Repeated caffeine intake suppresses cerebral grey matter responses to chronic sleep restriction in an A(1) adenosine receptor-dependent manner: a double-blind randomized controlled study with PET-MRI.

The study tested whether daily caffeine intake (300 mg/day, split into three doses of 100 mg each) could counteract or modify the effects of five nights of partial sleep deprivation (5 hours in bed per night) on brain structure and function. The comparator was a placebo group that underwent the same sleep restriction protocol but received identical-looking capsules containing cellulose. The primary outcome was change in cerebral grey matter volume measured by structural MRI. Secondary outcomes included cerebral blood flow, A₁ adenosine receptor availability measured by PET, subjective sleepiness (Karolinska Sleepiness Scale), objective vigilance (Psychomotor Vigilance Task), and sleep quality (polysomnography).

36 healthy adults (18 male, 18 female), aged 18–35 years (mean age 24.8 ± 3.8 years). All were right-handed, non-smokers, with no history of sleep disorders, neurological or psychiatric conditions, and no regular medication use. Habitual caffeine consumption was low to moderate (≤200 mg/day, equivalent to ~2 cups of coffee). Participants were recruited from the general population of the city of Lübeck, Germany, and surrounding areas. The study was conducted at the University of Lübeck and the University Medical Center Hamburg-Eppendorf.

**Structural MRI:** T1-weighted images at 3 Tesla, processed with FreeSurfer to quantify cortical thickness and grey matter volume in 68 cortical regions (Desikan-Killiany atlas).

**PET imaging:** [¹⁸F]CPFPX, a radioligand that binds specifically to A₁ adenosine receptors, to measure receptor availability (non-displaceable binding potential, BPND) in grey matter.

**Cerebral blood flow:** Arterial spin labelling (ASL) MRI.

**Subjective sleepiness:** Karolinska Sleepiness Scale (KSS, 1–9 scale, 1=extremely alert, 9=very sleepy, fighting sleep).

**Objective vigilance:** Psychomotor Vigilance Task (PVT), measuring reaction time and lapses (reaction times >500 ms) during a 10-minute test.

**Sleep monitoring:** Polysomnography (PSG) on nights 1, 3, and 5 of the sleep restriction period, including EEG, EOG, and EMG. Also wrist actigraphy throughout.

**Caffeine/placebo administration:** Double-blind, double-dummy design with capsules prepared by the hospital pharmacy.

**Design:** Double-blind, randomised, placebo-controlled, parallel-group trial. Participants were randomly assigned to either the caffeine group (n=18) or the placebo group (n=18). Randomisation was performed by the hospital pharmacy using a computer-generated random number list, and allocation was concealed from both participants and researchers.

**Protocol:** The study lasted 11 days total. Days 1–2 were baseline (habitual sleep, no intervention). Days 3–7 were the sleep restriction period: participants were allowed only 5 hours in bed per night (23:00–04:00). During this period, the caffeine group received 100 mg of caffeine three times daily (at 08:00, 12:00, and 16:00), while the placebo group received identical capsules. Days 8–11 were recovery (habitual sleep, no intervention). Brain imaging (MRI and PET) was performed at three time points: baseline (day 2), after 5 nights of sleep restriction (day 7), and after 3 nights of recovery sleep (day 10).

**Why this design matters:** The parallel-group design (caffeine vs. placebo) allows direct comparison of the effect of caffeine versus no caffeine under identical sleep restriction conditions. Double-blinding prevents expectation bias — participants don't know whether they're getting caffeine or placebo, which is critical because caffeine has noticeable subjective effects (alertness, jitteriness) that could otherwise influence reporting. The inclusion of a recovery period allows assessment of whether any brain changes are reversible. However, the design cannot prove causality at the molecular level — it can only show associations between caffeine intake, receptor binding, and grey matter changes. The study is also limited by the relatively short duration of sleep restriction (5 nights) and the single dose of caffeine (300 mg/day), so it cannot tell us about long-term effects or dose-response relationships.

**Statistical approach:** Linear mixed-effects models with fixed effects for group (caffeine vs. placebo), time (baseline, restriction, recovery), and their interaction, plus random intercepts for participants. Primary analysis focused on the group-by-time interaction for cortical thickness. Correction for multiple comparisons across 68 cortical regions was performed using false discovery rate (FDR) at q<0.05. Effect sizes are reported as Cohen's d.

**What this design can and cannot prove:** The design can prove that caffeine, compared to placebo, is associated with different patterns of grey matter change during sleep restriction. Because of randomisation and blinding, any differences between groups can be attributed to the caffeine intervention rather than pre-existing differences or expectation effects. However, the design cannot prove that the effect is *caused* by A₁ adenosine receptor blockade — that inference is supported by the PET data showing correlation between receptor availability and grey matter changes, but correlation is not causation. The design also cannot rule out that caffeine's effects are mediated through other mechanisms (e.g., adenosine A₂A receptors, phosphodiesterase inhibition). Additionally, because there was no "normal sleep + caffeine" control group, we cannot separate the effects of caffeine during sleep restriction from caffeine's effects during normal sleep.

**Primary outcome — cortical thickness:**

In the placebo group, 5 nights of sleep restriction caused significant thinning in 17 of 68 cortical regions (FDR-corrected), with an average reduction of 0.02–0.05 mm (approximately 1–3% of baseline thickness). The most affected regions were the prefrontal cortex, anterior cingulate, insula, and temporal poles.

In the caffeine group, only 2 regions showed significant thinning (right superior temporal sulcus and left pars opercularis), and the magnitude was smaller (0.01–0.02 mm).

The group-by-time interaction was significant for 12 regions (FDR-corrected), meaning caffeine significantly attenuated grey matter loss compared to placebo. Effect sizes ranged from d=0.4 to d=0.8 (small to large).

After 3 nights of recovery sleep, cortical thickness returned to baseline in both groups, with no significant differences between groups.

**Secondary outcomes — A₁ adenosine receptor availability:**

Baseline A₁ receptor binding potential (BPND) did not differ between groups.

After sleep restriction, the placebo group showed a significant increase in A₁ receptor availability in several cortical regions (mean increase ~8–12%), consistent with upregulation of receptors in response to increased adenosine levels from sleep deprivation.

The caffeine group showed no significant change in A₁ receptor availability, suggesting that caffeine blocked the receptor upregulation.

Importantly, the degree of A₁ receptor upregulation in the placebo group correlated positively with the degree of cortical thinning (r=0.52, p=0.03), meaning individuals who showed more receptor upregulation also showed more grey matter loss. This correlation was absent in the caffeine group.

**Secondary outcomes — cerebral blood flow:**

Sleep restriction reduced cerebral blood flow in both groups (mean reduction ~10–15% in frontal and parietal regions), with no significant difference between groups. Caffeine did not prevent the blood flow reduction.

**Secondary outcomes — subjective sleepiness (KSS):**

Both groups reported increased sleepiness during sleep restriction (from ~3 at baseline to ~6–7 at day 5, p<0.001).

The caffeine group reported slightly lower sleepiness than placebo on days 1–3 of restriction (mean difference ~0.8 points on the 9-point scale, p=0.04), but this difference disappeared by days 4–5.

After recovery, sleepiness returned to baseline in both groups.

**Secondary outcomes — vigilance (PVT):**

Both groups showed significant increases in reaction time (mean increase ~30–50 ms) and lapses (from ~2 lapses at baseline to ~8–12 lapses at day 5) during sleep restriction (p<0.001 for both).

There was no significant difference between caffeine and placebo groups at any time point. Caffeine did not prevent the cognitive performance decline.

**Secondary outcomes — sleep quality (PSG):**

Total sleep time during restriction was similar between groups (~280–290 minutes per night, as per protocol).

The caffeine group showed significantly lower sleep efficiency (percentage of time in bed spent asleep) on nights 3 and 5 (mean 82% vs. 88% in placebo, p=0.02), and more wake after sleep onset (mean 25 min vs. 15 min, p=0.03).

Slow-wave sleep (deep sleep) was reduced in the caffeine group compared to placebo (mean 18% of total sleep time vs. 22%, p=0.04).

The grey matter thinning observed in the placebo group (0.02–0.05 mm over 5 nights) is roughly equivalent to the cortical thinning seen in 1–2 years of normal aging. For context, healthy adults lose approximately 0.01–0.02 mm of cortical thickness per decade after age 30. So 5 nights of sleep restriction accelerated this process by a factor of roughly 10–20.

Caffeine's protective effect was substantial: it reduced the number of affected regions from 17 to 2, and the magnitude of thinning in those regions was halved. This is a large effect — comparable to the difference between a healthy sleeper and someone with chronic insomnia.

However, the cognitive benefits were negligible: caffeine reduced subjective sleepiness by less than 1 point on a 9-point scale (barely noticeable), and had zero effect on reaction time or lapses. So the brain structure was "protected" while brain function was not.

The sleep disruption caused by caffeine was modest but real: ~6% lower sleep efficiency and ~10 minutes more wake time per night. This is roughly equivalent to the effect of one standard drink of alcohol before bed.

**Author-acknowledged limitations:**

Small sample size (n=36 total, n=18 per group), which limits statistical power for detecting small effects and increases the risk of false positives.

Short duration of sleep restriction (5 nights) — real-world chronic sleep restriction often lasts weeks or months.

Single dose of caffeine (300 mg/day) — cannot assess dose-response relationships or whether lower doses would have similar effects.

No "normal sleep + caffeine" control group, so we cannot separate the effects of caffeine during sleep restriction from caffeine's effects during normal sleep.

The PET radioligand [¹⁸F]CPFPX measures A₁ receptor availability, not adenosine levels directly. The interpretation that caffeine blocks adenosine-induced receptor upregulation is indirect.

Participants were young, healthy, and low-to-moderate caffeine consumers. Results may not generalise to older adults, heavy caffeine users, or people with sleep disorders.

**Critical reader observations:**

The study was funded by the German Research Foundation (DFG) and the University of Lübeck, with no apparent industry funding. However, the lead author has previously published on caffeine and adenosine receptors, which could introduce subtle bias in interpretation.

The lack of cognitive benefit despite "brain protection" is puzzling and raises questions about what the grey matter changes actually mean. Cortical thinning could be a normal adaptive response (e.g., synaptic pruning) rather than pathology. The authors interpret thinning as "damage," but this is an assumption.

The correlation between A₁ receptor upregulation and cortical thinning in the placebo group (r=0.52) is moderate and based on only 18 participants. This could easily be a spurious finding.

The study did not measure caffeine withdrawal. Participants were low-to-moderate consumers, but even 200 mg/day habitual use could produce withdrawal symptoms during the placebo condition, potentially confounding results.

The recovery period was only 3 nights — too short to assess whether the grey matter changes fully reverse or whether there are lasting effects.

For someone running their own n=1 experiment:

**What to test:**

The core question: Does daily caffeine consumption (e.g., 2–3 cups of coffee) protect against the negative effects of chronic sleep restriction on brain health, or does it simply mask sleepiness while disrupting sleep quality?

A useful self-experiment would compare two conditions: (1) 5–7 days of restricted sleep (e.g., 5–6 hours/night) with your usual caffeine intake, versus (2) the same sleep restriction with no caffeine (or decaf as placebo). You could also test a third condition: normal sleep with caffeine, to separate the effects.

**Minimum meaningful duration:**

At least 5–7 days of sleep restriction, based on this study's protocol. Shorter periods (1–2 nights) may not produce detectable grey matter changes.

Include a 3–7 day recovery period to see if any changes reverse.

Total experiment: ~2–3 weeks per condition (baseline, restriction, recovery).

**What to measure:**

**Subjective sleepiness:** Karolinska Sleepiness Scale (1–9) or Stanford Sleepiness Scale, taken 3–4 times daily (morning, midday, afternoon, evening).

**Cognitive performance:** Psychomotor Vigilance Task (PVT) — free apps exist (e.g., PVT+). Measure reaction time and lapses daily at the same time (e.g., 10:00 and 16:00).

**Sleep quality:** Sleep diary (time to bed, time to sleep, wake after sleep onset, total sleep time). If you have a wearable (e.g., Oura Ring, Fitbit, Whoop), track sleep efficiency and deep sleep percentage. Note: consumer wearables are less accurate than PSG but sufficient for n=1 tracking.

**Mood/energy:** Simple 1–10 rating of "how well-rested do you feel?" each morning.

**Caffeine intake:** Record exact timing and dose (mg). 300 mg/day is ~3 cups of brewed coffee or 2 large energy drinks.

**Key confounds to control for:**

**Caffeine withdrawal:** If you normally drink coffee, going cold turkey during the placebo condition will cause withdrawal headaches and fatigue that mimic sleep deprivation. Solution: use a washout period (3–5 days of no caffeine before starting each condition), or use decaf as a placebo.

**Timing:** Caffeine has a half-life of 4–6 hours. Taking it after 16:00 will disrupt sleep. In this study, the last dose was at 16:00. Stick to morning and early afternoon only.

**Sleep schedule:** Keep bedtime and wake time consistent across conditions. Use an alarm to enforce the restriction.

**Diet and exercise:** Keep these constant across conditions. Both affect sleep quality and cognitive performance.

**Expectation effects:** If you know you're in the "caffeine" condition, you might feel more alert due to expectation. Use a blinded design if possible (e.g., have a friend prepare identical capsules of caffeine and placebo, labelled A and B, and reveal the code after the experiment).

**What a positive result would look like:**

If caffeine is "protective": You would see less decline in subjective sleepiness and cognitive performance during the caffeine condition compared to placebo, and better sleep efficiency during recovery. However, this study suggests the opposite — caffeine may worsen sleep quality without improving cognition.

If caffeine is "harmful": You would see worse sleep efficiency, less deep sleep, and possibly worse cognitive performance in the caffeine condition, despite feeling less sleepy initially.

A truly informative result would show that the short-term alertness boost from caffeine is offset by poorer sleep quality and no net