Temperature measurements in high thermal gradients: I. The effects of conduction.

Two temperature probes (a fluoroptic sensor and a metallic thermistor), which are both suitable for stereotaxic implantation, were used in comparative thermometry studies during interstitial microwave heating of the brain in vivo. Thermal distributions having large temperature gradients (5-10 degrees C/cm) were routinely observed. The temperature differentials (delta T) between the 2 probes were position dependent within the thermal field. The maximum difference in temperatures measured, using the 2 probes along identical tracks without a catheter, ranged between 0.5 degree C and 1.8 degree C. Near the brain/air surface, the thermistor measured lower temperatures than the optical probe; however, medial to the antenna, the thermistor temperatures were higher than the optic sensor. The measured temperature discrepancies are the result of smearing due to thermal conduction along the axial length of the metallic thermistor probe. These effects are significantly accentuated when the temperature probes are tracked in catheters. Experiments performed in a nonperfused phantom, heated with the interstitial microwave antenna, demonstrated similar conductive effects. Studies in a nonelectromagnetic environment (flow cell-thermal step gradient) additionally confirmed that thermal conductive artifacts were the major source of temperature error.

Temperature measurements in high thermal gradients: II. Analysis of conduction effects.

Measurement errors associated with thermal conduction along a temperature probe in regions of high thermal gradients are examined. An analysis of a conducting probe inserted into an insulating catheter for the purpose of temperature mapping gives a means for estimating the effects of thermal smearing on the measured distribution. A comparison is made between the theory and an experimental test case (flow cell-thermal step gradient). Also, an iterative algorithm is developed to correct thermally smeared temperature distributions in order to reconstruct the desired unsmeared distributions. The algorithm is checked for self consistency in the flow cell experiment and is applied to in vivo data obtained during interstitial microwave heating in normal dog brain. Data from flow cell measurements are used to make relative comparisons of the probe conduction artifact for several different temperature probes (2 thermocouple needle probes, a thermistor needle probe and an optical probe) and assorted teflon catheters (16, 18 and 20 ga).

Thermal conduction effects associated with temperature measurements in proximity to radiofrequency electrodes and microwave antennas.

The smearing effects due to thermal conduction along various, nonenergized, interstitial devices were quantified in a flow cell-thermal step gradient. An insulated cylindrical flow cell with a high (ca 45 degrees C, 1.12 cm i.d., 1.6 cm o.d.) temperature region surrounded by a low (ca 37 degrees C) temperature region was used to compare temperature profiles measured with a thermocouple sensor inside a Stanford radiofrequency (RF) hyperthermia/brachytherapy catheter, a BSD instrumented microwave (MW) antenna (i.e., thermistor integrated into a dipole antenna) and a Dartmouth MW antenna with a juxtaposed optical sensor. Two parameters were used to quantify the thermal smearing of each interstitial device in the flow cell: (a) the maximum temperature difference (MTD) and (b) the full- width- half-maximum (FWHM) of the high temperature region. The "true" temperature maximum (45.4 degrees C) and distribution (FWHM = 1.65 +/- 0.06 cm) were measured with an optical sensor. These data indicate that the BSD instrumented MW antenna significantly smeared the true temperature profile (MTD = 2.7 degrees C, FWHM = 2.1 cm), as did the Dartmouth MW antenna (MTD = 1.5 degrees C, FWHM = 1.7 cm). The Stanford RF catheter, when insulated, resulted in minimal smearing (MTD = 0.3 degrees C, FWHM = 1.9 cm). Moreover, when the insulation was removed so the RF electrode was exposed to the thermal step gradient, smearing was again minimal (MTD = 0.3 degrees C, FWHM = 1.9 cm).

Chronic histological effects of ultrasonic hyperthermia on normal feline brain tissue.

The histopathological changes associated with ultrasonic heating of normal cat brain have been correlated with thermal distributions. Ultrasound energy was applied for 50 min at different intensities to generate tissue temperatures from 42 to 48 degrees C. Animals were sacrificed at various intervals from 1 to 56 days. The organization and resolution of thermal damage was characterized by three stages of histopathological changes within the nervous tissue. The acute stage (Days 1-3) was defined by (1) extensive coagulation necrosis, (2) pyknosis of neuronal elements in the gray matter, (3) edema and vacuolation in the white matter, and (4) polymorphonuclear leukocytes. The subacute stage (Days 3-21) was characterized by (1) the appearance of lipid-laden macrophages, (2) liquefaction of the necrotic regions, (3) fibroblastic proliferation, and (4) vascular proliferation with some perivascular inflammatory infiltration (lymphocytes). Lastly, the chronic stage (Days 21-56) was defined by (1) fibrosis (reticulin and collagen formation) and (2) gliosis (reactive astrocytic proliferation) occurring around the fluid-filled necrotic center. Analysis of these data has also included a study of the lesion size versus the dose (temperature for 50 min) of heating. The results demonstrate a significant linear dose-response correlation. The results of this study indicate that the histological appearance and time course of repair of thermal injury in the normal brain tissue are analogous to acute brain necrosis resulting from cerebral infarction, except the thermal damage does not result in significant hemorrhage.

Response of a brachytherapy model using 125I in a murine tumor system.

The effects of low-dose-rate irradiation (brachytherapy) were investigated in vivo using a murine mammary adenocarcinoma (MTG-B) growing in the flank of C3H mice. For local tumor irradiations, a noninvasive cap was devised to cover the tumor and house three 125I seeds (average apparent activity 5.2 mCi each) located at 120 degree intervals around the circumference of the hemispherical cap (13 mm i.d.). Mice were secured during treatment in a tube allowing limited mobility while restricting access to the seeds. Tumors were exposed to a series of dose rates ranging from 14-40 cGy/h, and the total dose over the treatment interval (48 or 72 h) ranged from 830 to 2378 cGy. A total of nine experiments were conducted using the caps over a 10-week interval. In each experiment three groups (irradiated tumors, sham controls, and untreated controls) were analyzed, each containing 8-15 mice (N = 34, untreated control; N = 46, sham control; N = 91, brachytherapy irradiation). The brachytherapy results are compared to the effects of external beam irradiation in the same tumor system. A linear relationship was observed between the total radiation dose and doubling volume growth delay (GDDV) or treatment volume growth delay (GDTV) for the brachytherapy and external beam irradiation. The slopes of the dose-response curves are steeper for the acute dose (517 cGy/min) external beam irradiation (0.0072 day/cGy, GDDV; 0.00695 day/cGy, GDTV) than for the brachytherapy (0.0050 day/cGy, GDDV; 0.0057 day/cGy, GDTV) using both GDTV and GDDV end points. Comparison of the tumor volume regrowth slopes indicates that the tumor bed effect is larger for external beam irradiation than for brachytherapy, suggesting that the tumor bed effect may be dose-rate dependent.

Natural history of experimental intracerebral hemorrhage: sonography, computed tomography and neuropathology.

The evolution of intracerebral hemorrhage was investigated in a canine model by high resolution sonography, computed tomography (CT), and neuropathologic examination. In 12 dogs, a parietal lobe hematoma was introduced by craniotomy. The sonographic appearance of acute hemorrhage was characteristic and consisted of a sharply circumscribed, homogeneous, highly echogenic lesion, the size and shape of which correlated closely to the area of increased density seen on the CT scan. This changed within 3-4 days to an echogenic rim surrounding a hypoechoic center. Histologically, this change corresponded to a loss of integrity of individual red blood cells. This occurred earliest in the hemorrhage center causing a hypoechoic center, while intact red blood cells at the periphery accounted for the echogenic rim. Shortly after the red blood cells lost their biconcave shape they began to lose their hemoglobin causing the hemorrhage to become isodense with surrounding brain on the CT scan. Faint contrast enhancement by CT was noted at this early stage and was related primarily to a mononuclear perivascular infiltrate at the edge of the hemorrhage. A collagen capsule formed around the hemorrhage over a 2 week period. This capsule slowly replaced intact red cells as the cause of the now shrinking echogenic rim. This capsule was also responsible for the increasing ring contrast enhancement around the resolving hemorrhage. The sequence of image changes seen on both CT and sonography in this experimental model closely resembled the findings seen in intracerebral hemorrhage in patients.

Experimental brain abscess: enhanced sonography and pathologic correlation.

A new sonographic contrast agent, gelatin-encapsulated nitrogen microbubbles, was introduced intraarterially to enhance the high-resolution sonographic scan of an experimental brain abscess. The echogenicity produced by the microbubbles correlated closely to the site and distribution of abscess neovascularity. The contrast agent aided in the detection of small necrotic centers in the late stages of abscess evolution when these centers were not visualized on noncontrast sonograms. The echogenic effect of the microbubbles was maximum immediately after injection; it decreased by 5 min and had virtually disappeared at 15 min.

Feasibility of ultrasound hyperthermia in the treatment of malignant brain tumors.

In order to test the biological feasibility of using ultrasound-generated hyperthermia for the treatment of brain tumors, damage threshold studied and thermal dosimetry studies as a function of temperature were performed in 44 acute experiments in cats. Bilateral craniotomies were performed to expose the dural surface. Ultrasonic radiation was applied for 50 minutes at different intensities to generate temperatures up to 48 degrees C. Thermal fields were mapped using an electrode array of three triple-junction thermocouple probes. Each probe left a track easily identified histologically. Serial sections of each brain were cut and stained, allowing for precise correlation of histology and thermocouple location and temperature. At temperatures of less than 42 degrees C for 50 minutes, no evidence of damage could be detected in either gray or white matter. At 43 degrees C partial loss of neurons was seen in the brain adjacent to the probe, but at the same temperature in white matter, only edema was seen. At temperatures of 44-45 degrees C there was definite loss of both neurons in the gray matter and myelin tracts in the white matter. The lesions created by using ultrasound-generated hyperthermia were sharply marginated. This sharp demarcation histologically correlated well with the abrupt fall off in temperature as a function of distance from the lesion edge. The results of this study are important in two respects. First, it demonstrates that ultrasound can effectively heat the brain in an extremely controlled and precise manner. Second, the brain can withstand temperatures to 42 degrees C without showing histological evidence of damage, which is the temperature range at which neoplastic cells begin to show cytotoxic effects.

Thermal enhancement of low dose rate irradiation in a murine tumour system.

The effects of localized hyperthermia (HT) in combination with low dose rate irradiation (brachytherapy) have been investigated in vivo using a murine mammary adenocarcinoma. Flank tumours were grown to 0.45-0.70 cm3 in volume, at which time their treatment course was initiated. Tumours were locally heated in a water bath for 15 min at either 44 or 45 degrees C. For tumour irradiations a non-invasive cap was devised to permanently house three iodine-125 sealed sources located at 120 degree intervals around the circumference of the hemispherical cap. During treatment, mice were secured in a modified syringe tube allowing mobility while restricting access to the cap which was placed over the tumour. Calculated dose rates ranged from 15 to 40 cGy/h. Brachytherapy (BT) was delivered for 48 or 72 h to obtain a dose range of 830-2378 cGy. Mice were randomized into one of 10 treatment protocols: BT alone, HT-BT, BT-HT, HT-BT-HT, 1/2BT-HT-1/2BT, four control groups of HT alone and a sham treatment group. Normalized tumour doubling volume growth delays (GDDv) were used to calculate the thermal enhancement ratios (TER). In the 44 degrees C experiments, HT before BT (TER = 1.33 +/- 0.071) was more efficacious than HT after BT (TER = 1.07 +/- 0.042). Two HT treatments, one given before and one after BT (TER = 1.38 +/- 0.152), were not different from a single HT treatment given before BT. However, a single HT treatment given in the middle of an interrupted course of BT resulted in the greatest thermal enhancement (TER = 1.64 +/- 0.072) compared to any other treatment sequence. These data suggest that potentiation of low dose rate irradiation by a single heat treatment may be maximized if the HT is given either in the middle of, or simultaneously with, the BT.

Heat loss and blood flow during hyperthermia in normal canine brain. I: Empirical study and analysis.

The effects of blood flow and thermal conduction during microwave hyperthermia were investigated in normal canine brain. Heating was accomplished with an external microstrip spiral antenna and temperature measurements were made using a multichannel fluoroptic thermometry system. In order to determine cooling rates, temperature measurements made during cooling were fitted with a model consisting of a constant value and an exponential term. Data from experiments in both perfused and non-perfused brains could be fitted with this simple model. The resulting cooling rates indicated that heat loss by conduction is comparable to that by blood flow. In another series of experiments, temperature measurements were made during several 1 min cooling intervals in which the power was shut off intermittently during a 35 min heating episode. Results were consistent with a 2-3-fold increase in blood flow rate which occurred gradually throughout the course of heating. Parameters that affect the determination of cooling rates are discussed in terms of the bioheat transfer equation. These investigations demonstrate that a simple heat sink model provides a good representation of the cooling data for the thermal distributions obtained.

Heat loss and blood flow during hyperthermia in normal canine brain. II: Mathematical model.

A mathematical model for heating and cooling during hyperthermia has been developed from an appropriate solution of a bioheat transfer equation. Predicted cooling rates obtained from the model have been compared with cooling rates obtained from experiments performed on both perfused and non-perfused normal canine brain tissue. The agreement between the predicted and observed cooling rates in non-perfused tissue is satisfactory (within 6-11 per cent) and provides confidence that the conduction process is being accurately represented. The model is then used to estimate the relative contribution of conductive and convective (blood flow) heat loss during cooling for the in vivo experiments. Estimates of blood flow dynamics are made from cooling data taken early and late in a heating course using the model to correct for conductive heat loss. Simplified forms of the bioheat transfer equation are examined. An adequate model for the observed cooling data is one that treats heat loss (both conduction and blood flow) as a heat sink (i.e. an effective perfusion model) rather than an effective thermal conductivity model.

Immunohistochemical study of glial fibrillary acidic protein in avian sarcoma virus-induced gliomas in dogs.

Anaplastic astrocytomas, gliosarcomas and sarcomas were induced in 33 neonatal dogs by intracerebral injection of purified Schmidt-Ruppin strain of avian sarcoma virus (ASV). A total of 11 anaplastic gliomas, 7 gliosarcomas and 18 sarcomas were induced. Ten of the 11 anaplastic astrocytomas and all 7 gliosarcomas were positive for GFAP using the peroxidase-antiperoxidase method on formalin-fixed, paraffin embedded tissue. None of the sarcomas were positive for GFAP. One anaplastic astrocytoma was analyzed by gel electrophoresis and immunoblot and the amount of GFAP in the tumor was compared with comparable samples from normal white and gray matter. Densitometric analysis of the GFAP stained gels showed that 12-13% of tumor protein was GFAP compared with 2-3% for the non-neoplastic white and gray matter. The results of this study add further evidence to the astrocytic origin of many of the induced anaplastic primary brain tumors by ASV.

Short term, high dose corticosteroids in computed tomographic staging of experimental brain abscess.

A short term, high dose corticosteroid treatment protocol was investigated to determine its clinical utility in staging an experimental brain abscess. Corticosteroids were shown to decrease the degree of contrast enhancement of brain abscess 12 h after administration but the magnitude of the effect was not consistent enough to be clinically useful in staging. The corticosteroid effect progressively diminished as the inflammatory lesion encapsulated over time. This effect emphasized that a decreasing ring diameter and not diminished contrast enhancement should be the CT criterion for brain abscess resolution.

Correlation of neuropathologic findings, computerized tomographic and high-resolution ultrasound scans of canine avian sarcoma virus-induced brain tumors.

The location, size, geometry and neuropathological findings of anaplastic astrocytomas (AA), gliosarcomas and sarcomas induced by the avian sarcoma virus (ASV) in dogs were compared with images generated using computerized tomography (CT) and real time high-resolution ultrasound (HRUS). Seven AA showed a wide range of findings on CT. Pre-contrast CT scans showed that the tumors could be hyper, hypo, or isodense. Three of seven AA had no contrast enhancement; two of these tumors were also isodense which resulted in a false-negative CT exam. Partial enhancement was seen in one tumor. This resulted in a sensitivity of detection of 72%. Real time HRUS was able to define tumor location, size and geometry of the AA missed or incompletely imaged by CT. All tumors were hyperechoic. Inhomogeneity of the echo pattern was due to hemorrhage, cyst formation, and necrosis within the tumors. Such secondary tumor characteristics were more accurately defined by HRUS compared to CT. Vasogenic edema in the brain surrounding tumors was of low density on CT and hypoechoic or indistinguishable from normal brain on US. Similar findings were seen in six gliosarcomas, two of which were not visualized by either pre- or post-contrast enhanced CT scans (sensitivity of 66%). Sarcomas differed in that they were either hyper or isodense; none were hypodense. The area of increased density matched the tumor geometry and correlated with dense cellularity and reticulin deposition. All 13 sarcomas showed contrast enhancement (100% sensitivity), but in two tumors, contrast enhanced CT underestimated the size of the tumor. Because of the large size and multiplicity of the sarcomas, HRUS imaging was not able to resolve the entire tumor volume because of limited imaging access. Intravenously injected horseradish peroxidase (HRP) crossed the tumor blood-brain barrier (BBB) only in those tumors in which contrast enhancement was seen. These studies suggest that intraoperative HRUS imaging may be useful in detecting and delineating human AA incompletely visualized by CT.

Iridium-192 brachytherapy in combination with interstitial microwave-induced hyperthermia for malignant glioma.

A phase I clinical trial assessing the feasibility and safety of hyperthermia in combination with 192Ir brachytherapy (60 Gy) for the treatment of malignant glioma now includes 14 patients. Hyperthermia (42-43 degrees C at tumor margin for 60 min) has been induced using stereotactically implanted afterloading catheters and a microwave (915 or 2,450 MHz) antenna array. Thermometry recorded along each catheter confirms the general ability of the technique to heat such volumes, but thermal heterogeneities are documented. Transient or permanent worsening of previous neurologic deficit, seen in 7 patients, has been the most common morbidity.