Radiofrequency Thermal Ablation: Computer Analysis of the Size of the Thermal Injury Created by Overlapping Ablations

Transcription

1 Gerald. odd III 1 Mark S. Frank 1,2 Manohar ribandi 1 Shailendra hopra 1 Kedar N. hintapalli 1 Received February 1, 2001; accepted after revision pril 18, epartment of Radiology, The University of Texas Health Science enter at San ntonio, 7703 Floyd url r., San ntonio, TX ddress correspondence to G.. odd. 2 Present address: epartment of Radiology, Massachusetts General Hospital, 55 Fruit St., oston, M JR 2001;177: X/00/ merican Roentgen Ray Society Radiofrequency Thermal blation: omputer nalysis of the Size of the Thermal Injury reated by Overlapping blations OJETIVE. The purpose of this study was to perform a computer analysis of the size of the thermal injury created by overlapping multiple thermal ablation spheres. MTERILS N METHOS. computer-assisted design system was used to create three-dimensional models of a spherical tumor, a spherical tissue volume consisting of the tumor plus a 1-cm tumor-free margin, and individual spherical ablations. These volumes were superimposed in real-time three-dimensional space in different geometric relationships. The effect of the size and geometric configuration of the ablation spheres was analyzed with regard to the ability to ablate the required volume of tissue (tumor plus margin) without leaving untreated areas or interstices. RESULTS. The single-ablation model showed that if a cm tumor-free margin is included around the tumor targeted for ablation, radiofrequency ablation devices producing 3-, 4-, and 5-cm ablation spheres can be used to treat 1-, 2-, and 3-cm tumors, respectively. The six-sphere model, in which six ablation spheres are placed in orthogonal planes around the tumor, showed that the largest tumor that may be treated with a 3-cm ablation device is 1.75 cm, whereas 4- and 5- cm ablation spheres can be used to treat tumors measuring 3 and 4.25 cm, respectively. The 14- sphere model showed that addition of eight more spheres to the six-sphere model increased the treatable tumor size to 3, 4.6, or 6.3 cm, depending on the diameter of the ablation sphere used. For treating larger tumors, we found a cylindrical model to be less efficient but easier to control. ONLUSION. Our computer analysis showed that the size of the composite thermal injury created by overlapping multiple thermal ablation spheres is surprisingly small relative to the number of ablations performed. These results emphasize the need for a methodic tumor ablation strategy. R adiofrequency thermal ablation has shown promise as an effective minimally invasive technique for the treatment of primary and secondary malignant hepatic tumors [1 4]. The complete tumor-kill rate for ablated tumors has been reported to be as high as 98% [5]. However, a moderate to high recurrence rate in local tumors has been documented in multiple studies [6 9]. The single most important factor affecting the local tumor recurrence rate after hepatic resection is the presence and size of a tumor-free margin of hepatic parenchyma along the resection margin. It has been clearly documented that tumor-free resection margins of less than 1 cm are directly related to increased local hepatic tumor recurrence rates and decreased overall patient survival [10, 11]. Our experience and a review of the published studies on radiofrequency ablation of hepatic tumors suggest that the high failure rate in local tumors seen with radiofrequency ablation treatment may be due to a gross underablation of the tumors and a failure to create an adequate tumor-free margin [6 9] (Fig. 1). lthough it seems intuitively logical that adding a 1-cm tumor-free margin would result in an increase in the overall ablation volume required to adequately treat a tumor, no study to date has been conducted to determine how large an ablation should be to treat tumors of various sizes or how ablations should be overlapped to treat larger tumors effectively. Therefore, we have used a computer-generated three-dimensional (3) model to analyze the size of the thermal injury created by solitary and overlapping thermal ablation spheres. The model takes into account the volume of a tumor and the creation of an adequate tumor-free margin. The software renders an easily understood 3 model that facilitates a JR:177, October

2 odd et al. systematic approach to treatment using the least number of ablations. We present the results derived from the model and its impact on thermal ablation planning. Materials and Methods computer-assisted design system (esign Pro 2000; ViaGrafix, Pryor, OK) was used to create 3 models of a spherical tumor, a spherical tissue volume consisting of the tumor plus a cmthick tumor-free margin, and individual spherical ablations. These were superimposed in real-time 3 space in different geometric relationships. We analyzed the effect of size and geometric configuration of the ablation spheres on the ability to ablate the required volume of tissue (tumor plus margin) without leaving untreated areas or interstices. The geometric configurations of the models were chosen based on the efficiency of filling a spherical space. We created models with one, six, and 14 spheres and overlapped the ablation spheres so that the size of a composite sphere was maximized and all the space within the composite sphere was encompassed by one or more of the individual spheres. We also evaluated the usefulness of creating overlapping thermal cylinders or columns. In this model, multiple spheres were overlapped along a single axis to create a thermal cylinder. Multiple cylinders were then overlapped to create adjacent rectangular columns. s with the spherical models, ablation spheres were overlapped in the most efficient manner to maximize the composite volume. Results Single-blation Model To adhere to the surgical principle of having a 1-cm tumor-free margin, the ablation of a spherical tumor requires that the tumor be completely enveloped by an ablation sphere that is of sufficient size to encompass the entire tumor as well as a cm-thick tumor-free rind. This 1-cm rind adds 2 cm to the diameter of the ablation that must be created to ablate a tumor and achieve a surgical margin (Fig. 2). The adoption of a surgical margin has considerable impact on the size of a tumor that can be treated by a single ablation. If a radiofrequency ablation device produces a 3-cm solitary ablation, then the largest tumor that should be treated with a single ablation is 1 cm. For devices producing 4- and 5-cm ablation spheres, the largest tumors that can be treated adequately would be 2 and 3 cm, respectively. Six-blation Model The six-sphere model is assembled with four spheres in the x y plane and two along the z- axis (Fig. 3). y adjusting the degree of overlap of adjacent spheres with the computer-aided design program, we found that a 23% overlap of the diameter of the spheres produced the largest composite internal sphere without gaps. If correctly positioned, the largest diameter of the inner composite sphere is equal to 1.25 times the diameter of a single sphere. For single-ablation spheres that are 3, 4, and 5 cm in diameter, the six-ablation model yields composite spheres that are 3.75, 5, and 6.25 cm in diameter, respectively. If a 1-cm surgical margin is applied to this model, the largest tumor that should be treated with a 3-cm ablation device is 1.75 cm, whereas for devices producing 4- and 5-cm ablation spheres, the largest treatable tumor size would be 3 and 4.25 cm, respectively. The limiting factor in the size of the composite ablation volume is the depth of the pits at the junction of three spheres (Fig. 4). 14-blation Model To encompass a contiguous spherical composite ablation volume greater than that achievable with the six-sphere model, eight additional spheres must be added to cover the Fig year-old man with hepatocellular carcinoma., Enhanced T scan obtained before ablation shows 2.5-cm hepatocellular carcinoma., Enhanced T scan obtained at the same level as 6 months after tumor was treated with single 3-cm ablation shows ablation area equal to size of tumor with no residual tumor detected., Resected liver specimen 6 months after ablation shows residual tumor (arrows) not revealed at T. Fig. 2. omputer representation of single-ablation model. Effective ablation must encompass tumor plus cm tumor-free margin. omputer drawing depicts tumor plus half of effective tumor-free margin (red sphere). This 360 margin adds 2 cm to overall diameter of ablation sphere, depicting ablation volume encompassing tumor and tumor-free margin. 778 JR:177, October 2001

3 Size of Thermal blations of Liver Tumors Fig. 3. omputer representation of construction of six-sphere thermal ablation model. -, Six-sphere ablation model is constructed by performing four ablations in the x y plane ( sequentially) and two along the z- axis (). Green sphere represents total volume of tissue requiring ablation (tumor plus 1-cm tumor-free margin), and red spheres represent individual thermal ablation spheres that are being overlapped. Largest composite ablation sphere is created when all spheres are overlapped by approximately 23% of diameter. Fig. 4. omputer representation of six-sphere thermal ablation model depicting overlapping ablation spheres that will create composite spherical ablation encompassing tumor and tumorfree margin., Six-sphere model shows fissures at intersection of spheres., ross-section through middle of model shows that maximum composite spherical ablation (green area) does not touch margins at midsection., Size of composite spherical ablation (green area) is limited by pits at intersection of three spheres., ut section of model shows size of tumor that can be adequately treated (yellow area), taking 1-cm tumor free margin (green area) into account. JR:177, October

4 odd et al. eight pits in the six-sphere model (a total of 14 ablations, Fig. 5). With all 14 spheres positioned to maximize inner volume, this model yields a composite spherical ablation diameter that is only 1.66 times the diameter of a single-ablation sphere. Thus, a single-ablation sphere of 3 cm in diameter would yield a composite spherical ablation diameter of 5 cm. Incorporating a 1-cm tumor-free margin into the model would limit the maximum treatable tumor diameter to 3 cm. For devices producing 4- and 5-cm ablation spheres, the corresponding composite ablation sphere diameters would be 6.6 and 8.3 cm, which could be used to treat tumors measuring 4.6 and 6.3 cm, respectively. ylindrical blation Model Recognizing that the 14-sphere model is technically impossible to execute clinically, we created a model of overlapping thermal cylinders or columns that could be used to treat larger tumors (Fig. 6). onceptually, the spheres are overlapped to create a cylinder, and the cylinders are overlapped as necessary to systematically ablate a tumor. In analyzing the overlap of adjacent spheres, we found that optimal 3 positioning reduced the configuration of the effective volume of the spheres to cubes, the sides of which are equal to 58% of the diameter of the sphere. This model, although technically easier to execute than the 14-sphere model, is much less efficient geometrically. For example, an eight-sphere model that renders a cube with two spheres along each edge yields a composite inner spherical volume with a diameter that is only 1.16 times that of each single-ablation sphere. In comparison, the sixsphere model previously described yields a composite ablation volume with a diameter that is 1.25 times that of the individual ablation sphere. reating a composite spherical ablation equal to the 14-sphere model would require a cylindrical model with 27 overlapping ablations that form a cube with three ablations on each side. Fig. 5. omputer representation of 14-sphere thermal ablation model., Six-ablation model shows untreated tissue (green areas) at each pit., Eight additional spheres (in blue) are used to cover eight pits., ross-section through middle of 14-sphere ablation model shows maximum diameter of composite spherical ablation (green area) that is 1.66 times diameter of single ablation., ut section of model shows size of tumor that can be adequately treated (yellow sphere) taking 1-cm tumor-free margin (green area) into account. iscussion Surgical resection of primary hepatocellular carcinoma and hepatic metastases from primary neoplasms, such as colorectal adenocarcinoma, has been shown to increase the survival and the disease-free period in carefully screened patients [12]. Percutaneous ablation techniques provide a valuable alternative to patients who do not meet surgical criteria for hepatic resection. These methods include intratumoral ethanol instillation, cryoablation, laser ablation, microwave ablation, and radiofrequency thermal ablation [4]. Tumor recurrence is the major cause of morbidity and mortality in patients who have undergone curative resection of hepatocellular carcinoma or of liver metastases from primary tumors, such as colorectal cancer. In the case of liver metastases from colorectal cancer, Pinson et al. [13] found that the overall tumor recurrence rate at a mean period of 13 months was 71%, with the liver being involved in 55% of affected patients. The liver 780 JR:177, October 2001

5 Size of Thermal blations of Liver Tumors Fig 6. omputer representation of cylindrical thermal ablation model., Thermal ablation cylinders are created by overlapping multiple ablation spheres by 58%., Overlapping of ablation cylinders creates rectangular ablation columns made up of individual ablation cubes., Sides of each ablation cube are equal to 0.58 times diameter of individual ablation sphere., blation cylinders are systematically overlapped to ablate large tumors. was the sole site of recurrence in 33% of patients. The exact incidence of intrahepatic tumor recurrence along the surgical resection margin is difficult to assess because it has not been widely reported in the surgical literature. Sugihara et al. [14] reported a local failure rate of 17% after surgical resection of colorectal metastases. This failure rate was caused by positive resection margins in 9% of patients and recurrence along the resection margin despite apparently adequate 1- cm tumor-free margins in 8% of patients. ozzetti et al. [15] reported a local hepatic tumor recurrence rate of 9% after surgical resection of colorectal liver metastases. Matsumata et al. [16] reported that in patients who had undergone curative resection of hepatocellular carcinoma, 21% of all the hepatic recurrences were at the site of the resected hepatic stump, approximating a local hepatic recurrence rate of 6.7%. The recurrence rate for local tumors after radiofrequency thermal ablation of primary and secondary malignant hepatic tumors varies from as low as 1.8% [5] to as high as 34 55% [6 9]. Several causes have been suggested for the less impressive results reported for thermal ablation treatment [6 9]. These include the technical difficulties in localizing the tumor and accurately placing the ablation electrode, heterogeneous thermodynamic properties of tissue in the ablation field (e.g., the cooling effects of blood vessels in close proximity), unpredictable volume and geometry of ablations achieved with specific ablation devices in the same patient and in different patients, underestimation of the size of the tumor or overestimation of the extent of thermal injury as visualized by different imaging techniques, and underestimation of the volume of ablation necessary to completely eradicate the tumors of varying sizes. When surgical excision is used to remove a primary or secondary malignant hepatic tumor, the desired tumor-free margin that is, the tissue without malignancy that surrounds the tumor and is excised with it is at least 1 cm. Surgical excision with a margin of less than 1 cm results in a higher recurrence rate for local tumors and decreased rate of survival when compared with those rates seen in excisions with margins of 1 cm or more [10, 11]. We have observed that the strategy used in thermal ablation of liver tumors does not appear to be sufficient to both ablate the tumor and provide an adequate tumor-free margin [6 9]. When a tumor-free margin is considered and the standards of surgical resection are applied to thermal ablation, the required ablation volume increases dramatically over that required to treat the tumor alone. For example, a 1-cm tumor would require an ablation diameter of 3 cm, or, conversely, the largest tumor diameter that could be effectively treated with a single 3-cm ablation is 1 cm. If the largest ablation that a treatment device can render is 3 cm and if a 1-cm margin is desired, then, assuming a spherical model, six ablations would be required to treat a tumor larger than 1 cm. To our knowledge, none of the radiofrequency ablation studies has followed the strict JR:177, October

6 odd et al. modeling that we have proposed. However, the high recurrence rates for local tumors reported in many of those studies suggest that tumors in many patients were undertreated. This interpretation of the results is also supported by the fact that local-tumor recurrence was more often seen in larger tumors than in smaller ones, conceivably because including all of the tumor plus a tumor-free margin within the ablation volume is difficult [5, 9, 17]. We believe that an appreciation of the effect of tumor margin on ablation strategy can result in a more thorough therapy and a diminished recurrence rate for local tumors. Furthermore, all radiofrequency devices do not produce spherical ablations nor do they produce ablations of equal size [18] (Fig. 7). Given this variation, it is imperative that an ablation strategy be modified for the radiofrequency device that is used. Our study has several potential limitations. Our computer models are based on perfect spheres that are perfectly placed; obviously, neither this shape nor placement is a reality in clinical practice. However, the irregularity of thermal lesions and the inevitable error in needle placement are further evidence of the need for an effective ablation strategy. Second, our ablation models are based on a sum effect, and no consideration has been given to the possibility that contiguous ablations may produce a composite ablation that exceeds simple volumetric addition. Last, we evaluated radiofrequency ablation as an isolated treatment. The size and effectiveness of individual ablations may be increased substantially by adjuvant techniques, such as chemoembolization, alcohol ablation, or temporary impairment of liver blood flow [19 21]. In conclusion, if radiofrequency ablation is going to challenge surgical resection as a viable alternative treatment for hepatic tumors, the complete tumor-kill rate achieved with radiofrequency ablation must approximate the complete excision rate achieved with hepatic resection. To accomplish this goal, physicians performing radiofrequency ablation must adhere to the surgical requirement of a tumor-free margin that is at least 1 cm thick. They must know the precise size and configuration of the thermal injury created by the ablation device being used, adhere to a strict ablation protocol, and be well versed in the 3 volume of the cumulative ablation sphere created by multiple overlapping ablations. cknowledgment We thank avid aker, our graphics artist, for his superb illustrations. References Fig. 7. Varying sizes of thermal ablation injury by different radiofrequency ablation devices. irrhotic liver of 63- year-old woman shows varying sizes and shapes of ablations (arrows) created before liver resection with three different radiofrequency ablation devices from different manufacturers. Marked variation emphasizes need for systematic ablation protocol. 1. McGahan JP, odd G III. Radiofrequency ablation of the liver: current status. JR 2001;176: Goldberg SN, Gazelle GS, Mueller PR. Thermal ablation therapy for focal malignancy: a unified approach to underlying principles, techniques, and diagnostic imaging guidance. JR 2000;174: Gazelle GS, Goldberg SN, Solbiati L, Livraghi T. Tumor ablation with radiofrequency energy. Radiology 2000;217: odd G III, Soulen M, Kane R, et al. Minimally invasive treatment of malignant hepatic tumors: at the threshold of a major breakthrough. RadioGraphics 2000;20: urley S, Izzo F, elrio P, et al. Radiofrequency ablation of unresectable primary and metastatic hepatic malignancies: results in 123 patients. nn Surg 1999;230: Solbiati L, Goldberg SN, Ierace T, et al. Hepatic metastases: percutaneous radiofrequency ablation with cooled-tip electrodes. Radiology 1997;205: Solbiati L, Ierace T, Goldberg SN, et al. Percutaneous US-guided RF tissue ablation of liver metastases: long-term follow up. Radiology 1997;202: Kainumao O, sano T, oyama H, et al. ombined therapy with radiofrequency thermal ablation and intra-arterial infusion chemotherapy for hepatic metastases from colorectal cancer. Hepatogastroenterology 1999;46: Livraghi T, Goldberg SN, Lazzaroni S, et al. Hepatocellular carcinoma: radiofrequency ablation of medium and large lesions. Radiology 2000;214: ady, Jenkins RL, Steele G Jr, et al. Surgical margin in hepatic resection for colorectal metastases: a critical and improvable determinant of outcome. nn Surg 1998;227: Elias, avalcanti, Sabourin J, et al. Resection of liver metastases from colorectal cancer: the real impact of the surgical margin. Eur J Surg Oncol 1998;24: Neeleman N, ndersson R. Repeated liver resection for recurrent liver cancer. r J Surg 1996;83: Pinson W, Wright JK, hapman W, Garrard L, lair TK, Sawyers JL. Repeat hepatic surgery for colorectal cancer metastasis to the liver. nn Surg 1996;223: Sugihara K, Hojo K, Moriya Y, Yamasaki S, Kosuge T, Takayama T. Pattern of recurrence after hepatic resection for colorectal metastases. r J Surg 1993;80: ozzetti F, ignami P, Morabito, oci R, Gennari L. Patterns of failure following surgical resection of colorectal cancer liver metastases. nn Surg 1987;205: Matsumata T, Kanematsu T, Takenaka K, Yoshida Y, Nishizaka T, Sugimachi K. Patterns of intrahepatic recurrence after curative resection of hepatocellular carcinoma. Hepatology 1989;9: Lencioni R, Goletti O, rmillotta N, et al. Radiofrequency thermal ablation of liver metastases with a cooled-tip electrode needle: results of a pilot clinical trial. Eur Radiol 1998;8: de aere T, enys, Johns Wood, et al. Radiofrequency liver ablation: experimental comparative study of water-cooled versus expandable systems. JR 2001;176: Goldberg SN, Kruskal J, Oliver S, louse ME, Gazelle GS. Percutaneous tumor ablation: increased coagulation by combining radiofrequency ablation and ethanol instillation in a rat breast tumor model. Radiology 2000;217: Rossi S, Garbagnati F, Lencioni R et al. Percutaneous radiofrequency thermal ablation of nonresectable hepatocellular carcinoma after occlusion of tumor blood supply. Radiology 2000;217: Goldberg SN, Hahn PF, Tanabe KK. Percutaneous radiofrequency tissue ablation: does perfusion-mediated tissue cooling limit coagulation necrosis? J Vasc Interv Radiol 1998;9: JR:177, October 2001