Our current standard protocol for pre-treatment PSQA of TMI/TMLI patient treatment on Radixact HT include verification of absolute dose in five locations corresponding to brain, chest, pelvis, upper leg and lower leg using calibrated ionization chamber inserted in Cheese phantom and 2D fluence verification only for the chest target using ArcCHECK helical detector array. ArcCHECK allows measurement of treatment field length ≤20 cm in a single irradiation. Our protocol is in alignment with other studies wherein several authors have reported verification of absolute dose and fluence either section by section or in the junction of two consecutive arcs using various detectors and methods [7,11]. In addition, for each patient we also carried out in-vivo EBT3 film dosimetry a) at eleven pre-defined locations across the whole body and b) in the junction region of upper and lower body treatment plans to ensure delivery of homogeneous dose. The excellent agreement between planned and measured absolute dose (<±1.2%) and planar dose fluence (2Dγ>96%) in the five TMI/TMLI patients were within the internationally acceptable criteria. Off the many sub PTVs, we have limited the fluence verification only for the PTV chest due to logical and technical challenges prior to treatment and postulated that the results may be still applicable to other PTVs. We have chosen PTV chest because of its complicated shape with many OARs located as an island and hence represent the most complex intensity fluence. The probable shortcoming of our protocol is extensive time spend for the delivery QA, verification of limited dose fluence of a mega treatment volume, challenges in the in-vivo absolute dose verification in a highly modulated heterogeneous dose region and non-availability of on-line or prior-treatment verification results. Although the PSQA method reported by Takahashi [12] was carried out in a single irradiation, it requires a customized phantom with many slabs, three ion chambers and three films stitched one after the other to cover 100 cm length of the target. Besides extensive logistic requirements, it is very labor intensive, time consuming and co-registration of film measured and TPS calculated dose fluence is not straight forward and error prone.
The feasibility of using MVCT measured LOTS to reconstruct 3D dose distribution on to patient CT datasets have been reported by several authors either by using in-build or independent dose reconstruction algorithm in routine clinical cases [13–19]. The excellent agreement both in absolute dose (<±1.8%), 2Dγ (>97%), 3Dγ (>96%) and 2Dγ vs 3Dγ variation (≤2.5%) between reconstructed dose from LOTS as compared to planned, ion chamber and ArcCHECK measurement, in all ten non-TMI patients validated the accuracy and reliability of LOTS reconstructed method in Delivery Analysis against the standard methods. Our results are in agreement with previous publications [17–19]. Although MVCT measured LOTS reconstructed 3D dose distribution has been successfully implemented as an alternative PSQA method for regular clinical cases, its feasibility in TMI/TMLI has not been reported so far. The feasibility and validation of this method for TMI/TMLI is especially important as there is no direct approach and suitable detector or phantom to verify the delivery accuracy of this complex and highly modulated mega field.
In a big surprise to our retrospective investigation of LOTS reconstructed 3D dose distribution in five TMI/TMLI patients, 3Dγ to the majority of the PTVs in upper body plans were found much below the acceptable criteria of 95%. Amongst the four separate PTVs (Brain, Chest, Torso and limb), 3Dγ were slightly better for brain (>90%) except P12. The 3Dγ values (28%–94%) of PTV Chest of every patient were much lesser than corresponding 2Dγ (>96%) estimated from ArcCHECK measurement. For reasons not clear to us, we observed poor 3Dγ values from the analysis of the first patient (P11) itself. In an attempt to remove any possible error, we carried out a series of investigations including various email communication and data sharing with Accuray Medical Physics support team based in Europe. MVCT detector output was baselined again. As per the suggestions from Accuray medical physics support, the threshold of LOT was decreased from 0.7 to 0.5. Even after all these probable corrective measures also, the new 3D dose distribution reconstructed for the same patient (P11) from the newly measured LOTS resulted in no changes in the 3Dγ values of all PTVs. Despite this unsatisfactory result, we have continued the measurement of LOTS for the other four patients and reconstructed dose distributions were compared with the corresponding plans. Off the five patients, the upper body TMI/TMLI plan of P13, which was created using 5 cm field width, pitch of 0.3 and modulation factor of 3 showed the best agreement with 3Dγ values ≥93% in three separate PTVs. Although 3Dγ values were very poor for all PTVs and for every patient, the median deviation in D98% of all PTVs were within 2.5% except torso where the deviation is 3.78%. The deviation in D2% was relatively large for all PTVs and increase up to 9.48% for lower body PTV where 3Dγ were 100%. Overall, minimum PTV coverage (D98%) from the reconstruction method was within ±5% of corresponding plans except for PTV upper body where a reduction of up to 6.33% was observed. The reconstruction method increases hot spot (D2%) to all PTVs by up to 12.51% as compare to plan. The overall analysis results based on minimum and maximum dose to PTVs can still be considered acceptable, although not very satisfactory, based on the complexity of the target and treatment technique.
The selection of optimum field width, pitch and modulation factor determine both TMI/TMLI plan quality and treatment time. Hui et al [4] have investigated the effect of field width, modulation factor and pitch on the treatment plan outcome and delivery time. The authors have reported a reduction of treatment delivery time by 50% when the field width was increased from 2.5 cm to 5 cm in superior-inferior direction. For a 5 cm field width, earlier studies have recommended modulation factor and pitch ranging from 2.0 to 2.8 and 0.397 to 0.46 [4,7,8] respectively. The impact of HT planning parameters on the PSQA results especially with reconstruction from measured LOTS has not been reported in the literature. In all the lower body TMI/TMLI plans where the 3Dγ values were 100%, a field width of 5 cm, modulation factor from 2.15 to 2.5 and pitch from 0.4 to 0.41 were used, which is in agreement with the reported values [4,7,8]. However, in the upper body TMI/TMLI plans, modulation factor and pitch were customized from 2.49 to 3.5 and 0.3 to 0.43 to meet the set clinical goals. Although both modulation factor of 2.48 and 3.5 provides a poor 3Dγ values in patients P11 and P15, we observed a fairly better 3Dγ values of P13 plan created with modulation factor of 3 and pitch of 0.3. Subsequently, all the upper body TMI/TMLI plans including for patient P13 were re-optimized with 5 cm field width, pitch of 0.3 and modulation factor of 3. Similar to non-TMI planning protocol, we have ensured that the proportion of LOT bins up to 100 msec were less than 30% in TMI plan. We also have ensured that the maximum LOT bin does not exceed 30% of the total LOT bins to reduce “hockey stick effect”. Moreover, even after getting optimum dose distribution and MLC-LOT, we continue to run up to 1000 iteration, while simultaneously ensuring no change in the plan quality. The reconstructed dose from the new upper body plans thus created showed improvement both in 3Dγ and minimum target coverage and hot spot. As the 3Dγ was 100% for all plans and PTVs, we have tightened the evaluation criteria to 2%@2mm. Even at this stringent evaluation criteria also, almost all plans pass acceptance criteria of 95%. This also leads to improvement in the deviation in D98% and D2%.
Measured LOTS based reconstructed methods provide accurate and efficient verification of TMI/TMLI plan in a single irradiation. Reconstructed 3D dose calculation assumes that there is no change in the patient anatomy and tumor geometry. Moreover, the reconstruction method in Delivery Analysis does not explicitly check for differences between planned and delivered gantry angle, couch position, or treatment field position. Only variations in MLC-LOT are considered when calculating dose differences.