Defining the Parameters for Endovenous Microwave Ablation to Achieve Equivalence With Endovenous Laser Ablation, Using the Porcine Liver Model

Aims Endovenous microwave ablation (EMWA) is an endovenous thermoablation (EVTA) system to ablate incompetent truncal veins. Early results suggest that EMWA uses more power than endovenous laser ablation (EVLA) to get the same results. We aimed to define the parameters for EMWA, which give the same tissue ablation as EVLA, using the validated porcine liver model. Methods EVLA (1470 nm 600 micron radial fibre) treatments were performed at 6 W, 8 W and 10 W, at pullback speeds of 6, 7, 8 and 9 s/cm, giving Linear Endovenous Energy Densities (LEEDs) between 36 - 90 J/cm. Each combination of power and pullback was repeated 5 times. We then used EMWA in the same model. Powers of 35-75 W and pullback speeds of 4-9 s/cm were used (LEEDs 140-675 J/cm). Ablation tracts from both devices were analysed by 2 blinded observers, noting thermal spread and carbonisation. Results For each commonly used parameter for EVLA, we identified a range of parameters for EMWA that produced similar tissue ablation in the porcine liver model. To keep the pullback speeds within the usual range, powers of 35-75 W were needed with EMWA, with mean EMWA LEEDs 3.9 - 5.8 times higher than EVLA LEEDs. We found the quicker the pullback speed, the higher the multiple of EMWA LEED we needed to get the same effect. Conclusion We have identified parameters for EMWA that gave equivalent tissue ablation in the porcine liver model to commonly used parameters used for EVLA. These need to be validated clinically, but as the model used has already been validated against clinical outcome in endovenous thermal ablation, there is little reason to suspect that these results would not be valid. As the power during EMWA is higher than EVLA, EVMA LEEDs are approximately 4-6 times higher than EVLA LEEDs to achieve the same thermal effect on the tissues.


Introduction
][3] The most widely used methods of EVTA are radiofrequency ablation (RFA) and endovenous laser ablation (EVLA).Endovenous microwave ablation (EMWA) is a relatively new catheter-based EVTA system.Although reasonable ablation rates have been reported 95-100% at 12 months [4][5][6] and even 98% at 5 years, 7 the power parameters currently used vary widely, from 20 W-65 W, [4][5][6][7] with a future multicentre randomised controlled study planning to use a power of 65 W. 8 Doctors working with EVLA report the energy used for ablation as the Linear Endovenous Energy Density (LEED) 9 the energy used per cm of vein treated.However, the wide variation of power, pullback and intermittent ablation and rest cycle combinations reported by those using EMWA have not included the LEED, 4-8, making comparing devices and reported outcomes challenging.
For those wishing to use EMWA to emulate their EVLA practice, there is no independent comparison between the 2 methods to suggest which EMWA parameters will give the same ablation results as the EVLA parameters they currently use.
This study aims to use the validated porcine liver model to define which parameters for EMWA emulate the effects of the most commonly used parameters for EVLA.

Methods
1][12][13] The previously described protocol for assessment of thermal spread and the blinding of the observers using this model was followed. 14he fresh porcine liver was acquired from a local butcher and allowed to reach room temperature (21°C) before the experiment commenced.First, the liver was placed on a plastic sheet and covered with .9%normal saline solution.Next, the laser fibre or microwave antenna was placed on the liver's serosal surface and covered with heat-resistant glass (30 × 10 cm).Clamps held the glass with sufficient pressure to push the device into the liver, always ensuring contact with the porcine liver serosa on 3 sides.This pressure is essential for observation of the lateral thermal spread through the glass.Finally, we removed any air bubbles by irrigation with .9%normal saline using a plastic cannula and syringe.
A high-resolution digital camera (Nikon D90) was positioned on a tripod to take digital images.Each image also included a ruler placed on the glass and a code number representing the parameters used.Including these allowed for accurate measurements to be made later from the digital image and assessment of the ablation tract by blinded observers, unaware of the parameters used.
The EVLA system selected for the comparison was the Vari-Lase 1470 nm diode laser (Teleflex, Wayne, Pennsylvania, United States) with a 600-micron radial fibre.Five ablation tracts, each 3 cm long, were made at 6 W, 8 W and 10 W using pullback speeds of 6, 7, 8 and 9 cm/s.These combinations gave ablation tracts in the porcine liver model correlating to LEEDs between 36 J/cm and 90 J/cm at the 3 different powers tested.An experienced venous surgeon (MSW) performed the treatments as a continuous pullback.
We then performed a preliminary trial using the ECO-100E2 catheter (ECO, Nanjing, Jiangsu, P. R. China) using powers ranging from 30-80 W and various pullback speeds (4-9 s/cm, to identify adequate ablation tracts which had the potential to correlate with those formed using EVLA. We identified that using pullback times of 4-9 s/cm (corresponding closely to those used with EVLA), the ablation tracts that appeared most similar to those produced by EVLA were made using EMWA power settings between 35-75 W. Using this information, we then performed 5 ablations in the porcine liver model, each 3 cm long, with EMWA at each combination of powers 35, 40, 45, 50, 55, 60, 65, 70 and 75 W and with pullback speeds of 4, 5, 6, 7, 8 and 9 cm/s.We performed the treatments using an interrupted pullback technique.The catheter was held still during the heating phase, and when the heating cycle had finished, the device was withdrawn 1 cm and the process repeated.
The digital images for the EVLA and EMWA were cropped, straightened, and anonymised with a unique code for each image for blinded analysis by 2 observers.These observers had not been involved in performing the ablations or the image preparation.
The thermal spread was assessed on each digital image using SketchAndCalc™ (v: 6.2.4) (https://www.sketchandcalc.com/).As porcine liver changes colour from a deep purple/brown to light beige when the protein is denatured by heat, the observers were able to assess the lateral extent of the thermal ablation for each treatmentthe thermal spread.The 2 blinded observers made 5 measurements per image -maximum lateral spread, minimum lateral spread and 3 random lateral spreads.The observers were instructed not to make any measurements within 3 mm of either end of the ablation tracts.This restriction avoided the inclusion of any artefacts due to the beginning or end of the ablation.Interobserver agreement was assessed using the Intraclass Correlation Coefficient (ICC). 15They also recorded any carbonisation using the scale noted in Table 1.
Each observer made measurements of the total lateral spread of the ablated tissue in mm to 2 decimal places (.01 mm).As mentioned above, there were 5 measurements for each image.We combined the results from each observer, resulting in ten measurements of lateral thermal spread per image.Results from each of the EVLA parameters were compared to those from each EMWA parameters, using a oneway ANOVA with Tukey's multiple comparison tests (GraphPad Prism 9 -https://www.graphpad.com/scientificsoftware/prism/).
Once we had identified those EMWA ablation tracts that produced equivalent lateral thermal spreads to the EVLA tracts, we analysed carbonisation.The blinded observers graded each ablation tract using the scale in Table 1.Ablation tracts with a mean carbonisation score of 2.5 or more were considered to show excess carbonisation.

Interobserver Agreement
As the ablation tracts were not uniform, the observers measured and recorded the maximum and minimum thermal spread, and then 3 random measurements along the tract away from these 2 points.In order to compare the 2 observers, it was necessary to compare their results of fixed points.The only fixed points for each image were the observers assessment and measurement of the maximum and minimum thermal spread.Hence ICC was performed on each of these for both the EMWA and EVLA ablation tracts.For maximum and minimum thermal spread in EMWA tracts, ICC showed an agreement of .99 and .96respectively and for EVLA, .98 and .98.These results show "excellent reliability" of interobserver measurements. 15owever, we also assessed the ICC for the 3 areas picked at random by each observer, to assess whether these heterogeneously selected results were also in agreement.The ICC for EMWA for these measurements was .85 and for EVLA was .88.These results show "good reliability" of interobserver measurements. 15

EMWA v EVLA Thermal Spread
The results are tabulated in Table 2.
For each of the selected EVLA parameters studied, at least 1 EMWA combination of parameters produced very similar results.For example, to achieve similar thermal effects as EVLA with a LEED of 48 J/cm at 8 W power, the same lateral thermal spread can be achieved by EMWA using: However, we found excessive carbonisation of 2 of these parameters, leaving either 35 W at 7 s/cm or 40 W at 6 s/cm as optimal (Figure 1).However, It should be noted that the excessive carbonisation was in the EVLA and not the EMWA tract.
Similarly, to achieve an equivalent effect as obtained with EVLAwith a LEED of 64 J/cm at a power of 8 W, we have found EMWA with the following parameters to produce similar results: • 75 W 4 s/cm A higher power with a faster pullback -55 W at 5 s/cm (Figure 2) -or even 60 W ant 6 s/cm or 75 W at 4 s/cm -would be the most appealing, although all of these produced the same results.Again, although excessive carbonisation was found in the EVLA tract at 40 W and 8 s/cm pullback, it was not found in the EMWA tract, making it a viable, but slower, alternative.
We thought there might be a simple conversion from the LEED obtained with EVLA and that produced with EMWA, which would make selecting the optimal parameters easy.However, we found that the mean EMWA LEEDs varied their relationship with EVLA LEEDs depending on the speed of pullback of the EVLA device (Figure 3).At all 3 EVLA powers tested, the EMWA LEED needed to attain the same result was higher in proportion with the quicker pullbacks and lower with the slower ones.However, we could not find a trend with the increasing EVLA power.

Discussion
EVTA requires transmural death of the vein wall for successful long-term ablation by fibrosis. 16,17ublished research in this area has previously used explanted human great saphenous vein embedded in pork meat, followed by the treatment of fore limb veins and hind limb arteries in a live pig model. 18The authors then extrapolated the results from these 2 experiments to suggest the optimal settings to use in human great saphenous vein. 18However, using this approach, the authors only achieved a 65% occlusion rate at 1 year, although they claimed that 85% had reflux under .5 s.In addition, the use of a live animal model breaks the reduction criteria of the "3 Rs" principle of animal experimentation, when a viable alternative exists. 19ence, the porcine liver model, with its ability to measure thermal spread from an EVTA device, has proven very useful in assessing such devices, both RFA and EVLA, 10,13,14 and has been shown to correlate well both histologically with exvivo great saphenous vein 11 as well as with clinical resultsleading to an occlusion rate of 100% at 1 year. 12owever, the aim of this study is not to dictate optimal settings for EMWA, but to use the porcine liver model to show which settings of EMWA produce an equivalent effect as recognised power and pull-back combinations for EVLA.We have not suggested that the precise measurement of thermal spread in porcine liver is the same as might be measured in the human saphenous vein.However, as we have previously shown that there is a good correlation between the thermal spread in the porcine liver model and depth of ablation in the ex-vivo human vein, 11 and shown this to be relevant clinically, 12 then the use of the porcine liver model to look for equivalence of action between 2 EVTA device appears to be valid.The rationale for this study was that the original microwave generator that we were using gave very good thermal spread at 35 W and 40 W, leading us to recommend this power initially in our clinic.However, following a service of the machine, we found that this power was inadequate to close the vein and powers in the region of 60 W were needed.On enquiry of the service company, we were informed that the machine had initially been supplied with the set-up used for liver tumour ablation using a different probe.At service, the machine was re-programmed with the settings for venous treatments using the EMWA catheter.Therefore we needed to re-calibrate our recommended setting against those that we commonly used when utilising the EVLA device.
Hence, we have produced a table of parameters that doctors might refer to when using the ECO EMWA device to attain an equivalent treatment they would achieve with their preferred setting with the Vari-lase 1470 nm radial fibre.
Although this should be useful clinically, it does raise some interesting issues.
The fact that the power levels of EMWA are so different from EVLA to get the same thermal effect in the porcine liver model suggests that the power displayed on the console is not the same as the energy emerging from the tip of the device.An alternative conclusion would be that the microwave field at the device's tip causes less heat generation for the same power, but this is less likely.
In the EU, regulations allow laser consoles to have an error of +/À 20% from the displayed power. 20,21Furthermore, the power displayed on laser consoles is a measurement of the electrical input into the diode that produces the laser, not its output.Hence there is a loss of energy from several points between the electrical input into the diode and the emission of laser energy from the fibre tip, including the diode itself (especially if cold), 22 optical couplings, and from the fibre  itself -lateral energy loss, impurities in the fibre material and back reflection from the tip. 23uch differences might explain why 1 meta-analysis concluded that "commonly used parameters" are not important in EVLA. 24However, the authors used papers with "acceptable" ablation rates of around 92% and did not consider that the papers used reported outcomes from different EVLA systems.One would reasonably expect far better outcomes if optimal parameters were generated for each different EVLA system and modified for the size of the vein treated -both the diameter and the vein wall (ie: the mass of tissue to be ablated).
The relationship between the power displayed on the microwave console and the energy emerging from the EMWA device tip is less well studied.However, it is likely that the power displayed on the console is a measurement of the electrical input into the system.Losses will occur before the energy reaches the microwave antenna.Indeed, it is noticeable that during use, the endovenous catheter for the ECO microwave system heats up significantly, indicating power loss along the catheter.
5][6][7][8] It is quite possible that the power displayed and the amount of energy emerging from the device tip might vary between the manufacturers.This might explain why such a wide range of powers have been reported for treating varicose veins with EMWA.
There are several limitations of this study.The first and most obvious is that we have used an in vitro model.Even though the porcine liver model has been used previously and validated clinically, conclusions derived from an in-vitro study still need to be verified in clinical practice.Of course, as this model has already been validated against ex-vivo GSV treatment and histology, 11 and clinical outcomes at 1 year, 12 such verification would only be of interest if it showed that there was a discrepancy in the clinical outcomes of EMWA and EVLA in similarly sized veins and using the setting that are matched as equivalent in this study.
Secondly, we have only used 1 EMWA system and 1 EVLA system.As such, the results apply to those systems, but must be checked if other EMWA or EVLA systems are used.
Thirdly, the interrupted pullback method we used for the EMWA resulted in a less homogeneous tract, giving noticeable differences between the maximum and minimum lateral thermal spreads.Therefore, a continuous pullback might give slightly different results.However, using a continuous pullback method introduces another variable into the system, making it easier for inexperienced doctors to use the interrupted technique.
In conclusion, we have used the validated porcine liver model to define parameters (power and pullback speed) that can be used with EMWA, to emulate the tissue ablation found when using common parameters used for EVLA.These need to be validated in clinical practice, but as the model used has already been validated against clinical outcome in endovenous thermal ablation using bipolar radiofrequency, there is little reason to suspect that the results would not be valid.We have confirmed that EMWA requires a higher power than EVLA to get the same thermal effect in porcine liver tissue and found that the LEED for EMWA is approximately 4-6 times higher than that for EVLA.However, the difference between the LEEDs for EMWA and EVLA is not constant.It seems to be related to the speed of pullback of the EVLA rather than the power.

Figure 3 .
Figure 3.A graph showing how the EMWA LEED varies against the EVLA LEED and how this decreases with longer EVLA pullback time across all EVLA powers studied.The X-axis is the EVLA pullback in s/cm, and the Y-axis is the multiple of EVLA LEED found for the equivalent EMWA LEED (see text).

Table 1 .
The Carbonisation Scale Used to Assess Carbonisation in the Ablation Tracts.

Table 2 .
This Table Shows the EMWA Parameters That Have Equivalent Ablation Tracts to the EVLA Parameters Studied.