac response to exercise are well-documented, and these results may not apply to women
• Exclusion of people with diagnosed heart disease — the population most likely to benefit from cardiac improvements was excluded
• No measure of daily physical activity outside the intervention — control participants could have increased or decreased their activity, diluting or exaggerating the apparent effect

For someone running their own n=1 experiment:

What to test:

• HIIT on a stationary bike: 3-minute intervals at very hard effort (90% of max heart rate, or 8-9 out of 10 on perceived exertion), followed by 2 minutes of easy pedalling. Repeat 5 times. Total workout ~35 minutes including warm-up and cool-down.
• Do this 3 times per week for 12 weeks minimum.

Minimum meaningful duration:

• 12 weeks is the minimum to see cardiac structural changes. Some metabolic improvements (liver fat, HbA1c) may appear earlier (4-8 weeks), but heart remodelling takes longer.
• For a self-experiment, commit to 12 weeks with measurements at baseline, 6 weeks, and 12 weeks.

What to measure (specific metrics):

• Primary: Rest