Abstract
Background
The range of an ablation zone (AZ) plays a crucial role in the treatment effect of microwave ablation (MWA). The aim of this study was to analyze the factors influencing the AZ range.
Methods
Fourteen factors in four areas were included: patient-related factors (sex, age), disease-related factors (tumor location, liver cirrhosis), serological factors (ALT, AST, total protein, albumin, total bilirubin, direct bilirubin, and platelets), and MWA parameters (ablation time, power, and needle type). Multiple sequence MRI was used to delineate AZ by three radiologists using 3D Slicer. MATLAB was used to calculate the AZ length, width, and area of the largest section. Linear regression analysis was used to analyze influencing factors. Moreover, a subgroup analysis was conducted for patients with viral hepatitis.
Result
220 patients with 290 tumors were included between 2010–2021. In addition to MWA parameters, cirrhosis and tumor location were significant factors that influenced AZ (p < 0.001). The standardized coefficient (beta) of cirrhosis (cirrhosis vs. non-cirrhosis) was positive, which meant cirrhosis would lead to a decrease in AZ range. The beta of tumor location (near the hepatic hilar zone, intermediate zone, and periphery zone) was negative, indicating that AZ range decreased as the tumor location approached the hepatic hilum. For viral hepatitis patients, Fibrosis 4 (FIB4) score was a significant factor influencing AZ (p < 0.001), and the beta was negative, indicating that AZ range decreased as FIB4 increased.
Conclusion
Liver cirrhosis, tumor location, and FIB4 affect the AZ range and should be considered when planning MWA parameters.
Introduction
Liver tumors, including primary hepatocellular carcinoma (HCC) and metastatic lesions, are serious diseases that adversely affect human health [Citation1,Citation2]. HCC is the second most common cause of cancer-related deaths [Citation3], and the liver is a common metastatic site for colon cancer, breast cancer, and other malignant tumors [Citation4,Citation5]. For patients with liver tumors, complete local curative treatment is a crucial therapeutic step and can help reduce the risk of tumor recurrence, prolong their lives, and even cure their disease [Citation3,Citation6]. Currently, surgical resection and ablation are the mainstream methods for treating liver tumors, as recommended by guidelines [Citation1,Citation7].
Compared with surgical resection, ablation is relatively less traumatic and is suitable for patients in poor physical condition or those with multiple or recurrent tumors [Citation8,Citation9]. Currently, the main methods for intrahepatic ablation are radiofrequency ablation (RFA) and microwave ablation (MWA) [Citation10]. Both RFA and MWA can inactivate liver tumors by heating the tissue; however, compared with RFA, MWA has higher thermal efficiency and is less affected by intrahepatic blood vessels [Citation10–12].
However, acting as a double-edged sword, its high thermal efficiency also makes the ablation zone (AZ) of MWA difficult to control, and the AZ of MWA cannot be monitored by measuring tissue resistance, similarly as with RFA [Citation10]. Although there are some studies on the AZ range based on computer simulations and phantom and animal experiments [Citation13–15], the assumptions in the computer simulations are too ideal, and phantoms and animal livers are quite different from the real clinical environment. Moreover, the factors influencing the AZ analyzed in these studies (such as tissue density and electrical conductivity) are difficult for radiologists to directly translate into their typical clinical observations [Citation13,Citation14]. In contrast, the physicochemical properties of the liver parenchyma might vary with the disease, and blood perfusion also changes according to different intrahepatic zones [Citation16,Citation17]. All of these factors might influence the AZ, but cannot be perfectly simulated in a computer, phantom, and animal experiments. In this study, we collected postoperative MRI data from patients with different physical conditions and explored the factors associated with the AZ based on postoperative MRI rather than simulated, phantom, or animal experiments.
Therefore, the purpose of this study was to analyze the influence of noninvasive and easily available clinical factors on the AZ to assist in planning individual ablation parameters of each patient before treatment.
Methods
Patients
The Ethics Committee Board of Chinese PLA General Hospital approved this retrospective study (S2021-439-01) and waived the requirement for patient consent. The inclusion criteria were as follows: (1) Tumor size smaller than 2 cm, HCC or metastatic cancer; (2) Ultrasound-guided curative intent MWA was performed using the following parameters: single needle, ablation power of 50 W or 60 W, and ablation time of 240 s, 300 s, or 360 s; (3) The MW antenna was a KY-2450 11 mm type or KY-2450 05 mm type, (4) Postoperative MRI within three days after MWA, and (5) The primary lesion was treated in patients with metastatic cancer. Exclusion criteria were as follows: (1) Liquefied necrotic area inside the tumor, (2) Palliative MWA; (3) Multiple overlapping ablations; (4) Missing or poor-quality images; and (5) Artificial pleural effusion or ascites were used because of the tumor was located near the liver capsule ().
Figure 1. Study flow chart. MWA:microwave ablation; AZ: ablation zone.
We collected postoperative contrast-enhanced MRI data within three days after MWA and 14 factors potentially influencing the AZ. The demographic factors included sex and age. Serological factors included alanine aminotransferase (ALT), aspartate aminotransferase (AST), total protein (TP), albumin, total bilirubin (TB), direct bilirubin (DB), and platelet (PLT). Disease-related factors included tumor location (according to the distance between the tumor and the hepatic hilum, the tumor location was divided into near the hepatic hilar zone, the intermediate zone, and the periphery zone) and liver cirrhosis. The MWA parameters included the ablation power and time and the MW antenna type. Detailed information on the diagnostic criteria for cirrhosis [Citation18,Citation19], tumor location, MW antenna type, and the type of ultrasound and MRI device are provided in the Supplementary material. The ablation zone of the MW antenna in the phantom is listed in Supplementary Table 1.
Delineated AZ on Postoperative MRI
The AZ was delineated on the postoperative MRI by 3Dslicer, and a 3D model was generated. To reduce subjectivity, we used the threshold function in 3Dslicer to help the radiologist complete the AZ delineation (the delineating method is described in the Supplementary material). Then, we calculated the area of the AZ section at each angle by continuously rotating and projecting the 3D model, and defined the section with the largest area as the standard section (). This process can help reduce the influence of local deformation caused by the heat sink effect on the choice of AZ standard sections [Citation8] because the AZ area of a section with local deformation is smaller than that of a section without local deformation. This processing was performed using MATLAB (version 2018) to ensure the full utilization of each case and the robustness of this study (, corresponding text in the Supplementary material).
Figure 2. Postoperative MRI image processing. (a) Selecting MRI sequence; (b) Using the threshold function that comes with 3Dslicer to delineate AZ; (c) Correcting the outline of AZ manually; (d) Generating the 3D-shape of AZ automatically; (e) Rotating and projecting the 3D-shape to find the section with the largest area as the standard section of AZ; (f) Calculating the AZ length width and area. AZ: ablation zone.
Consistency of the AZ in different MRI sequences
Because the AZ display format in different MRI sequences was inconsistent [Citation20,Citation21], we evaluated the consistency of AZ in different MRI sequences and chose the most stable MRI sequence for AZ delineation. Five radiologists (three with 3+ years of experience and two with 5+ years of experience) delineated 10 randomly selected AZs on five MRI sequences (T2WI, T1WI, early arterial period, late arterial period, and delayed period), calculated the length, width, and area of the AZ, and calculated the coefficient of variation (CV) for each sequence. The sequence with the smallest CV was chosen as the standard sequence, and the remaining AZs were delineated on the standard sequence by three radiologists (two with 3+ years of experience and one with 5+ years of experience). The delineation process was completed by three radiologists and the final result was recorded as the average of the three radiologists. We also calculated the CV for each remaining AZ, and when the CV of the AZ was more than 1.5 times the CV in the pre-experiment, delineation of the AZ was completed through consultation with all five participating radiologists.
Analysis of factors influencing the AZ
Multiple linear regression and stepwise linear regression were used to analyze the influence of the various factors on the AZ length, width, and area, and to calculate the standardized coefficient (beta) of each factor. We then calculated the Fibrosis 4 (FIB4) score and performed a subgroup analysis for patients with viral hepatitis.
Statistics
Continuous variables are represented as mean ± standard deviation, and categorical variables are represented by numerical values (n%). In linear regression analysis, cirrhosis was set to 1, non-cirrhosis was set to 2, tumor located in the periphery zone was set to 1, in the intermediate zone was set to 2, in the near hepatic hilar zone was set to 3; and MW antenna 11 mm-type was set to 1 and 05 mm-type was set to 2. A normal probability distribution plot (P-P plot), standardized residual histogram, standardized prediction-residual plot, Durbin–Watson test, and variance inflation factor (VIF) were used to ensure the precondition of linear regression: the residual distribution conformed to normality, independence, and homogeneity of variance, and there was no multicollinearity for multiple independent variables. Multiple linear regression and stepwise linear regression were used for statistical analysis, and the F value for stepwise linear regression was set at 0.05. The statistical software used was SPSS version 26.
Results
A total of 220 patients with 290 tumors were included in this study between January 2010 and December 2021. The baseline characteristics are presented in .
Table 1. Baseline of all cases.
In this study, we analyzed sex, age, cirrhosis, tumor location, ALT, AST, TP, ALB, TB, DB, PLT, ablation power, time, and needle type, but not tumor type or size. All cases in this study used a curative ablation strategy, which required the AZ range to cover more than 5 mm beyond the tumor boundary. Therefore, the effect of tumor size on AZ range was significantly attenuated. Additionally, there was a high collinearity between HCC and liver cirrhosis, and including similar factors in the linear regression would have seriously interfered with the analysis results.
Five radiologists completed the delineation of 10 AZs on five postoperative MRI sequences and calculated the CV for each sequence (, Supplementary Table 2). The delay period had the lowest CV (the CVs of the AZ length, width, and area were 4.21 ± 0.80%, 4.39 ± 0.77%, and 8.38 ± 1.08%, respectively) and was selected as the standard sequence. Three radiologists completed the delineation of the remaining cases on the delayed-period MRI.
Figure 3. CV of multiple MRI sequences for AZ length (a); width (b); area (c). CV: coefficient of variation; AZ: ablation zone.
A precondition analysis of the linear regression is shown in Supplementary Figure 1. The residual distributions of each model conformed to normality and homogeneity of variance. The results of the Durbin–Watson test for each model were approximately 2.0, indicating that the residuals of the models were independent. The VIFs of ALT, AST, TB, and DB were less than 5, and the VIFs of the other variables were less than three, indicating that there was no significant multicollinearity among these variables.
Multiple linear regression and stepwise linear regression showed that liver cirrhosis, tumor location, ablation time, ablation power, and MW antenna type were factors influencing AZ length, width, and area (). The adjusted R2s of the stepwise linear regression of the AZ length, width, and area were 0.431, 0.483, and 0.579, respectively. The beta of the ablation time and power in all models was positive, indicating that the AZ length, width, and area increased as the ablation time and power increased. The beta of the MW antenna type was negative, indicating that the AZ caused by the 11 mm-type MW antenna was greater than that of the 05 mm-type antenna. The beta of liver cirrhosis was positive, indicating that the AZ of noncirrhotic patients was larger than that of cirrhotic patients. The beta of the tumor location was negative, indicating that the AZ decreased as the tumor location approached the hepatic hilum.
Table 2. Linear Regression for AZ in all cases.
We then performed subgroup analyses of viral hepatitis patients (172 patients with 207 tumors) and introduced FIB4 to represent the degree of liver fibrosis. Baseline data are presented in Supplementary Table 3. Since FIB4 was calculated using serological factors, to prevent collinearity, we excluded any related factors (age, ALT, AST, and PTL) and cirrhosis. Multiple linear regression and stepwise linear regression showed that FIB4 was the major factor influencing the AZ length, width, and area (, precondition analysis in Supplementary Figure 2), and the betas were all negative, indicating that with an increase in FIB4, the AZ decreased.
Table 3. Linear Regression for AZ in hepatitis patients.
Discussion
Ablation is a minimally invasive treatment recommended by many guidelines for liver tumors, with the advantages of less trauma, faster recovery, and broader indications [Citation9,Citation22]. Compared with surgical resection, ablation is especially suitable for patients with metastatic cancer, multiple tumors, recurrent tumors, or poor physical condition [Citation8,Citation23]. Currently, the main ablation methods used in the liver are RFA and MWA [Citation24]. Although both methods use thermal ablation, their thermogenic mechanisms are different. Heat generation by RFA uses the heating effect of high-frequency alternating current to heat the tumor and surrounding areas, whereas heat generation from MWA uses microwaves to make water molecules move back and forth rapidly, causing friction between the molecules [Citation10,Citation12]. The difference in the thermogenic mechanisms produces a higher thermal efficiency of MWA than of RFA but also introduces difficulties in controlling the AZ range of MWA.
In previous studies, two methods were used to analyze the AZ range. First, theoretical studies were based on the Pennes biological heat conduction equation and computer simulations of thermal fields [Citation25]. Second, practical studies were based on phantom and animal experiments [Citation26]. However, neither of these methods can accurately simulate or reconstruct tumors and the surrounding tissues in the human body. In addition, both types of studies focus on a description of the AZ range but do not analyze the factors influencing the AZ range.
Although some clinical investigations have attempted to explore variations in the ablation zone between different tumor types or different ablation needle types, these studies have two limitations [Citation27,Citation28]. First, these studies did not exclude cases of multi-needle ablation or needle movement during the process. For multi-needle ablation, the energy between the ablation needles is significantly different from that outside the ablation needles. Moreover, the influence of the ablation needle on the AZ during its movement could not be quantified. Second, these studies were not rigorous enough to determine the AZ. AZ can be affected by blood vessels and ligaments; therefore, symmetry and integrity cannot be ensured in all directions. Thus, the measurement of the AZ volume would be affected by the surrounding structure, and this influence cannot be quantified. However, in our study, we selected the maximum cross-section of the AZ as the standard section to measure its area. The AZ sections affecting the surrounding structure were automatically excluded because the area of the affected section would obviously be smaller than that of the unaffected section. This process ensured that the measurement of the AZ area would not be affected by the surrounding structure.
To ensure the robustness of the AZ range calculation, we compared the CVs of the AZ range on different MRI sequences and found that the CVs of the late arterial and delayed periods were significantly smaller than those of the T2WI, T1WI, and early arterial periods. There might be two reasons for this: First, the layer thickness of T2WI was higher than that of the other MRI sequences, resulting in a rougher delineation on T2WI; Second, during the process of inactivating tumors, MWA led to the destruction of blood perfusion in the AZ, so the contrast between inactivated AZ and activated liver parenchyma would be more pronounced in MRI sequences that can reflect blood perfusion. Therefore, this contrast was obvious in the late arterial and delayed periods, but not in the T2WI, T1WI, or early arterial period.
Based on the AZ delineated on the delayed-period MRI, we analyzed the factors influencing the AZ by linear regression. In terms of the MWA parameter, we found that the AZ range increased with the ablation time and power, and the AZ of the 11 mm-type MW antenna was larger than that of the 05 mm-type MW antenna. This might be because the 11 mm-type MW antenna is longer than the 05 mm-type, resulting in a larger microwave radiation field (Supplementary Figure 3).
In terms of disease-related factors, we found that the AZ range in cirrhotic patients was smaller than that in noncirrhotic patients. Combining the Pennes biological heat conduction equation and pathological alterations of the liver parenchyma, we speculated that there may be two reasons for this result. First, when cirrhosis occurs, fibrous tissue proliferates within the liver parenchyma, resulting in increased tissue density. Tissue density was negatively correlated with the AZ range according to the formula provided in previous studies [Citation16]. Second, when cirrhosis occurs, the water content in the liver parenchyma increases [Citation17], which results in an increase in tissue-specific heat capacity. Therefore, under the same energy, the range of tissues heated by microwaves is reduced. Third, the liver parenchyma is subject to substantial shrinkage upon MW heating owing to the massive vaporization of its water content. The higher the initial water content, the more tissue contracts because of water vaporization, and the water content in cirrhosis parenchyma is higher than that in non-cirrhosis parenchyma, so the AZ would be smaller in the cirrhosis parenchyma than in the non-cirrhosis parenchyma. Moreover, a higher water content leads to more water vapor being generated during the heating process, and gasification consumes more energy [Citation26]. Furthermore, as the generated water vapor enters the liver tissue, the tissue conductivity further decreases [Citation14]. All these factors act together and decrease the AZ range in cirrhotic tissue.
In terms of disease-related factors, we also found that the AZ range decreased when the tumor was close to the hepatic hilum. Combining the Pennes biological heat conduction equation and the blood flow distribution of the liver parenchyma [Citation29,Citation30], we speculated that the liver parenchyma near the hilar region had richer blood perfusion than that in the peripheral zone, which led to more obvious liquid heat dissipation by blood flow. This effect could remove more heat and cause a reduction in the AZ range when liver tumors are located near the hepatic hilar zone.
In the subgroup analysis of viral hepatitis patients, we included FIB4, an index that has been confirmed to be correlated with the degree of liver fibrosis, in linear regression. The beta value of FIB4 was negative, indicating that the AZ range decreased with increasing FIB4. We speculated that since fibrosis and liver cirrhosis are continuous processes in pathology, the influence of fibrosis on AZ may be similar to that of liver cirrhosis, and this influence could be reflected through FIB4.
Our study had some limitations. First, our study did not analyze the relationship between the AZ range on MRI and the actual AZ range in the human body. Therefore, the range of tissue inactivation actually caused by MWA may not exactly match the range of AZ observed on MRI. However, our study aimed to analyze the factors influencing the AZ and not to predict or rebuild the AZ shape. Second, different types of MRI devices were used, which may have caused inconsistencies in AZ imaging. However, the MRI devices used in all cases were 1.5 T or 3.0 T, which could provide sufficient clarity for the AZ range, and our research carried out a rigorous consistency test on the MRI images. Third, this was a retrospective single-center study. Due to different medical practices, there are some differences in ablation parameters, such as power 70w–100 w, which we did not include. However, a detailed precondition test was performed before the linear regression to ensure the reliability of our results.
In this study, we collected data on 14 common clinical factors and ablation parameters and evaluated their effects on AZ. We found that the AZ range decreased in the presence of liver cirrhosis and as the tumor location approached the hepatic hilum. In patients with viral hepatitis, the AZ range decreased as FIB4 increased. This result can guide doctors in planning MWA parameters individually before treatment and provide a basis for fundamental and artificial intelligence research on AZ.
Conclusion
Our study analyzed the factors influencing the AZ range and found that the AZ range decreased in the presence of liver cirrhosis and as the tumor location approached the hepatic hilum. In patients with viral hepatitis, the AZ range decreased as FIB4 increased.
Ethical statement
The Ethics Committee Board of Chinese PLA General Hospital approved this retrospective study (S2021-439-01) and waived the requirement for patient consent for this retrospective review.
Supplemental Material
Download PDF (470.6 KB)Disclosure statement
None of the authors have a conflict of interest to disclose.
Additional information
Funding
References
- European Association for the Study of the Liver. Electronic address: [email protected], EASL clinical practice guidelines: management of hepatocellular carcinoma. J. Hepatol. 2018;69(1):182–236.
- Thompson SM, Welch BT, Kurup AN. Ablation for oligometastatic colorectal carcinoma in extrahepatic, extrapulmonary sites. Int J Hyperthermia. 2022;39(1):633–638.
- Marrero JA, Kulik LM, Sirlin CB, et al. Diagnosis, staging, and management of hepatocellular carcinoma: 2018 practice guidance by the American association for the study of liver diseases. Hepatology. 2018;68(2):723–750.
- Filippiadis DK, Efthymiou E, Palialexis K, et al. Interventional oncology procedures for breast cancer metastatic disease: current role and clinical applications. Diagn Interv Radiol. 2022;28(3):249–256.
- Zhang ZY, Jiang AN, Yang W, et al. Percutaneous radiofrequency ablation is an effective method for local control of liver metastases from lung cancer. Front Oncol. 2022;12:877273.
- Mansur A, Garg T, Shrigiriwar A, et al. Image-guided percutaneous ablation for primary and metastatic tumors. Diagnostics . 2022;12(6):1300.
- Vogel A, Cervantes A, Chau I, et al. Hepatocellular carcinoma: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2018;29(Suppl 4):iv238–iv255.
- Ahmed M, Solbiati L, Brace CL, Guided tumor ablation:standardization of terminology and reporting criteria—a 10-year update. Radiology. 2014;273(1):241–260.
- Wang Z, Liu M, Zhang DZ, et al. Microwave ablation versus laparoscopic resection as first-line therapy for solitary 3–5-cm HCC. Hepatology. 2022;76(1):66–77.
- Llovet JM, De Baere T, Kulik L, et al. Locoregional therapies in the era of molecular and immune treatments for hepatocellular carcinoma. Nat Rev Gastroenterol Hepatol. 2021;18(5):293–313.
- Zhu F, Rhim H. Thermal ablation for hepatocellular carcinoma: what’s new in 2019. Chin Clin Oncol. 2019;8(6):58–58.
- Yu J, Yu XL, Han ZY, et al. Percutaneous cooled-probe microwave versus radiofrequency ablation in early-stage hepatocellular carcinoma: a phase III randomised controlled trial. Gut. 2017;66(6):1172–1173.
- Cavagnaro M, Pinto R, Lopresto V. Numerical models to evaluate the temperature increase induced by ex vivo microwave thermal ablation. Phys Med Biol. 2015;60(8):3287–3311.
- Ji Z, Brace CL. Expanded modeling of temperature-dependent dielectric properties for microwave thermal ablation. Phys Med Biol. 2011;56(16):5249–5264.
- Wang X, Gao H, Wu S, et al. RF ablation thermal simulation model: parameter sensitivity analysis. Technol Health Care. 2018;26(S1):179–192.
- O’Rourke AP, Lazebnik M, Bertram JM, et al. Dielectric properties of human normal, malignant and cirrhotic liver tissue: in vivo and ex vivo measurements from 0.5 to 20 GHz using a precision open-ended coaxial probe. Phys Med Biol. 2007;52(15):4707–4719.
- Sy S, Huang S, Wang YX, et al. Terahertz spectroscopy of liver cirrhosis: investigating the origin of contrast. Phys Med Biol. 2010;55(24):7587–7596.
- Danielsen KV, Wiese S, Busk T, et al. Cardiovascular mapping in cirrhosis from the compensated stage to hepatorenal syndrome: a magnetic resonance study. Am J Gastroenterol. 2022;117(8):1269–1278.
- Zhang XQ, Zheng RQ, Jin JY, et al. US shear-wave elastography dispersion for characterization of chronic liver disease. Radiology. 2022;305(3):597–605.
- Huang X, Zhou Y, Wang C, et al. Development of a novel MR-conditional microwave needle for MR-guided interventional microwave ablation at 1.5T. Magn Reson Med. 2022;88(4):1886–1900.
- Hensen B, Drenkmann U, Frericks B, et al. Detection of ablation boundaries using different MR sequences in a swine liver model. Cardiovasc Intervent Radiol. 2022;45(7):1010–1018.
- Ding W, Yu J, Liu F, et al. Percutaneous microwave ablation versus robot-assisted hepatectomy for early hepatocellular carcinoma: a real-world single-center study. Dig Liver Dis. 2022;54(2):243–250.
- Nieuwenhuizen S, Dijkstra M, Puijk RS, et al. Microwave ablation, radiofrequency ablation, irreversible electroporation, and stereotactic ablative body radiotherapy for intermediate size (3–5 cm) unresectable colorectal liver metastases: a systematic review and meta-analysis. Curr Oncol Rep. 2022;24(6):793–808.
- Xu Y, Shen Q, Wang N, et al. Microwave ablation is as effective as radiofrequency ablation for very-early-stage hepatocellular carcinoma. Chin J Cancer. 2017;36(1):14.
- Ren H, Guo W, Sam Ge S, et al. Coverage planning in computer-assisted ablation based on genetic algorithm. Comput. Biol. Med. 2014;49:36–45.
- Ai H, Wu S, Gao H, et al. Temperature distribution analysis of tissue water vaporization during microwave ablation: experiments and simulations. Int J Hyperthermia. 2012;28(7):674–685.
- Heerink WJ, Solouki AM, Vliegenthart R, et al. The relationship between applied energy and ablation zone volume in patients with hepatocellular carcinoma and colorectal liver metastasis. Eur Radiol. 2018;28(8):3228–3236.
- Ruiter SJS, de Jong JE, Pennings JP, et al. Comparison of two 2.45-GHz microwave ablation devices with respect to ablation zone volume in relation to applied energy in patients with malignant liver tumours. Cancers. 2022;14(22):5570.
- Sakamoto Y, Kokudo N, Kawaguchi Y, et al. Clinical anatomy of the liver: review of the 19th meeting of the Japanese research society of clinical anatomy. Liver Cancer. 2017;6(2):146–160.
- Ziessman HA, Thrall JH, Gyves JW, et al. Quantitative hepatic arterial perfusion scintigraphy and starch microspheres in cancer chemotherapy. J. Nucl. Med. 1983;24(10):871–875.