INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chronic obstructive pulmonary disease (COPD) is a major cause of respiratory morbidity and mortality (1), with exacerbations of COPD accounting for over half of all acute medical admissions for respiratory diseases (2). Although COPD is a slowly progressive disease, acute infective exacerbations are associated with worsening hypoxia, hypercapnia, and respiratory acidosis, and have a high mortality (3).

Patients with chronic stable hypoxia and hypercapnia have grossly normal mental function compatible with daily living. Normal cerebral cell function is critically dependent on intracellular energy metabolism via oxidative phosphorylation and the production of adenosine triphosphate (ATP). We hypothesized that for normal function to be preserved under hypoxic conditions, alternative anaerobic pathways of energy production must be utilized. We hypothesized that there is an increasing dependence on glycolysis as such a pathway. This would be associated with increased activity of the enzyme phosphofructokinase, overproduction of glycolytic pathway metabolites, and an increased proportion of intracellular phosphate in the cytosol. This last change would occur because in the cytosol, glycolysis produces ATP, whereas the mitochondrion is the site of oxidative phosphorylation. In patients with stable COPD, the indices of total energy production, such as phosphocreatine (PCr), would be expected to be normal.

Energy production via glycolysis is relatively inefficient and produces lactate, which in turn requires intracellular buffering to prevent acidosis. An understanding of cerebral bioenergetics may thus illuminate the biochemical compensatory processes that occur in stable hypoxic patients.

Cerebral ³¹P magnetic resonance spectroscopy (³¹P MRS) offers the first opportunity to investigate abnormalities of intracerebral biochemistry in chronically hypoxic patients. ³¹P MRS is a noninvasive technique that allows cerebral metabolism to be studied in vivo in humans, providing information about cerebral phospholipids, glycolytic intermediates, and high-energy phosphates such as PCr and ATP (4). The aims of the present study were to identify changes in ³¹P MR spectra obtained from the brains of patients with stable COPD, and to compare the results with MR spectra from the same areas of the brain in healthy control subjects.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ten patients (six males and four females) with stable COPD and five healthy controls (three males and two females) were recruited for study. COPD was defined as evidence of chronic and generalized airflow limitation (FEV₁ < 80% predicted and FEV₁/FVC < 70% predicted) with long-term variability in FEV₁ or peak expiratory flow rate of < 20% and acute ₂-agonist reversibility of < 15% (5). No patient had an acute exacerbation of COPD (defined as hospital admission or prescription of antibiotics/systemic corticosteroids by a general practitioner) over the 3 mo prior to the study. Patients with other, coexistent pulmonary diseases, congestive heart failure, cirrhosis, dementia, and recent or old stroke, and those with claustrophobia, cardiac pacemakers, and ferromagnetic implants were excluded. All patients had 42 ± 10 pack-years (mean ± SD) of smoking, and were taking bronchodilators (seven through regular home nebulizers), but none was receiving long-term home oxygen therapy. The control subjects were asymptomatic, healthy nonsmokers (never smoked/> 20 yr ex-smokers), and were taking no medications. All subjects gave written informed consent, and ethical approval for the study was obtained from the local ethics committees.

Spirometry and arterial blood gas measurements were done on the day of the study for all patients. Spirometry was not available for controls. Arterial blood gases were measured in all but one control subject, who denied the necessary arterial puncture.

All MRS data were obtained with a Picker (Cleveland, OH) prototype spectroscopy system, using a whole-body magnet (Oxford Magnet Technology, Oxford, UK) operating at 1.5 T. A ¹H/³¹P enveloping birdcage coil was used, which comfortably encompassed the entire head. The proton signal was used for shimming and to acquire T₁-weighted imaging (relaxation time [TR] = 300 ms, excitation time [TE] = 22 ms) to verify spectral localization. Localized ³¹P MR spectra were acquired simultaneously from a series of parallel transverse planes with a nominal resolution of 2 cm, using a one-dimensional chemical shift imaging (CSI) technique (6) with a TR of 5,000 ms. The total examination time was approximately 20 min. The MR technique has a coefficient of variation of less than 3% (7)

The ³¹P MR spectra were processed with a 120-ms exponential filter, and were manually phased. The baseline roll was removed (8) and the peaks were fitted to inverse polynomial functions by a single observer (blinded to the subject group) using the NMR1 spectral processing program (New Methods Research Inc., East Syracuse, NY). Metabolite peak-area ratios were expressed with respect to ATP, and in accord with current normal practice, relative percentages of phosphomonoesters (PMEs), inorganic phosphate (Pi), phosphodiesters (PDEs), PCr, and ATP were also measured with respect to the total ³¹P MRS signal (9). Intracellular brain pH was also calculated (10). For each subject, spectral parameters from the three central 2-cm transverse planes, at the level of the basal ganglia, lateral ventricles, and centrum semiovale of the brain, were averaged. Changes in patient spectra were compared with those of controls.

Statistical tests were done through multiple regression, unpaired t tests, and multivariate analysis, as appropriate, using the SPSS-PC software system (SPSS Inc., Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary Function

Table 1 gives the characteristics of patients and controls.

MR Spectroscopy

All subjects had an MRS examination performed at the same time of day (in the afternoon). When Pi an PMEs were expressed as percentages of the total MR-detectable phosphorus signal, COPD patients had a significant increase in both, but a lower signal for PDEs (Figure 1). COPD patients also had higher Pi-to-ATP and PME-to-ATP ratios, but a lower PCr- to-Pi ratio (Figure 2) than did controls. Other variables, including intracellular brain pH, were not significantly different for patients and controls (Table 2).

View larger version (35K): [in this window] [in a new window]   Figure 1.   Differences between patients and controls in mean percentage metabolites.

View larger version (34K): [in this window] [in a new window]   Figure 2.   Differences between patients and controls in mean metabolite ratios.

Although the cerebral intracellular pH was similar in patients and controls, there was a significant correlation between intracellular pH and Pa_O2 (R = 0.8, p < 0.002) (Figure 3) in the COPD patient group. There was a weaker correlation between cerebral intracellular pH and Pa_CO2 (R = 0.7, p = 0.02) and between cerebral intracellular pH and arterial pH in this group (R = 0.5, p = 0.05).

View larger version (8K): [in this window] [in a new window]   Figure 3.   Correlations between PO2 and intracellular brain pH in the patient population.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients with mild to moderate hypoxia have previously been found to have abnormalities on testing for psychometric functions, activities of daily living, and electroencephalography (11). We therefore postulated abnormalities in brain bioenergetics in patients with chronic stable hypoxia. The present study was the first in vivo cerebral MRS study to be undertaken in patients with COPD and hypoxia.

³¹P MRS Examination: Technical Considerations in COPD

The MRS examination requires an obligatory period during which each subject lies flat within the bore of the magnet, which presents technical difficulties for COPD patients. Our study protocol was therefore intended to minimize the time in the magnet, but did allow for MR localization to a set of 2-cm slices of brain tissue. Because the spectra from three different slices did not show any significant regional variation, the spectral information was averaged to reduce any errors from background noise. This is standard methodology in chemical shift imaging techniques, in which there is no variation in metabolite signaling between the tissue slices. Obtaining whole-brain spectra would have involved a different technique, and would have been contaminated by a large amount of information from soft tissues, such as muscles of the head and neck. This was obviated by the technique we used. The a priori collection of localized data allowed us to look for regional variations, but since these were inapparent, the spectra were averaged.

³¹P MRS: Biochemical Background

The ³¹P nucleus has proved particularly valuable in clinical MRS, since resonances from PCr, ATP, and Pi are readily observed, and the chemical shift of Pi depends on intracellular pH. These parameters are of central importance in energy metabolism.

A typical ³¹P MR spectrum (Figure 4) contains seven resonances (19), which can be assigned to PMEs, Pi, PCr, PDEs, and nucleoside triphosphates (gATP, aATP, and ATP). Notably absent from this spectrum are signals from membrane phospholipids such as phosphatidylcholine. The phosphorus nuclei in such large molecules have reduced mobility, and the signals are invisible on nuclear MR (NMR). Similarly, the signal from adenosine diphosphate (ADP) bound to proteins is not detectable using NMR methods, and the small ADP signal observed therefore represents only free ADP.

View larger version (18K): [in this window] [in a new window]   Figure 4.   Cerebral 31P MR spectrum (TR-5,000 ms) from (a) patient with stable respiratory failure and (b) a healthy volunteer.

The PME peak contains at least 10 components (20), including glucose-6-phosphate, glycerol-1-phosphate, and ribose-5-phosphate, which are intermediates in carbohydrate metabolism; coenzyme A, which is important in the metabolism of fatty acyl moieties; phosphoethanolamine and phosphocholine, which are metabolites in the pathway of cell membrane synthesis; adenosine monophosphate (AMP), which is an intermediate of ATP and ADP turnover; and 2,3-diphosphoglycerate, which is present in the blood, although it is unclear how much of this in circulating blood contributes to the in vivo NMR spectrum.

The Pi signal represents only about 40% of the intracellular levels of Pi (21), because the Pi bound to mitochondrial inner membrane is invisible on NMR. Pi, together with ATP and ADP, plays a key role in energy metabolism, and changes in cerebral energy state may therefore be reflected by a change in the ATP/Pi ratio.

The PDE peak contains contributions from at least three water-soluble metabolites, including glycerophosphoethanolamine and glycerophosphocholine. These are intermediates in phospholipid catabolism. In addition, phosphoenolpyruvate, although not a PDE, resonates in this region.

The dominant contribution to the nucleotide triphosphate peaks is from ATP, but 10 to 20% of the peak area may be due to uridine triphosphate and guanosine triphosphate. Further signals can be identified overlapping with peaks from the three phosphate groups. For example, a-nucleoside triphosphate resonance overlaps with signals from aADP and nicotinamide adenosine dinucleotide or its reduced form. The g-nucleotide triphosphate signal overlaps with signals from ATP. Nicotinamide adenine dinucleotide in conjunction with its reduced form (NAD/NADH) plays a pivotal role in electron transport in redox reactions, but it remains unclear what proportion is bound to proteins and is therefore undetectable by MRS in vivo.

³¹P MRS: Quantification of Results

The results of MRS are reported here in accord with standard practice in the MR literature, by expressing metabolite changes either relative to ATP or Pi, but also expressing results in terms of the total phosphorus signal when these two denominators also change, because this signal remains unchanged (22). It was not possible to measure absolute concentrations of metabolites in mmol/g wet weight of tissue in the present study because absolute quantitation of metabolites with reference to a known concentration of an external reference standard requires further sequences. This would have unavoidably prolonged the examination time beyond that which was ethically permissible and practically possible for the patient population studied.

Cerebral ³¹P MRS Results in Hypoxia

Cerebral ³¹P MRS has been used to measure ATP biosynthesis in relation to hypoxia and ischaemia in animal studies, neonates, and patients with cerebrovascular accidents (23). In hypoxic-ischemic encephalopathy there is an accumulation of cerebral lactate and a reduction in N-acetyl aspartate in proton MRS spectra, whereas in ³¹P MRS spectra there is a reduction in the PCr-to-Pi ratio and reduced ATP levels. This is a different situation from hypoxia without ischaemia, such as that in patients with COPD.

³¹P MRS of the Skeletal Muscle in COPD

Several other investigators have examined the bioenergetics of skeletal muscle in hypoxia. Most (32) but not all (40) found augmented glycolysis and decreased aerobic metabolism in skeletal muscles of patients with COPD.

Cerebral ³¹P MRS in COPD

Our results suggest several bioenergetic abnormalities in COPD patients with chronic stable hypoxia. In these patients, as compared with normal volunteers, we found a significant increase in PMEs and Pi, with a significant reduction in PDEs expressed as a percentage of the total MR-detectable phosphorus signal. However, no change was noted in the percentage of PCr or in the percent gATP, percent aATP, and percent ATP resonances. The Pi-to-ATP and PME-to-ATP ratios were therefore increased, and the PCr-to-Pi ratio was correspondingly reduced. A possible explanation for these findings would be impaired mitochondrial ATP production from oxidative phosphorylation, owing to hypoxia and leading to accelerated cytoplasmic glycolysis in order to maintain cellular high-energy phosphate levels.

Changes in PME and Pi Resonances

Under hypoxic conditions, mitochondrial oxidative phosphorylation is reduced. Glycolysis in the cytoplasm becomes steadily more important in maintaining cellular ATP levels through a positive feedback loop with increased activity of glycolytic enzymes, particularly phosphofructokinase (41). The increase in percentage of the PME signal that we noted in patients with COPD may well have represented accelerated glycolysis, because the multicomponent resonance of PMEs is chiefly composed of signals from cell membrane precursors and glycolytic intermediates (20).

Because intramitochondrial Pi is membrane bound and thus MR-invisible, any increase in the percentage Pi signal represents an increase in the cytoplasmic Pi concentration. Impaired mitochondrial oxidative phosphorylation may also shift Pi from mitochondria to the cytoplasm. It is therefore noteworthy that effectors of phosphofructokinase activity, such as fructose-2,6-diphosphate, require increased cytosolic Pi levels, such as we noted, for accelerating glycolysis (42, 43). Increased levels of fructose-2,6-diphosphate, itself a phosphomonoester, acts as an effector of fluxes between glycolytic and gluconeogenetic pathways (43). This may further contribute to increases in the percentage PME signal and induction of 6-phospho-fructo-1-kinase. The increase in PME signal therefore occurs because of an increased cytoplasmic concentration of glycolytic intermediates, increased concentration of fructose-2,6-diphosphate, and increase in cell membrane precursors.

However, the net effect of these changes would be to keep cerebral ATP levels unchanged. An alternative mechanism to maintain cellular homeostasis would be a reduction in ATP consumption (44). Because the percentage ATP and percentage PCr levels were unaltered in our study, we conclude that the system must be fully compensated overall.

Changes in PDE Resonance

The reduction in PDE signal found in our study is more difficult to explain. The multicomponent resonance produced by PDE is largely composed of cell membrane degradation products and also signal from mobile phospholipids in the endoplasmic reticulum (45, 46). In diseases in which there is increased cell repair, there is a reduction in PDE resonance. This is because the normal equilibrium that exists between cell membrane synthetic products (like phosphocholine and phosphoethanolamine) and breakdown products (like glycerophosphocholine and glycerophosphoethanolamine) is altered in favor of cell membrane synthesis. Under these circumstances the cell membrane degradation products would be recirculated into synthetic pathways, and would therefore no longer contribute to the PDE signal. This would therefore result in increasing PME resonance and decreasing PDE resonance (47).

Clinical Correlations

Although our study did not establish causality, it is likely that the changes we observed in cerebral bioenergetics in COPD patients were a consequence of chronic changes in arterial blood gases. It is also unclear whether these changes in cerebral bioenergetics are a consequence of chronic hypoxia, hypercapnia, or both. The precise intracellular pH, even within the normal range, may well be determined by many factors. In normal human subjects, intracellular brain pH may be expected to become acidotic with both arterial hypercapnia (48) and with changes in cerebral cellular oxygenation (31). Our findings indicate that the cerebral pHi is tightly regulated in vivo and that it was fully compensated in COPD patients despite chronic ventilatory abnormalities. Multivariate analysis suggested that Pa_O2 (and not Pa_CO2) correlates better with pHi, accounting for 65% of the variability in pHi. This raises the possibility that although pHi is buffered and within the normal range, its tendency to decrease is related to hypoxia.

This association in COPD patients between intracellular pH and arterial PO₂ suggests that at least as far as these patients are concerned, the metabolites of anaerobic metabolism do have an impact on intracellular pH. In a hypoxic intracellular environment, the cell relies chiefly on glycolysis for ATP generation as both Kreb's cycle and oxidative phosphorylation become inefficient. With acceleration of glycolysis, intracellular lactate is expected to accumulate. This is expected to result in a decrease in intracellular pH as the cell becomes overwhelmed with excess lactate. The linear relationship observed between Pa_O2 and pHi (although still within normal range) in our study seems to support this notion. This finding supports our speculation that the switch to anaerobic metabolism in hypoxic patients tends to result in lactate accumulation, which is in turn buffered by cationic proteins. Patients with chronic stable hypoxia seem to fully compensate for developing intracellular brain acidosis either through active ion transport of protons outside the cell (49, 50) or by retention of bicarbonate (51). It is interesting to speculate that as hypoxia increases, the ability of the neuron to buffer lactate, pump out protons, and retain bicarbonate might fail, resulting in intracellular acidosis. This may be the mechanism underlying the decompensation in cerebral function with agitation, confusion, and ultimate unconsciousness that is characteristic of severe respiratory failure.

The present study offers a first insight into the complex biochemical changes in the brains of patients with severe COPD and chronic stable respiratory failure. Further work needs to be done (with ³¹P MRS and ¹H MRS) to correlate the cerebral bioenergetic abnormalities in these patients with other evaluations of cerebral function, and to directly measure levels of intracellular brain lactate. It is now possible to combine MRS with near-infrared spectrophotometry in the same examination in order to assess changes in cerebral blood flow (52, 53). Further studies should combine both modalities, since it is not known whether the spectroscopic abnormalities observed in hypoxic patients are correctable with supplemental oxygen or, indeed, are inducible in normal human subjects with controlled experimental hypoxia and hypercapnia. Correction of such cerebral bioenergetic abnormalities with administered pharmacological agents, independent of any changes in ventilatory status and both in patients with stable COPD and patients with acute exacerbation of this disease, may pave the way for novel modalities of therapy to support cerebral compensation in these patients.