Neurosurgical application of a flow-directed oximetry thermodilution catheter for evaluation of cerebral blood flow--technical note

The neurosurgical application was evaluated of a flow-directed oximetry thermodilution catheter for measurement of oxygen saturation in the jugular vein, which reflects cerebral blood flow (CBF). The catheter allows estimation of changes in CBF during carotid endarterectomy and therapeutically induced hypertension in the management of delayed vasospasm after subarachnoid hemorrhage.

Cerebral blood flow measurements using stable xenon CT with very short inhalation times

A noninvasive, simplified method using inhalation of stable xenon (Xe(s)) and computed tomographic (CT) scanning to estimate regional cerebral blood flow (rCBF) and regional partition coefficient (r lambda) is described. Twenty-four patients with cerebrovascular occlusive disease and six volunteer controls inhaled 30% Xe(s) and 70% oxygen for 180 seconds and exhaled for 144 seconds during serial CT scanning without denitrogenation. The end-tidal Xe(s) concentration was continuously monitored with a thermoconductivity analyzer to determine the build-up range (A value) and build-up rate constant (K value) for arteries with the curve fitting method. The time-CT number (Hounsfield unit) curve for cerebral tissue during the Xe(s) washin and washout phases was used to calculate r lambda and rCBF using least squares curve fitting analysis. The resultant r lambda and rCBF map demonstrated a reliable distribution between the gray and white matter, and infarcted areas. rCBF was high in gray matter, low in white matter, and much lower in infarcted areas than in white matter, r lambda was high in white matter, low in gray matter, and much lower in infarcted areas. Xe(s) CT-CBF studies with very short inhalation of 180 seconds is a clinically useful method for evaluation of rCBF in patients with cerebrovascular diseases.

Hemodynamic evaluation in patients with superficial temporal artery-middle cerebral artery anastomosis--stable xenon CT-CBF study and acetazolamide

Sixteen patients with minor completed stroke in the chronic stage underwent superficial temporal artery-middle cerebral artery (STA-MCA) anastomosis. The acetazolamide-activated regional cerebral blood flow (rCBF) was measured 20 minutes after the injection using inhalation of stable xenon and computed tomographic scanning (Xes CT-CBF study) pre- and postoperatively. Eleven patients (Group 1) showed immediate improvement in neurological state within a few days of the operation, while five (Group 2) showed no improvements. Preoperative rCBF in the ischemic areas without infarction was 30.8 +/- 3.0 ml/100 gm/min in Group 1 and 53.0 +/- 5.2 ml/100 gm/min in Group 2. Preoperative vasodilatory capacity with acetazolamide in Group 1 was 5.7 +/- 8.6 and significantly increased to 19.8 +/- 4.9 after surgery. In Group 2, pre- and postoperative vasodilatory capacity was 12.7 +/- 3.1 and 14.9 +/- 2.9, respectively, and there was no significant change. These results suggested that minor stroke patients with moderate decrease of affected side rCBF (less than 40 ml/100 gm/min) and with hemodynamic impairment may have the surgical indication for STA-MCA anastomosis.

Mechanism of the re-buildup phenomenon in moyamoya disease--analysis of local cerebral hemodynamics with intra-arterial digital subtraction angiography

The authors investigated the mechanism of the re-buildup phenomenon on electroencephalogram in 14 patients of moyamoya disease with superficial temporal artery-middle cerebral artery anastomosis. Visualization of the lateral view of the common carotid angiography was performed with intra-arterial digital subtraction angiography (IA-DSA), using a 4/sec x 3 sec + 2/sec x 5 sec + 1/sec x 5 sec film sequence. The catheter tip was inserted into C5/6 level and 250 mgI/ml of iopamidol was used as the contrast agent; 6 ml in total was injected over 1.5 seconds. Circulation times of the common carotid artery (C3 portion)-ascending parietal vein (delta TTPs) and common carotid artery-internal cerebral vein (delta TTPD) were measured before hyperventilation (HV), immediately after HV, and 3 minutes after HV during pre- and postoperative periods. delta TTPD in the preoperative period was prolonged by HV and was normalized at 3 minutes after HV but delta TTPs were prolonged immediately after and 3 minutes after HV. In the postoperative period, however, these values did not change significantly immediately after and 3 minutes after HV. These findings indicate that delayed cerebral blood flow response to HV is a pathogenetic factor of the re-buildup phenomenon in moyamoya disease.

Addition computed tomography with stable xenon--special reference to ischemic cerebrovascular diseases

Stable xenon (Xes) is used as a contrast agent because it freely diffuses to cerebral tissues through the blood-brain barrier. In this study, 2 axial levels for Xes enhancement analysis were selected from a baseline series of computed tomographic (CT) scans and 6 serial CT scans were obtained every 20 seconds for each scan level during the 240 seconds inhalation period of 30% Xes in 10 volunteer controls and in 52 patients with ischemic cerebrovascular diseases (ICVD). The serial CT scans were added and averaged in each pixel. This was used to make a new CT picture (addition CT scan). The CT scans before the Xes inhalation, the scan at the end of the Xes inhalation, and the addition CT scan were compared to see whether gray matter and ischemic areas could be differentiated from white matter. The addition CT scans could differentiate the three structures very well in both the acute and chronic stages of ICVD. This technique is thought to be a very simple and useful method to detect the small infarcted areas and low perfusion areas that cannot be visualized on precontrast CT scans.

Hemodynamic evaluation before and after the STA-MCA anastomosis--with special reference to measurement of regional transit time with intra-arterial digital subtraction angiography

Twenty-seven patients with minor completed and major stroke in the chronic stage underwent superficial temporal artery-middle cerebral artery (STA-MCA) anastomosis. The regional cerebral blood flow (rCBF), using inhalation of stable xenon and computed tomographic scanning (Xes CT-CBF study), and the mode of transit time (MTT) in the MCA territory using intra-arterial digital aortography were measured. Activated rCBF and MTT was measured 20 minutes after the administration of acetazolamide (10 mg/kg) in 14 patients. Nineteen of the 23 patients with minor stroke (Group 1) showed immediate improvement in their neurological state within a few days of the operation, while four patients with minor stroke (Group 2) and four patients with major stroke (Group 3) showed no improvement. Based on the rCBF obtained with the Xes CT-CBF study, affected side rCBF/unaffected side rCBF and %f [(peak DSA number/affected side MTT)/(peak DSA number/unaffected side MTT)] were compared. There was a significant positive correlation. Affected side MTT in Group 1 was 6.41 +/- 1.16 sec, preoperatively, and significantly decreased to 5.13 +/- 0.91 sec after the operation. On the other hand, preoperative MTT in Group 2 was 4.40 +/- 0.81 sec and 4.76 +/- 0.89 sec, postoperatively. Preoperative %f in Group 1 was 0.514 +/- 0.143 and significantly increased to 0.739 +/- 0.154, postoperatively. Group 2 showed no change. Vasodilatory capacity with acetazolamide showed a marked improvement in Group 1, postoperatively. Our study indicated that if MTT is moderately lengthened, %f is moderately decreased, and vasodilatory capacity is impaired, in patients with minor ischemic stroke will benefit from STA-MCA anastomosis.

Vertebrobasilar insufficiency: stable xenon computed tomography-cerebral blood flow study in posterior circulation revascularization

Preoperative and postoperative local cerebral blood flow were measured by the stable xenon computed tomography-cerebral blood flow technique in 15 patients with vertebrobasilar insufficiency. The surgery included end-arterectomy or angioplasty of the vertebral artery (five cases), superficial temporal artery-superior cerebellar artery anastomosis (eight cases), and superficial temporal artery-posterior cerebral artery anastomosis (two cases). Fourteen (93.3%) of the 15 patients improved in the post-operative period. Low local cerebral blood flow in the ischemic area without infarction manifested a constant and significant increase postoperatively. In summary, the stable xenon computed tomography-cerebral blood flow technique is thought to be a simple and useful method for assessing local cerebral blood flow in posterior circulation perioperatively.

Clinical application of stable xenon CT-CBF studies without denitrogenation

Noninvasive and simplified methods for estimating regional cerebral blood flow (CBF) and regional partition coefficient (lambda) using the inhalation of stable xenon (Xes) and computed tomographic (CT) scanning are described. Thirty percent Xes in 70% oxygen was inhaled for 240 seconds and exhaled for 160 seconds during serial CT scanning without denitrogenation in 26 patients with cerebrovascular diseases and four volunteer controls. During the investigation, the end-tidal Xes concentration was continuously monitored with a thermoconductivity analyzer to determine the build-up range (A value) and build-up rate constant (K value) of the artery by the curve fitting method. Calculated A and K values were corrected by the following formulae reported previously: for patients aged 0-20 years, Ae = 0.75Aa + 2.15, Ke = 0.67Ka + 0.69; 21-40 years, Ae = 0.56Aa + 3.24, Ke = 0.38Ka + 1.12; 41-60 years, Ae = 0.91Aa + 1.95, Ke = 0.38Ka + 1.32; over 61 years, Ae = 0.52Aa + 3.81, Ke = 0.31Ka + 1.55 (Ae and Ke were calculated with end-tidal Xes concentration, Aa and Ka were calculated by direct sampling of arterial blood). A CBF map (f map) and lambda map made with corrected A and K values demonstrated reliable distribution. The CBF was high in the gray matter, low in the white matter, and much lower in the infarcted area. lambda was high in the white matter, low in the gray matter, and much lower in the infarcted area. Eight patients were examined with and without denitrogenation. Both the f map and lambda map with denitrogenation were compatible with those without denitrogenation.(ABSTRACT TRUNCATED AT 250 WORDS)

Measurement of energy expenditure in acute stage of cerebrovascular diseases

Oxygen consumption (VO2), carbon dioxide production, respiratory quotient, urinary nitrogen excretion (UN), resting energy expenditure (REE), REE rate (%REE) and consumption rates of carbohydrate (%C), fat (%F) and protein (%P) were pre- and postoperatively measured in the acute stage of 17 patients with cerebrovascular diseases. They included six of ruptured intracranial aneurysms, six of cerebrovascular occlusion, and five of hypertensive intracerebral hemorrhage. Preoperative VO2, UN, REE, and %REE were 185.1 ml/min, 7.1 gm/day, 1280 cal/day, and 132.6, respectively. %C, %F, and %P were 40.9, 38.5, and 20.6, respectively. Especially the preoperative %P had markedly increased. Postoperatively, %C significantly decreased to 28.3, and %P showed a significant increase up to 30.3. Patients with ruptured intracranial aneurysms manifested a significant increase of %P and %F and a decrease of %C, postoperatively. In summary, the catabolism of fat and protein increases in the acute stage, especially postoperatively, and this hypermetabolic state should be considered when caring for patients with cerebrovascular diseases as malnutrition might prevent wound healing and cause weight loss and a decrease in immunity.

Hypermetabolism in the acute stage of hemorrhagic cerebrovascular disease

Oxygen consumption (VO2), carbon dioxide production (VCO2), urinary nitrogen excretion, respiratory quotient, resting energy expenditure (REE), %REE, and the consumption rates of carbohydrate, fat, and protein (%CHO, %Fat, %Prot, respectively) were determined pre- and postoperatively by indirect calorimetry in 13 patients with ruptured intracranial aneurysms and 11 patients with hypertensive intracerebral hemorrhage in the acute stage. The preoperative VCO2, VO2, urinary nitrogen excretion, respiratory quotient, REE, and %REE were, respectively (mean +/- standard deviation): 171 +/- 46 ml/min, 203 +/- 56 ml/min, 10.3 +/- 1.7 gm/day, 0.84 +/- 0.01, 1397 +/- 389 Cal/day, and 129% +/- 8%. The values for VCO2, VO2, REE, and %REE were all increased above normal levels. The %Prot was increased to 26.1% +/- 9.1%. In the postoperative period, the VCO2, VO2, urinary nitrogen excretion, REE, and %REE significantly increased to: 186 +/- 44 ml/min, 229 +/- 56 ml/min, 14.8 +/- 2.9 gm/day, 1557 +/- 384 Cal/day, and 141% +/- 21%, respectively. The %Fat and %Prot also increased significantly, but the %CHO significantly decreased. Preoperatively, in the patients with ruptured intracranial aneurysms, there was a greater increase in %Prot in eight patients classified (according to Fischer) as having a Group 3 or 4 subarachnoid hemorrhage (SAH) on computerized tomography than in five patients classified as having a Group 1 or 2 SAH. In summary, increased metabolic expenditure, especially increased catabolism of protein and fat, is characteristic of accompanying hemorrhagic cerebrovascular disease, and there is an increase in consumption of fat and protein in the postoperative period. Lack of precise knowledge about the cause and consequences of these metabolic responses makes it impossible at present to judge the optimal extent of nutritional replacement. The hypermetabolic state should be taken into consideration when caring for these patients as it may cause weight loss, poor wound healing, and susceptibility to infection.

[Transbrachial artery approach for selective cerebral angiography]

Transaxillary or transbrachial approaches to the cerebral vessels have been reported, but selective angiography of all four vessels has not been possible through one route. In this report, a new technique for selective cerebral angiography with transbrachial approach is described. One hundred and twenty three patients with cerebral infarction, vertebrobasilar insufficiency, intracerebral hemorrhage, epilepsy, or cerebral tumor were examined. Those patients consisted of 85 outpatients and 38 inpatients whose age ranged from 15 years old to 82 years old. The patients were examined via the transbrachial approach (97 cases via the right brachial, 29 cases via the left). Materials included a DSA system (Digital Fluorikon 5000, General Electric Co.), a 4 French tight J-curved Simmons 80-cm catheter, a 19-gauge extra-thin-wall Seldinger needle, and a J/Straight floppy 125-cm guidewire. Generally, the volume of the contrast agent (300 mgI/ml iopamidol) used in the common carotid artery angiogram was 6 ml, while that used in the vertebral artery angiogram was 4ml. If catheterization of the vertebral artery or right common carotid artery was unsuccessful, about 8ml of the contrast agent was injected into the subclavian or brachiocephalic artery. Definitive diagnosis and a decision on proper treatment of the patients can be easily obtained, and the results were clinically satisfactory. Moreover, no complications were encountered in this study. This new technique making a transbrachial approach to the cerebral vessels using the DSA system is introduced here. Neurosurgeons can use this technique easily, and they will find that it provides them with all the information they need about the patient.

Neurogenic pulmonary edema in the acute stage of hemorrhagic cerebrovascular disease

Extravascular lung water (EVLW) was measured by the double-indicator dilution method in 25 patients with hemorrhagic cerebrovascular diseases. EVLW had a significantly positive correlation with both alveolar-arterial oxygen difference (AaDO2) and intrapulmonary shunt. The value of EVLW in the acute stage in 15 patients with increased AaDO2 more than 20 mm Hg was 7.8 +/- 2.2 ml/kg and that in the chronic stage 4 weeks after onset significantly decreased to 4.6 +/- 0.7 ml/kg (P less than 0.001). The value of EVLW in the acute stage in 10 patients with normal AaDO2 less than 20 mm Hg was 4.7 +/- 1.1 ml/kg and that in the chronic stage 4 weeks after onset was 4.5 +/- 0.2 ml/kg. There was no significant difference between them. Pulmonary arterial blood pressure, pulmonary capillary wedge pressure, central venous pressure, cardiac index, systemic vascular resistance index, and pulmonary vascular resistance index in the acute stage in the 25 patients were all within the normal range. Three patients with neurogenic pulmonary edema had markedly increased EVLW without abnormalities in pulmonary arterial blood pressure, pulmonary capillary wedge pressure, central venous pressure, cardiac index, systemic vascular resistance index, and pulmonary vascular resistance index. From these facts, the main cause of the increase in EVLW cannot be explained by left ventricular failure, but can be explained by high permeability pulmonary edema.

[Hypoxemia in the acute stage of hypertensive intracerebral hemorrhage, with special reference to increased extravascular lung water]

Respiratory responses to acute brain damage have been well documented. Severe hypoxemia of unknown origin has often been seen in the acute stage of intracerebral and subarachnoid hemorrhage. Neurogenic pulmonary edema is a particular concern in the neurosurgical intensive care setting. In this study, the extravascular lung water volume (EVLW), alveolar-arterial oxygen difference (AaDO2), and intrapulmonary shunt (QS/QT) were measured in 17 patients with hypertensive intracerebral hemorrhage in the acute stage. The mean AaDO2, QS/QT, and EVLW values were 19.1 mmHg, 7.7%, and 5.49 ml/kg, respectively. The mean pulmonary arterial, central venous, and pulmonary capillary wedge pressures were all within the normal ranges. Eight patients with an AaDO2 of 20 mmHg or more manifested abnormal increases in QS/QT and EVLW, the mean values of which were 10.3% and 6.52 ml/kg, respectively. Nine patients with an AaDO2 of less than 20 mmHg exhibited normal QS/QT and EVLW, with mean values of 5.4% and 4.57 ml/kg, respectively. There was a significant, direct correlation between AaDO2 and EVLW. AaDO2 and EVLW in 11 measured cases either stayed normal or normalized within 4 weeks. These results suggest that acute congestive heart failure and/or pulmonary hypertension were not the main causes of the increase in AaDO2 and EVLW; the increase in EVLW might be explained by increased permeability of pulmonary vessels. The results of this study suggest that an increase in EVLW is causally related to hypoxemia. Further studies of lung water are necessary if we are to better understand respiratory abnormalities associated with acute-stage hypertensive intracerebral hemorrhage.

[Stable xenon CT-CBF study with measurement of end-tidal xenon concentration--correction by arterial concentration of xenon]

Non-invasive methods with monitoring end-tidal stable xenon (ETXes) are described for estimating local cerebral blood flow (LCBF) and local partition coefficient (L lambda). 30% of Xes in oxygen was inhaled for 240 sec and exhaled for 160 sec during serial CT scannings after denitrogenation with pure oxygen breathing. During the examination, serial samplings of arterial blood and continuous monitoring of ETXes were performed to determine build up range (A) and build up rate constant (K) of artery. Calculated A and K using the arterial sampling (Aa and Ka, respectively) were compared with the calculated A and K using the continuous monitoring of ETXes (Ae and Ke, respectively) in 109 patients with epilepsy, head trauma, or cerebrovascular diseases. Ae and Ke had significantly positive correlation with Aa and Ka, respectively.(ABSTRACT TRUNCATED AT 250 WORDS)

[Diaphragmatic pacing by electrical stimulation of the phrenic nerve--a case report]

A four-year-old boy had traffic accident and was transferred to our hospital with cardiorespiratory arrest, on March 24, 1986. After resuscitation, quadriplegia and apnea were sustained. Computed tomography (CT) scan on the day of admission only demonstrated subarachnoid hemorrhage and plain roentgenograms of the skull and cervical spine showed no abnormalities. He opened his eyes to verbal order one week later, but apnea and quadriplegia sustained. CT scan obtained on that day demonstrated no abnormalities but roentgenogram of the cervical spine showed atlanto-axial dislocation. On May 26, 1986, he was treated with densectomy through transoral approach. Furthermore, on June 19, he underwent laminectomy of the atlas, partial craniectomy of the occipital bone, and posterior fusion with iliac bone graft. However, apnea continued, and bilateral diaphragmatic pacemaker (Avery Laboratory Inc.) was planed: On Sept. 18, platinum and silicone rubber electrodes were surgically implanted on the both phrenic nerves at the cervical portion. Each electrode was connected by lead wire and connector to a radio-frequency receiver which was implanted in a subcutaneous abdominal pocket. An external transmitter supplied electrical power and information of the stimulation to the implanted receiver by radiofrequency across the skin. Pacing time was gradually prolonged and continuous pacing for four hours could be successfully gained by March, 1987. Pneumonia once occurred in the course of gradual prolongation of the pacing time. The ideal candidates for diaphragmatic pacing are patients with normal phrenic nerves, diaphragm, and lungs.(ABSTRACT TRUNCATED AT 250 WORDS)

Regional cortical blood flow during extra-intracranial bypass surgery in young patients with moyamoya disease

Regional cortical blood flow (rCBF) was measured with a thermal diffusion flow probe in 14 cortical regions of 12 young patients with moyamoya disease before and after completion of superficial temporal-middle cerebral artery anastomosis. The prebypass rCBF value was low in most patients, especially in the frontal regions. On temporary occlusion of the cortical artery during the surgical procedure, no significant drop in rCBF occurred. rCBF increased immediately after anastomosis in all but one region and continued to increase for 5-10 minutes in four of the eight regions measured. In the other four regions, rCBF declined gradually after the initial increase, and the final increase was slight. The average increase in rCBF was significant at 1-2 minutes and at 5-10 minutes after anastomosis. However, the latter increase did not bring the flow into the normal range. It may be that the initial increase in postbypass rCBF is determined solely by the pressure gradient between the donor and recipient arteries, and that subsequent rCBF is controlled by other factors within the brain. In one patient who underwent double anastomoses to the frontal and temporal lobes, neither anastomosis increased rCBF in the non-corresponding lobe to recipient artery. This suggests that there is no direct connection between supra- and infra-Sylvian arteries and supports the concept of nonuniform epicerebral microcirculation in moyamoya disease.

[Problems in the embolization therapy of spinal arteriovenous malformation]

Six cases with spinal arteriovenous malformation refractory to repetitive embolization therapy were reported. In these, de novo feeding arteries appeared after the emboli obliterated the proximal portion of the blood feeding system. Though the malformation disappeared after the embolization therapy in the angiograms of the original feeding arteries, the lesion was visualized in the follow-up angiogram from different segmental arteries via various collaterals. Through angiographic evaluation and magnetic resonance imaging were necessary to point out the "false angiographic disappearance" of the spinal arteriovenous malformation after the embolization therapy.

[Measurement of local cerebral blood flow by computerized tomography with inhalation of stable xenon and curve-fitting method]

Non-invasive methods are described for estimating local cerebral blood flows (LCBF) and local partition coefficients (L lambda) during inhalation of 30% stable xenon gas (Xe) in oxygen during CT scanning. After the denitrogenation with pure oxygen breathing, 30% Xe is inhaled for four minutes to minimize subanesthetic effects with a rubber face-mask and the delivery system of Xe. Local time-delta Hounsfield units curve during the Xe wash-in and wash-out phase is utilized in order to calculate L lambda and LCBF using a least squares curve fitting analysis. Calculated L lambda and LCBF with the new method manifested reasonable distribution between the grey and white matters, and reproducibility was excellent in our study. Several case studies of patients with cerebral infarction are presented to demonstrate the characterization of L lambda and LCBF patterns in various tissues and theoretical grounds underlying the new method of curve fitting analysis are discussed.

Mapping of local cerebral blood flow with stable xenon-enhanced CT and the curve-fitting method of analysis

A noninvasive method is described for estimating local cerebral blood flow (LCBF) and local partition coefficients by means of computed tomographic scanning during inhalation of 30% stable xenon gas in oxygen. Time-dependent xenon concentrations in arterial blood and brain tissue during the wash-in and washout phases are used to calculate partition coefficients and LCBF values by means of a least-squares curve-fitting analysis. Control values for partition coefficient and LCBF obtained from control subjects with minor head trauma in the chronic stage were compatible with those in several past reports, and reproducibility was satisfactory. The theoretic grounds underlying this new method of curve-fitting analysis are discussed.

Pitfalls in the surgical treatment of moyamoya disease. Operative techniques for refractory cases

Eleven cases of moyamoya disease refractory to indirect non-anastomotic revascularization, including encephalomyosynangiosis in two, encephaloduroarteriosynangiosis in seven, and encephalomyoarteriosynangiosis in two, are described. The patients suffered from recurrent cerebral ischemic symptoms, and further operative intervention, including superficial temporal artery-middle cerebral artery anastomosis and intracranial omental transplantation, was performed. The choice of operative maneuver depended on the availability of scalp arteries and on the nature of the ischemic symptoms. Although indirect non-anastomotic revascularization procedures have the advantage of technical ease and most patients respond to these procedures alone, there are some patients like the 11 presented here who are not cured by such procedures. In such cases, direct anastomotic revascularization is necessary for the prevention of stroke.

Intraarterial digital subtraction angiography for definitive diagnosis of intracranial aneurysms

Comparisons between digital subtraction angiography and conventional angiography have frequently been made in the radiologic literature, and the high quality and several advantages of the former have been reported. In this study, 101 patients with intracranial aneurysms were examined only by intraarterial digital subtraction angiography; no conventional angiography was used. High-quality images were consistently obtained, facilitating an accurate and definitive diagnosis of intracranial aneurysm. Magnification radiography and stereography using intraarterial digital subtraction angiography were done to obtain a more precise diagnosis. Five small intracranial aneurysms with diameters of 1.0 to 2.0 mm could be detected. The procedure was considered to be as reliable and as safe as conventional angiography, used previously. Important advantages of intraarterial digital subtraction angiography include reduced procedural time and decreased contrast agent burden, factors that will ensure broader application of this procedure for definitive diagnosis of intracranial aneurysms as experience with the technique accumulates.