Caffeine in cerebrovascular research: To withdraw or not to withdraw?

Caffeine is among the most widely used substances, with ∼80% of the global population consuming an average of 200 mg/day (Samoggia & Rezzaghi, 2021). Following consumption (e.g., through coffee or tea), caffeine reaches maximal plasma concentrations within 30–120 min and has a half-life of 4–6 h (Fredholm et al., 1999). The physiological effects of caffeine are largely attributed to its action as an adenosine receptor antagonist, particularly on the A1 and A2 subtypes (Ribeiro & Sebastião, 2010). In this respect, adenosine is a multifaceted neuromodulator, in addition to a nucleoside with potent vasoactive properties, depending on the target receptor and organ (Benarroch, 2009). Of particular importance in the field of cerebrovascular physiology is to consider: (1) the direct cerebrovascular modulatory effects of caffeine; (2) the indirect cerebrovascular effects (cerebral perfusion pressure and ventilation) of caffeine; and (3) the effects of withdrawal. We argue that this last component is often ignored in research settings, because most studies adhere to a 6–24 h withdrawal period. Indeed, controlling for caffeine is complicated by its ubiquity in use, which begs the question of what the ‘true’ cerebral blood flow (CBF) baseline response is. The physiological impacts of withdrawal must be considered, and it is essential to evaluate whether the normal consumption or, in most cases, abnormal abstention of caffeine in research best reflects lived physiology. Caffeine is a non-selective antagonist of adenosine and interacts with cerebral A1, A2A and A2B receptors, leading to neural activation in the A1 subtype and vascular activation in the A2 subtypes (Addicott et al., 2009; Dodd et al., 2015). Importantly, the A2A and A2B receptors are present in the smooth muscle of the cerebrovasculature, with the A2A subtype located predominantly in pial arteries and arterioles, whereas the A2B receptors are situated primarily on parenchymal vessels (Addicott et al., 2009; Pelligrino et al., 2010). In short, when caffeine binds to A2 subtypes, adenosine-induced vasodilatation (Addicott et al., 2009) is inhibited, leading to increased cerebrovascular tone (Dodd et al., 2015). Adenosine receptors (primarily A2A) are also abundant in the peripheral vasculature. Caffeine-mediated inhibition of these receptors results in an increase in systemic vascular resistance and blood pressure (Pincomb et al., 1988; Umemura et al., 2006). In addition to the antagonistic effects of caffeine, it also stimulates the CNS, thereby increasing circulating catecholamines, such as noradrenaline and adrenaline, which can increase heart rate, myocardial contractility and alveolar ventilation (Papadelis et al., 2003; Pincomb et al., 1988). Conceptually, caffeine can alter CBF by: (1) cerebrovascular constriction via inhibition of adenosine-mediated dilatation; (2) increased cerebral perfusion pressure via increased arterial blood pressure; and (3) increased cerebrovascular resistance if an increase in respiration lowers the arterial partial pressure of CO2 ( P aC O 2 ${P_{{\mathrm{aC}}{{\mathrm{O}}_2}}}$ ). En masse, the direct cerebrovascular effects of adenosine inhibition combined with a small reduction in P aC O 2 ${P_{{\mathrm{aC}}{{\mathrm{O}}_2}}}$ appear to outweigh the influence of increased cerebral perfusion pressure. For example, acute caffeine consumption of 250 mg reduces resting CBF by 20%–30% (Addicott et al., 2009; Turnbull et al., 2017). Importantly, the magnitude of the decline is contingent upon the amount of caffeine regularly consumed; i.e., inexperienced or low (e.g., 45 mg/day) users exhibit heightened sensitivity to the effects of caffeine (Addicott et al., 2009; Turnbull et al., 2017). Chronic caffeine exposure leads to the development of tolerance and a reduction in cardiovascular perturbations (Addicott et al., 2009; Turnbull et al., 2017). Indeed, participants who were administered caffeine equivalent to their usual daily intake (an average of 333 mg/day), with half the dose administered in the morning and the other half in the afternoon, did not exhibit any significant differences in CBF when compared with baseline measurements after the afternoon dose (Jones et al., 2000). Likewise, caffeine consumption of 150 mg resulted in a decrease in CBF 30 min after consumption; however, it returned to values comparable to ‘baseline’ 2 h after intake (Couturier et al., 1997). Furthermore, although naive users exhibit considerable decreases in CBF after 75 mg of caffeine, this dose had no impact on habitual users (Kennedy & Haskell, 2011). The reduction in CBF with caffeine consumption appears to be mediated primarily by the direct vasoconstricting effects of inhibiting adenosine reception and only secondarily from the increase in minute ventilation. For example, an acute dose of 250 mg of caffeine has been found to increase minute ventilation by ∼6% (Richmond, 1949), which would correspond to a reduction in partial pressure of arterial carbon dioxide ( P aC O 2 ${P_{{\mathrm{aC}}{{\mathrm{O}}_2}}}$ ) by slightly <6% (where P aC O 2 = V ̇ C O 2 / V ̇ C O 2 V ̇ A V ̇ A ${P_{{\mathrm{aC}}{{\mathrm{O}}_2}}} = {{{{\dot V}_{{\mathrm{C}}{{\mathrm{O}}_2}}}} \mathord{/ {\vphantom {{{{\dot V}_{{\mathrm{C}}{{\mathrm{O}}_2}}}} {{{\dot V}_{\mathrm{A}}}}}} \kern-\nulldelimiterspace} {{{\dot V}_{\mathrm{A}}}}}$ (rate of alveolar ventilation), i.e., the alveolar ventilation equation), but only if assuming that rate of carbon dioxide production ( V ̇ C O 2 ${\dot V_{{\mathrm{C}}{{\mathrm{O}}_2}}}$ ) remains constant. Yet, caffeine increases the V ̇ C O 2 ${\dot V_{{\mathrm{C}}{{\mathrm{O}}_2}}}$ (Powers et al., 1986). Ultimately, there are limited data where end-tidal partial pressure of CO2 (or better, P aC O 2 ${P_{{\mathrm{aC}}{{\mathrm{O}}_2}}}$ ) is measured after caffeine ingestion (and breathing room air), except following a large dose of 650 mg, whereby end-tidal partial pressure of CO2 decreased by ∼4 mmHg (D'Urzo et al., 1990). On average, this alone would decrease the CBF by 4%–12% (Willie et al., 2014), but it should be noted that 650 mg of caffeine is a very high dose. As such, at least within normal doses, the ventilatory impacts of caffeine on CBF are likely to be minor. The inability of the increase in cerebral perfusion pressure to offset reductions in CBF with caffeine is likely to be related to the static cerebral autoregulatory mechanisms, whereby CBF remains relatively constant with gradual changes in mean arterial pressure within ∼±10 mmHg (Brassard et al., 2021). An acute 200–250 mg dose of caffeine increases systolic pressure by only 3–14 mmHg and diastolic pressure by 4–13 mmHg (Grant et al., 2018). Importantly, similar to the direct cerebrovascular responses, the range of the pressor response is dependent on the tolerance to caffeine, whereby it is less in habitual users (Turnbull et al., 2017). Chronic caffeine use appears to increase the number and activity of adenosine receptors (Addicott et al., 2009). Specifically, chronic consumption of caffeine leads to an enhancement in the expression of A1A receptors in mice (Johansson et al., 1993; Shi & Daly, 1999), in addition to elevated plasma levels of adenosine (Ribeiro & Sebastião, 2010) and increased adenosine sensitivity (Addicott et al., 2009). Consequently, caffeine must then compete with adenosine, which has a high affinity for A1A and A2A receptors (Ribeiro & Sebastião, 2010), ultimately resulting in diminished effects (tolerance). Although this attribute of caffeine tolerance might reduce the physiological effects of caffeine, such as decreased vasoconstriction, the increase in adenosine receptors might also amplify the effects of adenosine during caffeine withdrawal (Addicott et al., 2009). Increased adenosine signalling occurs, owing to an increased probability of receptor–ligand interactions (i.e., caffeine-induced increased adenosine sensitivity) that might exacerbate the physiological symptoms of withdrawal. For example, in regular caffeine users, withdrawal leads to an amplification of CBF corresponding to the severity of withdrawal headaches (Addicott et al., 2009). For regular users, withdrawal symptoms can appear within 12–24 h and can persist for 2–9 days, resulting in reduced reaction times, decreased alertness and flu-like symptoms (Juliano & Griffiths, 2004; Juliano et al., 2019). Withdrawal is also frequently associated with symptoms such as irritability, anxiety, insomnia, fatigue, and moderate to severe headaches (Couturier et al., 1997; Dews et al., 2002). The psychological dependence on caffeine alone can also influence physiological withdrawal symptoms. For instance, participants who received decaffeinated coffee but were informed that it contained caffeine reported fewer withdrawal symptoms in comparison to those who were told it was decaffeinated (Juliano et al., 2019). With respect to CBF, caffeine withdrawal (i.e., 24 h) has been demonstrated to increase cerebral blood velocity in the middle, basilar and posterior cerebral arteries by ∼15% and to decrease pulsatility in the middle cerebral arteries (Couturier et al., 1997; Jones et al., 2000). Habitual caffeine consumers also demonstrate a reduction in blood pressure within 12–24 h, whereby mean arterial pressure declines by ∼5–6 mmHg on average (Lane, 1997; Phillips-Bute & Lane, 1997). Not surprisingly, the impact of caffeine withdrawal on CBF is also dependent upon the quantity that is regularly ingested, with increased consumption leading to worsened withdrawal symptoms and elevated resting CBF (Addicott et al., 2009; Dodd et al., 2015). Given the implications of both caffeine consumption and withdrawal on CBF, the question remains: should research participants refrain from or be allowed caffeine consumption prior to experimental studies on cerebrovascular haemodynamics? Common practice is to refrain from caffeine consumption for 6–12 h prior to experimentation to capture the ‘baseline’ state of human physiology. However, this assumption does not hold true for habitual caffeine consumers. Instead, at minimum, it reflects a state of withdrawal, but it also fails to represent accurately the ‘normal physiology’ encountered in the daily life of a caffeine user. Cerebrovascular tone is largely regulated by four interconnected factors: blood gases, metabolism, perfusion pressure and autonomic tone. Cerebrovascular reactivity, neurovascular coupling (NVC) and cerebral autoregulation (CA) testing are widely recognized assessments used to evaluate these four factors through vascular responsiveness to blood gases (CO2 and O2), neuronal activation/metabolism and perfusion pressure, respectively (Willie et al., 2014). With respect to cerebrovascular CO2 reactivity, ∼150–300 mg of caffeine does not appear to influence the change in CBF (Blaha et al., 2007; Chen & Parrish, 2009b). Although adenosine has been implicated as one of many tenable signalling molecules for cerebral vasomotor changes in response to alterations in P aC O 2 ${P_{{\mathrm{aC}}{{\mathrm{O}}_2}}}$ (Hoiland et al., 2019), these data indicate that it is likely to be of lesser importance than other known factors, including nitric oxide and arachidonic acid metabolites. Likewise, and perhaps even more surprisingly, theophylline, a potent adenosine antagonist with properties similar to caffeine, does not impact the vasomotor response to hypoxaemia (Bowton et al., 1988; Hoiland et al., 2017). In turn, these findings suggest that caffeine consumption might not be a major concern in studies where cerebrovascular reactivity is a primary outcome. However, the complete cessation of caffeine consumption for 3 months has been shown to improve cerebrovascular reactivity to CO2 (Gil et al., 2022), at least in participants suffering from migraines. Regarding NVC, the process by which increased neuronal activity triggers localized increase in CBF, caffeine causes a neurovascular ‘uncoupling’ effect (Chen & Parrish, 2009b; Dodd et al., 2015). Collectively, despite reductions in CBF, whole-brain cerebral metabolism is fully compensated for via an increase in the fraction of oxygen extracted (Xu et al., 2015). Yet, caffeine withdrawal can be equally problematic, particularly in chronic users. The elevated adenosine sensitivity that comes with chronic caffeine use might alter NVC during caffeine withdrawal because heightened adenosine signalling could enhance CBF through vasodilatation (Zhu et al., 2022). Additionally, adenosine is thought to inhibit glutamate release in presynaptic neurons, which is essential to the NVC response (Nam et al., 2012). Consequently, an increase in adenosine signalling might enhance these vasodilatory effects while simultaneously inhibiting glutamate release, resulting in a discrepancy between CBF and neuronal activity (Nam et al., 2012). Lastly, there are no studies directly examining the effects of caffeine on CA. However, the known effects of caffeine to increase mean arterial pressure in a dose-dependent manner (Mesas et al., 2011) or of caffeine withdrawal to reduce mean arterial pressure must be considered. Moreover, the stimulatory effects of caffeine on the sympathetic nervous system might alter the dynamic elements of CA within the autoregulatory zone (Brassard et al., 2021; Hamner & Tan, 2014). Attempts to control the cardiovascular effects of caffeine have become common practice, but the protocols vary considerably, potentially resulting in varying physiological states and thereby leading to inconsistent findings. In an effort to control for caffeine, the specified duration of abstention from caffeine typically ranges from 1 to 24 h (Grant et al., 2018). Notably, protocols for abstention that correspond to the peak duration of caffeine concentrations in the bloodstream (for example, 30–120 min) are inadequate, because caffeine will be most impactful during this time (Grant et al., 2018; Nowaczewska et al., 2020). Conversely, caffeine withdrawal symptoms typically manifest ∼12 h after cessation of use, and as such, protocols requiring a duration of >12 h will also produce unfavourable outcomes. Therefore, the optimal protocol to limit the effects of caffeine appears to be ∼6 h (Grant et al., 2018). Certainly, the duration of abstention depends largely on the time of day when the experimentation occurs. For instance, even if a protocol requires participants to refrain from consumption of caffeine for a duration of 6 h prior to testing, participants are generally less inclined to ingest caffeine in the evening and tend to postpone it until the following morning. As such, the period of abstention extends beyond the anticipated duration, probably entering a time frame of withdrawal. Abstinence for 6 h is feasible provided that the experimental setting can occur in the afternoon (e.g., the final ‘dose’ of caffeine taken at 09.00 h, with the trial commencing at 15.00 h); however, this approach would not be appropriate for research requiring overnight fasting or where diurnal variations apply (e.g., with cerebrovascular research). When morning testing is necessary, we argue that the caffeinated state is a more accurate depiction of the everyday physiological condition (at least in habitual users), and caffeine should, therefore, be allowed prior to testing, as long as the amount is minimized (<100 mg), such as one cup (250 mL) of brewed coffee (∼92 mg of caffeine) or a 30 mL espresso (∼63 mg of caffeine) (van Dam & Hu, 2022). Limiting the caffeine to <100 mg prevents participants from entering a withdrawal state without a major impact on CBF (Chen & Parrish, 2009a). The research question at hand must be considered initially, but comparably important is the population being studied. For instance, older adults typically exhibit heightened sensitivity to the cardiovascular effects of caffeine in comparison to younger participants, as evidenced by a more pronounced increase in blood pressure (Arciero & Ormsbee, 2009). Additionally, obese individuals tend to have a diminished metabolic clearance rate and an extended half-life of caffeine in comparison to non-obese individuals (Starling-Soares et al., 2023). The source of caffeine, and substances commonly present in caffeinated beverages, including l-theanine, sucrose and taurine (Costa et al., 2023; Dodd et al., 2015), should also be considered, because it will impact the pharmacokinetics and bioavailability. A substantial proportion of the population regularly consumes caffeine, with an average intake of ∼200 mg per day (Samoggia & Rezzaghi, 2021). Given the impracticality of exclusively recruiting participants who abstain from caffeine and acknowledging the significant prevalence of caffeine consumption, an essential question arises when conducting cerebrovascular research: what defines the ‘normal’ physiological state? In other words, does research requiring participants to abstain from caffeine accurately depict the daily physiological conditions experienced by the general population? This question is indeed complex and lacks a straightforward answer. The type and quantity of caffeine use vary considerably, but little is known about the resultant quantity and severity of withdrawal symptoms and their effects on study variables. In habitual caffeine consumers, we propose that being in a caffeinated state better represents daily physiology and has a lesser impact on research results in comparison to withdrawal, provided the caffeine intake is <100 mg within 6 h before the study. Nevertheless, experimental protocols should be tailored to align with the primary research questions, giving thorough consideration to the potential impacts of both caffeine intake and withdrawal on the measured outcomes. Brooke R. Shepley and Anthony Richard Bain contributed to the conception of the Editorial. Each author participated in designing, writing and revising the Editorial, approved the final version and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed. The authors declare no conflicts of interest. None.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.