Smoking normalizes cerebral blood flow and oxygen consumption after 12-hour abstention

Abstract

Acute nicotine administration stimulates [¹⁴C]deoxyglucose trapping in thalamus and other regions of rat brain, but acute effects of nicotine and smoking on energy metabolism have rarely been investigated in human brain by positron emission tomography (PET). We obtained quantitative PET measurements of cerebral blood flow (CBF) and metabolic rate of oxygen (CMRO₂) in 12 smokers who had refrained from smoking overnight, and in a historical group of nonsmokers, testing the prediction that overnight abstinence results in widespread, coupled reductions of CBF and CMRO₂. At the end of the abstention period, global grey-matter CBF and CMRO₂ were both reduced by 17% relative to nonsmokers. At 15 minutes after renewed smoking, global CBF had increased insignificantly, while global CMRO₂ had increased by 11%. Regional analysis showed that CMRO₂ had increased in the left putamen and thalamus, and in right posterior cortical regions at this time. At 60 and 105 minutes after smoking resumption, CBF had increased by 8% and CMRO₂ had increased by 11-12%. Thus, we find substantial and global impairment of CBF/CMRO₂ in abstaining smokers, and acute restoration by resumption of smoking. The reduced CBF and CMRO₂ during acute abstention may mediate the cognitive changes described in chronic smokers.

Introduction

Tobacco remains the most widely abused substance worldwide, and overwhelming evidence links its abuse to the nicotine content of tobacco.¹ The nicotinic acetylcholine receptor (nAChR) is a ligand-gated cation channel permissive to neuronal membrane depolarization and calcium influx; as such, acute nicotine increases brain energy metabolism, although longer-term nicotine exposure may desensitize the receptors. Indeed, autoradiographic studies with [¹⁴C]deoxyglucose show that acute nicotine stimulates the cerebral metabolic rate of glucose (CMR_glc) by approximately 50% in specific brain regions, notably in the thalamus, the interpeduncular nucleus, and in components of the visual system.^{2, 3, 4} In humans, the effects of nicotine and smoking on cerebral energy metabolism are poorly documented; in one study, intravenous nicotine evoked a global 10% decrease in CMR_glc in healthy male smokers,⁵ whereas other studies showed a global decrease but relative activations in the right thalamus and left cortical regions,⁶ or an association between reaction time and activation of increased [¹⁸F]fluorodeoxyglucose uptake in left thalamus.⁷ An early study based on the N₂O washout method and arteriovenous differences of oxygen revealed parallel 25% increases in global cerebral blood flow (CBF) and oxygen consumption (CMRO₂) after smoking cigarettes, indicative of coupled stimulation of perfusion and energy metabolism,⁸ although a still earlier study based on the radioactive Xenon method showed no changes in CMRO₂ after smoking.⁹ Regional CBF (rCBF) changes can provide a surrogate measure for coupled brain energy metabolism changes; acute nicotine administration evoked relative increases in rCBF in left visual cortex, thalamus, and in cerebellum.¹⁰ Reports of global CBF after smoking show a decline,⁸ rise,¹¹ or bimodal effects¹² in groups of young smokers. Smoking of nicotine cigarettes focally increased the normalized rCBF in visual cortex, thalamus, and cerebellum,¹³ as well as relatively decreased CBF in the anterior cingulate cortex ventral striatum,¹⁴ which roughly matches the pattern of activations in rat [¹⁴C]deoxyglucose autoradiography.

The equivocal state of knowledge of the effects of acute and chronic smoking on CBF and energy metabolism in human brain calls for further investigation. As a measure of changes of energy metabolism, CMRO₂ is held to be a more accurate index than CBF of the work of populations of neurons,^{15, 16} insofar as CMRO₂ mainly serves the maintenance of postsynaptic depolarization, while CBF changes mainly reflect afferent impulse activity.^{17, 18} For these reasons, stimulus-evoked changes of CBF and CMRO₂ are not always closely coupled, such that changes of CBF typically exceed those of CMRO₂. Consequently, evidence of CBF changes during and after smoking must be interpreted in the context of possible uncoupling of CBF changes from the CMRO₂ changes that may result from smoking-evoked alterations in net depolarization of brain tissue.

In the present study, we tested the hypothesis that overnight abstention from smoking by habitual smokers impairs CBF and metabolism. We first compared resting CBF and CMRO₂ in habitual smokers in a condition of abstention for 12 hours with historical positron emission tomography (PET) data from age-matched nonsmokers. We next tested the effects of resumed smoking on CBF and CMRO₂ in paired PET examinations at 15, 60, and 105 minutes after smoking one or two cigarettes, to test the hypotheses that resumption of smoking evokes coupled increases in flow and energy metabolism in withdrawn smokers, consistent with preclinical findings.

Based on a consensus of preclinical results and known cognitive effects of smoking in humans, we predicted coupled changes in a limited number of brain regions, i.e., thalamus (rich in nicotinic receptors), caudate nucleus, putamen, and cerebellum, cingulate cortex, and hippocampus. We also made a comparison of cortical thickness of the subjects' magnetic resonance imaging data, to exclude cerebral atrophy as a factor in the anticipated differences in CBF/CMRO₂ between smokers and nonsmokers.

Materials and methods

Historical PET and magnetic resonance imaging (MRI) control data were recovered from a group of 12 self-reported right-handed nonsmokers (6 men) aged 21 to 31 years (mean 26±3.5 (s.d.) years) who had participated in previous protocols.¹⁸ Previous exclusion criteria had included any present or past use of drugs or medications acting on the central nervous system, or any history of neuropsychiatric disorders or head injury.

Fourteen healthy right-handed smokers (9 men, 5 women) aged 22 to 33 years (mean 27±4 (s.d.) years) gave written informed consent to the protocol, approved by the Research Ethics Committee of County Aarhus (Central Region), Denmark. All experiments were performed in accordance with the guidelines and regulations described in Good Clinical Practice (GCP) documentation. The subjects underwent physical and neurologic examinations and reviews of medical history, and were all medication free except for oral contraceptives. They reported a mean daily consumption of 15±5 cigarettes per day for a mean of 10±5 years; 13 out of 14 smoked a national brand of cigarette delivering 1 mg nicotine and 10 mg tar; one subject smoked selfrolled cigarettes. All subjects were habitual morning smokers, reporting the first cigarette of the day within an hour of waking. Subjects agreed to abstain from smoking after midnight before the day of PET scanning. At 10:00 h of the day of scanning, overnight abstention was verified by carboxyhemoglobin measurement. Plasma nicotine or cotinine was not measured. At approximately 2 weeks before the PET study, we measured anxiety and depression with the version of the Bech Rating Scale of Mood Disorders in Danish (BRS).¹⁹ The BRS is a selfreported multiple-choice questionnaire consisting of 42 items, including the 21 items of the Beck Depression inventory and the 14 items of the Beck Anxiety Rating Scale.²⁰

Experimental Design

After completion of a brief baseline attenuation scan, the smoking subjects underwent baseline CBF and CMRO₂ sessions, each with a 3-minute dynamic emission recording. The two sessions were completed approximately 10 minutes apart (five half-lives), in a randomized sequence (Figure 1). On completion of the first session (withdrawal condition), subjects left the tomograph and resumed smoking ad libitum. All subjects reported cigarette craving at this time, and smoked 2±1 cigarettes in 30 minutes, at the end of which period they returned to the tomograph. Subjects then underwent a series of three pairs (in a randomized order) of CBF and CMRO₂ tomographies at 15±5, 60±6, and 105±6 minutes after completion of the last cigarette (Figure 1). During all emission acquisitions, subjects fixed their gaze on a cross-hair shown inside the tomograph, and noise in the scanning room was kept to a minimum. The entire postsmoking sequence of emission recordings was concluded by a second attenuation scan. Blood levels of carbon dioxide, oxygen, and carboxyhemoglobin were measured in blood samples collected during each of the CMRO₂ recordings.

Figure 1.

Flow chart depicting the study design. CBF, cerebral blood flow; CMRO₂, cerebral metabolic rate of oxygen.

Positron Emission Tomography Measures

We used the ECAT EXACT HR47 (CTI/Siemens) whole-body tomograph, operating in 3D-acquisition mode. Details of image reconstruction and filtering are described elsewhere.¹⁸ Each subject was positioned in the tomograph with the head comfortably immobilized in a custom-made head holder. A short indwelling catheter was placed into the left radial artery for blood sampling and blood gas examination. During each dynamic recording, the radioactivity concentration in arterial blood was measured continuously using an online gamma counter cross-calibrated to the tomograph. Subjects inhaled 1,000 MBq of [¹⁵O]O₂ in one breath at the start of each CMRO₂ emission recording, and received 500 MBq of H₂¹⁵O intravenously as a single bolus at the start of each CBF study. Parametric maps of CMRO₂ and CBF were calculated with an application of the two-compartment method.^{21, 22} To reduce the effect of [¹⁵O]O₂ distribution in the nasal sinus, which might bias the sensitivity for detecting small focal changes in brain uptake, subjects used a nasal clamp to minimize intranasal tracer accumulation. Each subject also underwent MRI in a 1.5-T magnet (GE) for structural-functional (MRI-PET) correlation with a T1-weighted, 3D fast-field echo sequence consisting of 160 256 × 256 slices of 1 mm thickness. The historic data of control subjects were also acquired on the same PET scanner and with the identical conditions to the current study.

Data Analysis

Magnetic resonance images were transformed into common stereotaxic coordinates using an automatic registration algorithm, and reconstructed PET images were coregistered with individual MRI scans using a program based on the AIR (Automatic Image Registration) method. Between-scan subject movements were also corrected using automatic PET-to-PET registration (details described in ref. 18). Global gray-matter CBF and CMRO₂ were determined for each subject using a binarized brain mask by exclusion of extracerebral and white-matter voxels followed by averaging all of intracerebral voxels (gray matter). We also determined gray-matter CBF and CMRO₂ values in 20 regions-of-interest outlined by means of anatomic masks generated in the previous studies. The gray-matter regions included frontal lobe (left and right), temporal lobe (left and right), occipital lobe (left and right), parietal lobe (left and right), cerebellum (left and right), putamen (left and right), caudate nucleus (left and right), cingulate (left and right), hippocampus (left and right), and thalamus (left and right). For the purposes of comparison, the global values of the baseline scans of a group of previously studied non-smoking subjects were also determined with the same procedure. Reconstructed PET images were normalized for global CBF and CMRO₂ averaged across subjects, transformed into stereotaxic coordinates and blurred with a Gaussian filter (14 × 14 × 14 mm). Mean subtracted-image volumes (postsmoking state minus withdrawn state) were converted to z-statistic. For this purpose, we used a modified statistical analysis based on the general linear model with correlated errors²³ to accommodate PET data. This analysis overcomes the weaknesses associated with pooled standard deviations by initially smoothing the standard deviation using a regularized variance ratio to increase the degrees of freedom. The result is mean subtracted, normalized image volumes (postsmoking minus withdrawn state) with relative changes of CBF and CMRO₂ calculated and presented as activation clusters measured in blood samples collected during each of the CMRO₂ recordings.

Cortical Thickness

We determined cortical thickness with the CIVET processing pipeline (Montreal Neurological Institute). T₁-weighted images were registered to the ICBM152 nonlinear sixth-generation template with a 9-parameter linear transformation and corrected for inhomogeneity.²⁴ Deformable models were then used to create white and gray-matter surfaces for each hemisphere separately, resulting in four surfaces of 40,962 vertices each.²⁵ From these surfaces, the t-link metric was derived for determining the distance between the white and gray surfaces. Non-normalized, native-space thickness values were used in all analyses owing to the poor correlation between cortical thickness and brain volume. To normalize for global brain size when it has little relevance to cortical thickness risks introducing noise and reducing power. A mid-surface (between pial and white-matter surfaces) was nonlinearly normalized using a novel depth potential function to a minimally-biased surface. All vertex-wise analyses were performed using the RMINC (https://launchpad.net/rminc) package and corrected for multiple comparisons using a false discovery rate of 10%.

Results

At the time of first interview, without previous instructions to abstain, smoking subjects were tested for the blood carbon monoxide (CO) content, based on the CO concentration in exhaled breath (DUOTEC, Glostrup, Denmark); the measured carboxyhemoglobin concentration was 4.5±1.1% at the time of interview. Blood gases were measured in arterial blood samples collected just before each PET session. One smoking volunteer had carboxyhemoglobin level suggestive of recent smoking (>3%), and was therefore excluded from the study, and another smoking subject was excluded due to technical failure, leaving a total of 12 subjects. In the historical control group, the mean (s.d.) blood gas results were PCO₂ 5.1±0.5 kPa, PO₂ 13.6±0.8 kPa, and oxyhemoglobin 95.5±2.8%. Corresponding blood gas and oximetry results for the smoking group did not significantly differ from these values at any of the four time points. More specifically, and to exclude the effect of PCO₂ on CBF, those parameters were measured for each scan, i.e., presmoking and postsmoking (P1, P2, and P3) conditions; there were no significant intrascan differences (PCO₂: pre 5.5±0.6 kPa, P1 5.4±0.4 kPa, P2 5.4±0.5 kPa, P3 5.2±0.4 kPa). Mean (s.d.) carboxyhemoglobin saturation was 0.9±0.2% for the retrospective control group, thus confirming their attestation of nonsmoking status. In the case of smokers, carboxyhemoglobin saturation was 1.7±0.1% in the withdrawn baseline, and 3.7±0.2% at 15 minutes, 3.4±0.2% at 60 minutes, and 3.3±0.2% at 105 minutes after smoking. Blood pH (7.4±0.1 in nonsmokers and in smokers at baseline) was unaffected by smoking (7.4±0.1). The mean hematocrit was 44±7% for smokers and 42±8% for nonsmokers, which did not differ significantly between the two groups. No smoking subject was excluded for depression or anxiety disorders according to the BRS scores (∑BRS<9).

The mean parametric maps of absolute CBF and CMRO₂ in the nonsmokers and smokers are shown in Figure 2. Visual inspection suggests that the global CBF and CMRO₂ of withdrawn smokers were distinctly lower than among age-matched nonsmokers, and rose toward the range for nonsmokers in the 105 minutes after smoking one or two cigarettes.

Figure 2.

Average (A) cerebral blood flow (CBF) and (B) cerebral metabolic rate of oxygen (CMRO₂) maps in 12 nonsmokers and in 12 smokers in a condition of abstinence and at intervals after smoking; (a) the nonsmokers, (b) the smokers after 12 hours abstinence, (c) at 15 minutes after smoking, (d) 60 minutes after smoking, and (e) 105 minutes after smoking.

Region of Interest Analysis

There was no effect of gender on measures of global gray-matter CBF or CMRO₂ (i.e., the average values of voxels in the whole-brain minus white matter) in either group by two-tailed t-test (P=0.2), and therefore the data of men and women were combined for all analyses. At abstinence baseline, global gray-matter CBF and CMRO₂ were 17% reduced relative to the nonsmoking control group (Table 1); global CMRO₂ (but not CBF) had significantly increased by 11% at 15 minutes after resumption of smoking; both global CBF and CMRO₂ continued to increase at 60 and 105 minutes of postsmoking. Although statistical tests revealed that increases in each of the conditions were not significantly different from each other (one-way repeated measures ANOVA; P>0.05), the ad hoc statistical tests showed a linear trend of increased global gray-matter CBF and CMRO₂ after smoking (Figure 3). At 105 minutes after resumption of smoking, the average global gray-matter values of CBF and CMRO₂ of smokers had approached the corresponding means for nonsmokers.

Table 1. Average global brain CBF and CMRO₂ values in nonsmoking controls and in habitual smokers before and after smoking (mean±standard deviation).

Condition/Group	CBF (ml/hg/min)	CMRO2 (μmol/hg/min)
Nonsmokers (N=12)	59±7	169±15
Smokers (N=12), 12 hours abstinence	49±7a (−17%)	140±13a (−17%)
15 minutes after smoking	51±5 (+4%)	155±17b (+11%)
60 minutes after smoking	53±7b (+8%)	153±19b (+8%)
105 minutes after smoking	55±6b (+12%)	161b±16b (+11%)

CBF, cerebral blood flow; CMRO₂, cerebral metabolic rate of oxygen.

^aPresmoking condition significantly different compared with nonsmoking control group.

^bStatistically significant compared with pre smoking condition (one-way ANOVA, P<0.05).

Figure 3.

The global (gray matter, GM) values of cerebral blood flow (CBF) and cerebral metabolic rate of oxygen (CMRO₂) of smokers at withdrawal and postsmoking states compared with those of nonsmokers. The data are presented as mean +/− standard deviation.

We also performed two-way ANOVA test to compare the average regional CBF and CMRO₂ differences in regions of interest (a) and forebrain cortices (b) addressed by the hypothesis as shown in Table 2 (and Supplementary Figure S1). For CBF, the two-way ANOVA of regions of interest addressed by the hypothesis, i.e., thalamus, as well as four forebrain cortices yielded no statistically significant differences between the values in any region or condition tested, nor did regional CBF at withdrawn baseline differ from that in nonsmokers. For CMRO₂, as shown in Tables 2a and 2b (and in Supplementary Figure S1), two-way ANOVA indicated significant increases in left thalamus, left caudate nucleus and in right occipital/parietal cortex at 15 minutes after smoking. Significantly increased CMRO₂ was seen in the majority of regions at brain 60 and 105 minutes after smoking, compared with abstinent baseline.

Table 2. Summary of two-way ANOVA results of the regional CMRO₂ differences in subcortical regions of interest (upper) and forebrain cortices (lower) addressed by hypothesis when baseline values of smokers were compared with CMRO₂ values from smoking after 15 minutes (pre vs. 1), 60 minutes (pre vs. 2), 105 minutes (pre vs. 3), and also with the baseline values of non- smokers (pre vs. non-smoker).

	Thalamus		Caudate		Putamen		Cerebellum		Hippo
	L	R	L	R	L	R	L	R	L	R
pre vs. 1	**	ns	ns	ns	*	ns	ns	ns	ns	ns
pre vs. 2	ns	ns	ns	ns	ns	*	ns	ns	ns	ns
pre vs. 3	***	ns	*	ns	**	**	*	*	ns	ns
pre vs. non smokers	***	*	**	ns	***	***	***	***	*	ns

	Cingulate		Frontal		Temporal		Occipital		Parietal
	L	R	L	R	L	R	L	R	L	R
pre vs. 1	ns	ns	ns	ns	ns	ns	ns	*	ns	*
pre vs. 2	ns	ns	ns	ns	ns	ns	ns	ns	ns	ns
pre vs. 3	**	*	*	**	*	**	**	***	**	**
pre vs. non smokers	**	***	*	**	*	**	**	***	***	***

*Denotes the significance. ns: denotes not significant.

Voxel-Based Analysis

Nonsmokers vs. smokers

Statistical parametric analysis (t-maps) of the contrast between nonsmokers and abstinent smokers revealed highly significant clusters of reduced CBF (Figure 4a) in two cortical regions of abstinent smokers, i.e., fronto-parietal (x=−50; y=49; z=−36; Talairach coordinates of the peak voxel of the cluster) and occipital (x=23; y=−93; z=−39) regions, and a significant CMRO₂ cluster (Figure 4b) encompassing almost the entire cerebral cortex (x=−35; y=−11; z=73y), and extending into subcortical regions.

Figure 4.

Statistical parametric comparison (t-maps) of baseline scans of nonsmokers vs. baseline scans of smokers (A CBF; B CMRO₂). As seen in the figure, two significant clusters of CBF in the frontal/parietal and occipital regions (A) and one highly significant cluster of CMRO₂ change in the entire cortical mantel were observed (B). CBF, cerebral blood flow; CMRO₂, cerebral metabolic rate of oxygen.

Postsmoking condition vs. baseline of smokers

During the first (15 minutes) and second (60 minutes) postsmoking, CBF and CMRO₂ there was increases in several frontal and occipital cortical voxel clusters relative to abstinence baseline, which did not reach the chosen threshold of statistical significance. During the third postsmoking tomography session (105 minutes) compared with baseline, normalized CBF remained increased in the voxel clusters seen at 15 and 60 minutes after smoking, whereas CMRO₂ was significantly increased in a cluster of voxels (P=0.02) in the supra-lateral surface of brain (close to the central sulcus, at x=0; y=−34; z=84).

Structural Analysis

To rule out significant morphologic changes of a magnitude sufficient to affect the quantitative results, we tested the cortical thickness by analysis of MRI scans of smokers vs. nonsmokers covaried for age and sex and corrected for multiple comparison at 10% false discovery rate. The analysis revealed one area of significantly different thickness, a reduction in the left cingulate gyrus of the smokers compared with the nonsmokers (Figure 5).

Figure 5.

Cortical thickness analysis of smokers vs. nonsmokers covaried for age and sex. Correction for multiple comparisons at 10% false discovery rate (FDR) revealed only significant change in the thickness of left cingulate gyrus of smokers compared with the nonsmokers.

Discussion

Consistent with the hypothesis that abstention from smoking by habitual smokers would impair brain perfusion and energy metabolism, we made the observation of substantial global declines of gray-matter CBF and CMRO₂ in a group of habitual smokers after 12 hours of smoking abstinence, compared with nonsmoking control subjects. Cortical thickness analysis of magnetic resonance images showed cortical atrophy in only one region (the left cingulate cortex) of smokers, such that the global decrease in abstemious smokers cannot be attributed to cortical atrophy. In contrast, a previous study of global brain perfusion showed a global 12% decrease in CBF and an accelerated age-dependent decline in global CBF among smokers,²⁶ which, in a cross-sectional study, normalized after 1 year of abstinence.²⁷ The alternative interpretation that habitual smokers have lower CBF and CMRO₂ also when they smoke was ruled out by the rapid return to values close to those of the control group after resumption of smoking. The rapid reversal is consistent with the stimulant effect of nicotine or other components of tobacco smoke on flow-metabolism coupling in brain, as also indicated by the reversible effects of smoking on cardiovascular fitness.²⁸

Previous reports of effects of chronic smoking on CMRO₂ are more equivocal, with only two disparate reports, showing global CMRO₂ to be reduced⁸ or unaffected,⁹ both at variance with the present study which revealed a 17% reduction of CMRO₂ in gray matter after 12 hours of abstinence. The decline of global CMRO₂ of the abstaining smokers is comparable to the decline seen in four decades of healthy aging.^{29, 30} This extraordinary result indicates that brain energy metabolism is distinctly compromised in withdrawn smokers, which may have implications for the cognitive function of abstaining smokers. Nicotine's support of cognition is well recognized, and acute deficits in sustained attention and working memory occur in nicotine withdrawal.³¹ The present observations predict a correlation between cognitive function, especially executive and visual functions, and regional brain energy metabolism (CMRO₂) in abstaining smokers.

We note that the exact mechanisms by which nicotine or smoking change CBF and CMRO₂ are not understood clearly. The presynaptic nicotinic acetylcholine receptors on nitrergic fibers may mediate nicotine-induced increases in CBF in rat cerebral cortex.³² The most abundant subtype of nicotinic acetylcholine receptors in brain is α₄β₂, which are of notably high concentration in the thalamus, where they presumably drive local cerebrometabolic effects of nicotine, as seen in [¹⁴C]deoxyglucose autoradiographic studies. The α₄β₂ nAhRs are rather homogeneously distributed elsewhere. The next most abundant type is the α₇ homomeric receptor, which make up about 10% of nicotinic acetylcholine receptors .³³ The efficacies of nicotine at the several nicotinic acetylcholine receptors receptor subtypes with respect to CBF and cerebrometabolic effects are unknown. Furthermore, nicotine possesses both rewarding and aversive stimulus properties, which are mediated by different components of the mesolimbic dopamine pathway,³⁴ and likely involving the α₆* nAChR subtype.

The lack of widespread cerebral atrophy in the present group of young smokers rules of the possibility that impairment of global CBF and CMRO₂ arises from loss of gray matter. Previous volumetric studies have shown frontal cortical and subcortical atrophy in brain of elderly smokers,³⁵ and atrophy is also reported in the left anterior cingulate of smokers,³⁶ close to where we identify a focal decline in cortical thickness. Atrophy of this domain of the cingulate gyrus is associated with the cognitive impairment in patients with schizophrenia,³⁷ which may predict specific cognitive and attentional deficits in psychiatrically normal young smokers, and may also be consistent with observation that part of the ‘executive' cognitive deficits seen in schizophrenic patients are attributed to effects of smoking per se.³⁸ However, the rapid 2-hour reversal of the flow and metabolism effects upon resumed smoking makes atrophy an unlikely explanation of the decline observed in abstaining smokers.

The study is the first to investigate the time course of smoking-induced global gray-matter CBF changes, and is the first study of effects of smoking on cerebral oxygen consumption in more than 40 years. The arterial plasma nicotine concentration peaks within minutes of smoking, and subsequently declines with a half-life of 10 minutes. In contrast, the venous nicotine concentration declines with a half-life of approximately 2 hours after smoking, a pharmacokinetic difference related to the rapid deposition of arterial nicotine into brain and other tissues.³⁹ Although we did not attempt to measure plasma nicotine, these studies together suggest that the arterial plasma nicotine concentration exceeded 50 nmol/L during the entire PET recording interval after resumption of smoking. Single photon emission computed tomography studies have revealed 70% occupancy of α₄β₂ nicotinic receptors in human brain persisting for at least 5 hours after smoking to satiety.⁴⁰ Thus, we argue that a significant occupation of nicotinic receptors persisted in the present study after resumption of smoking ad libitum.

In the first PET session after resumption of smoking, we found rapid reversal of the CMRO₂ decline in regions of interest of the left thalamus and putamen, and in the right occipital and parietal cortices. The reversal is in partial agreement with the predicted acute effects of nicotine on energy metabolism, based on [¹⁴C]deoxyglucose autoradiography in rats.^{2, 3, 4} There is some evidence of tolerance of the metabolic responsiveness of rat brain to nicotine challenge after chronic nicotine administration.⁴¹ The phenomenon of tolerance may account for the small increases in global CBF and CMRO₂ (<12%) seen after resumption of smoking by the habitual smokers after the overnight abstinence (Table 1). However, another factor potentially limiting the sensitivity of present methods is the very focal extent of activations reported in the [¹⁴C]deoxyglucose autoradiography literature; although activations in rCMRglc are as high as 50% to 100% in some regions, the larger increases generally were confined to structures, such as divisions of the thalamus, the superior colliculus, and the interpeduncular nucleus, which are too small to discern with present PET methods; what emerges from the present study is the general finding of global and coupled increases in CBF and CMRO₂, developing during nearly 2 hours after resumption of smoking.

Discrepancies in the rat and human literature of cerebrometabolic effects of nicotine and smoking on brain energy metabolism may also reflect physiologic distinctions between glucose and oxygen consumption. Whereas CMRglc measures likely reflect the composite of aerobic glycolysis and oxidative phosphorylation,^{17, 42} cerebral metabolic rate of oxygen is an indirect measure of the work done by a population of neurons, which is uniquely responsive to local rates of ATP turnover in dendritic spines and other sites rich in cytochrome oxidase.^{17, 43} We cannot account for the earlier reports of globally decreased CMRglc in human brain after intravenous nicotine administration,^{5, 6} nor for the report of inverse correlation between craving and reductions in normalized rCMRglc in ‘reward' brain regions (ventral striatum, orbitofrontal cortex) upon withdrawal from nicotine cigarettes.⁴⁴ We note above that the rather limited quantitative [¹⁸F]fluorodeoxyglucose-PET literature in humans does not perfectly match the focal activations of rCMRglc seen in rat autoradiographic studies with nicotine challenge. However, the general pattern of acutely increased CMRO₂ in the present study is consistent with one of the two earlier reports on global oxygen consumption (conducted more than three decades ago), and may suggest a rectification of ATP generation in the hours after smoking.

We found that 15 minutes after smoking there was a numerical, but statistically insignificant, increase of CBF in the occipital region (cuneus), a site where others found significant relative (globally normalized) increases at a similar brief interval after smoking nicotine-containing cigarettes.¹³ Others have reported that nicotine infusion increased relative CBF in occipital cortex, as well as the anterior cingulate cortex and cerebellum.⁴⁵ Indeed, in our study there were no significant changes of CBF in any brain region at 15 minutes after smoking, although there were significant global increases at 60 and 105 minutes that followed the insignificant trend of a 4% increase at 15 minutes. This finding of a gradual increase in global CBF in the hours after smoking stands in particular contrast to a study with radioactive xenon study, which showed an acute 40% decrease in global perfusion.⁴⁶ This difference seems too large to be attributable to methodological factors, but may point to important differences between the CBF effects of nicotine infusion vs. smoking of cigarettes. The trends toward regional and global CBF increases at 15 minutes and onwards after smoking paralleled the rapid activation of CMRO₂; this is apparently related to the inherently greater variance of absolute CBF between individuals, which favored the detection of CMRO₂ changes. We chose to use quantitative methods with strict statistical threshold so as to gain physiologic accuracy, with a certain penalty in precision and sensitivity relative to that with normalization. This in turn may also explain the regional differences between CBF and CMRO₂ seen in Figure 4 since CBF is inherently more variable between individuals, so fewer regional differences will be picked up due to the lesser statistical power. The parallel upwards drifts in regional and global CBF and CMRO₂ in the 2 hours after resumption of smoking is consistent with the hypothesis and with the physiologic effects of nicotine, which is known to mediate fast signalling and depolarization in brain, both of which are activators of ATP consumption. Thus, in the third postsmoking session, CMRO₂ was restored to the magnitude seen in nonsmokers. Some factor in addition to the psychopharmacology of nicotine may have contributed to these observations. We speculate that an additional effect may have been mediated by another constituent of tobacco smoke, such as carbon monoxide, which was several-fold elevated in blood during this same interval after smoking. In addition to persistent association with hemoglobin, carbon monoxide has a signalling role in nervous tissue, and at low doses directly stimulates mitochondrial cytochrome c by an unknown mechanism, which conceivably can account for the present observation of slow restoration of oxidative metabolism in brain of habitual smokers. The magnitude of the coupled increase of global CBF and oxygen consumption may parallel normalization of cognitive function in smokers after a period of abstinence; a proper understanding of the underlying mechanisms might present new therapies for smoking cessation, other than generally ineffective nicotine substitution.

The authors declare no conflict of interest.