**4. Adaptive proton therapy of lung cancers**

#### **4.1. Rationale**

Adaptive radiation therapy (ART) is a closed‐loop process where the treatment plan will be re‐optimized for treatment variations such as patient anatomy change using a systematic feedback of measurements. [13]. Thanks to the advancement of imaging modalities available for treatment planning and delivery, such as 4D CT and onboard imaging, ART has been feasible and clinically implemented at many cancer centers. The main goal of plan adaptation is to adjust the treatment plan to the change of patient anatomy, tumor motion, or setup, so that the target coverage and normal tissue sparing remain optimal for each individual patient during the whole course of treatment. For lung cancer patients, anatomy change is often inevitable due to tumor regression, pleural and pericardial effusions, or atelectasis. Adaptive photon therapy has been shown to be beneficial in lung cancer treatment, resulted in a mean reduction of 21% for the volume of ipsilateral lung receiving 20 Gy (V20) [14], and an average of 65 cGy reduction in mean lung dose and reductions in cord max dose, mean esophageal dose, and heart dose [15]. It was reported that ART has the potential to improve the accuracy of radiation treatments, thus reducing the exposure of organs at risk and facilitating safe dose escalation, leading to potentially better local control and overall survival [16–19].

Because a proton beam has a finite range and sharp distal dose fall off, the dose distribution of a proton plan is very sensitive to anatomy change; therefore, the need for lung cancer treatment adaption in proton therapy is even greater than photon therapy. Hui *et al.* found that the effects of inter‐fractional motion and anatomic change could lead to a result of up to 8% reduction of the CTV coverage, a mean 4% dose increase of the volume of the contralateral lung receiving at least 5 CGE, and a mean 4.4 CGE increase in spinal cord maximum dose [20]. Koey *et al.* reported that without adaptive planning, target coverage could be dropped to below 60% com‐ pared with adaptive planning for some lung cancer case undergoing proton therapy [21]. The potential considerable dose change in proton therapy due to anatomy variation indicates that plan adaptation is essential in proton therapy of lung cancer.

#### **4.2. Process for adaptive proton planning**

**4. Adaptive proton therapy of lung cancers**

Adaptive radiation therapy (ART) is a closed‐loop process where the treatment plan will be re‐optimized for treatment variations such as patient anatomy change using a systematic feedback of measurements. [13]. Thanks to the advancement of imaging modalities available for treatment planning and delivery, such as 4D CT and onboard imaging, ART has been feasible and clinically implemented at many cancer centers. The main goal of plan adaptation is to adjust the treatment plan to the change of patient anatomy, tumor motion, or setup, so that the target coverage and normal tissue sparing remain optimal for each individual patient during the whole course of treatment. For lung cancer patients, anatomy change is often inevitable due to tumor regression, pleural and pericardial effusions, or atelectasis. Adaptive photon therapy has been shown to be beneficial in lung cancer treatment, resulted in a mean reduction of 21% for the volume of ipsilateral lung receiving 20 Gy (V20) [14], and an average of 65 cGy reduction in mean lung dose and reductions in cord max dose, mean esophageal dose, and heart dose [15]. It was reported that ART has the potential to improve the accuracy of radiation treatments, thus reducing the exposure of organs at risk and facilitating safe dose

**Figure 3.** Dose comparison of Uniform Scanning proton plan and IMRT plan. (a) Proton plan, (b) IMRT plan, and (c) DVH comparison (solid line—proton, dashed line—IMRT). The prescribed dose was 74 cobalt gray equivalent (CGE) at

escalation, leading to potentially better local control and overall survival [16–19].

Because a proton beam has a finite range and sharp distal dose fall off, the dose distribution of a proton plan is very sensitive to anatomy change; therefore, the need for lung cancer treatment adaption in proton therapy is even greater than photon therapy. Hui *et al.* found that the effects

**4.1. Rationale**

50 Radiotherapy

2 CGE per fraction for 37 fractions.

A typical adaptive planning process includes measuring the treatment variations such as ana‐ tomic change, evaluating their dosimetric and clinical impact, and adapting the radiation treat‐ ment to the updated information as necessary. In proton therapy of lung cancer, anatomic change is of main concern. Repeated CT scans are commonly used to measure the anatomic change dur‐ ing the treatment course. Ideally, the repeated CT scans should be performed frequently with a 4D CT scan so that patient anatomy and motion can be accurately evaluated. However, depend‐ ing on facility resources and patient compliance, in room CBCT or slow CT scans can also be used. The repeated CT will be registered to the initial CT, and a QA plan will be generated by applying the same beam configuration from the initial plan to the registered repeated CT data, which will be evaluated on dosimetric change and potential clinical impact. The physicist and physician will then determine whether and how the plan will be adapted. If plan adaptation is determined necessary, plan change will be made according to the physician/physicist instruc‐ tion, and the new plan will be changed and go through the process of plan review, QA, and approval before beam delivery similar to the initial plan. In addition to deciding whether a plan adaptation is needed, one should also decide whether any other change is needed for the patient. For example, if the patient anatomy is likely to change significantly before the next scheduled CT scan, we may want to increase the imaging frequency for the patient.

A clinical workflow of the adaptive planning for lung cancer treatment at our center is shown in **Figure 4**. After initial 4D CT imaging, treat planning, and beam delivery, QA CT (i.e., repetitive CT) will be performed after a patient receives 14, 30, and 50 CGE of proton dose, that is, after the 7th, 15th, and 25th fraction for most patients treated with 2 CGE per fraction. The repeated aver‐ age CT was fused to the original average CT based on bony anatomy by a dosimetrist using the VelocityAI software system (Version 3.1.0, Varian Medical Systems, Palo Alto, CA). A quality assurance (QA) plan was generated after each CT scan by applying the same proton beams and hardware (apertures and compensators) in the original plan to the registered new CT dataset using the XiO TPS (Version 5.0, Elekta Inc., St. Louis, MO). A physicist will first review the CT fusion to evaluate the anatomic change and check the correctness of the fusion. The physicist will then review the QA plan to evaluate the dosimetric change and the correlation between the dosimetric change and the anatomy variation. Together with the attending physician, the physicist will make a recommendation on whether plan adaption is needed. If plan adaptation is determined to be necessary, a dosimetrist will make the plan change, and treatment with the new plan will start as soon as possible. The process of treatment, QA CT, QA planning, and plan adaptation will be repeated until the patient complete the treatment course.

**Figure 4.** A clinical workflow of adaptive planning for lung cancer treatment using uniform scanning proton therapy.

#### **4.3. Strategies for plan adaptation**

One straightforward way of plan adaptation is to re‐plan based on the newly obtained CT, repeating the same process as how the initial plan is created. Re‐planning has been used for most adaptive treatment in both photon therapy and proton therapy and generally includes target contouring, beam placement, dose optimization, plan review and approval, documen‐ tation and billing, calendar adjustment, QA, and so on. For adaptive proton therapy where PSPT and USPT are used, new patient specific devices such as apertures and compensators are also needed during re‐planning, which can lead to added cost and long turnaround time due to the manufacturing process. Substantial effort is needed from the dosimetrists, physi‐ cist, and machinists, and it can take several days to make the new plan available for treatment. Before the new plan becomes available, the patient can either continue to be treated with the initial plan or have a treatment break, depending on the extent of anatomy change and its impact on dose distribution and potential clinical effect. On the other hand, re‐planning can fully adapt a plan and achieve the best optimization of dose distribution based on the new CT data set. **Figure 5** shows an example of re‐planning with new patient specific hardware. Substantial tumor shrinkage was observed on the repeated CT scan, which led to a large increase in lung and cord dose (**Figure 5b**). A new plan was created based on the new 4D CT (**Figure 5c**), with an improved normal tissue sparing while maintaining target coverage similar to the initial plan.

Another way of plan adaptation is to make some simple changes in beam parameters, such as range, modulation, or beam weight of any combination. Because a uniform scanning or passive scattering proton beam delivers a uniform dose to patients, it is possible to adjust the range and/or modulation for a proton beam to shift the depth of the spread out Bragg peak (SOBP) region so that the adjusted beam would conform to the target after the water equivalent thickness (WET) changes due to anatomy change. For uniform scanning proton beams, such parameter change is very easy and can be made for the TPS and R&V in minutes plus some additional work on documentation. **Figure 6** shows an example of such case that patient developed pleural effusion at the 25th fraction. After simply increasing the proton Adaptive Radiotherapy for Lung Cancer Using Uniform Scanning Proton Beams http://dx.doi.org/10.5772/67445 53

**4.3. Strategies for plan adaptation**

52 Radiotherapy

similar to the initial plan.

One straightforward way of plan adaptation is to re‐plan based on the newly obtained CT, repeating the same process as how the initial plan is created. Re‐planning has been used for most adaptive treatment in both photon therapy and proton therapy and generally includes target contouring, beam placement, dose optimization, plan review and approval, documen‐ tation and billing, calendar adjustment, QA, and so on. For adaptive proton therapy where PSPT and USPT are used, new patient specific devices such as apertures and compensators are also needed during re‐planning, which can lead to added cost and long turnaround time due to the manufacturing process. Substantial effort is needed from the dosimetrists, physi‐ cist, and machinists, and it can take several days to make the new plan available for treatment. Before the new plan becomes available, the patient can either continue to be treated with the initial plan or have a treatment break, depending on the extent of anatomy change and its impact on dose distribution and potential clinical effect. On the other hand, re‐planning can fully adapt a plan and achieve the best optimization of dose distribution based on the new CT data set. **Figure 5** shows an example of re‐planning with new patient specific hardware. Substantial tumor shrinkage was observed on the repeated CT scan, which led to a large increase in lung and cord dose (**Figure 5b**). A new plan was created based on the new 4D CT (**Figure 5c**), with an improved normal tissue sparing while maintaining target coverage

**Figure 4.** A clinical workflow of adaptive planning for lung cancer treatment using uniform scanning proton therapy.

Another way of plan adaptation is to make some simple changes in beam parameters, such as range, modulation, or beam weight of any combination. Because a uniform scanning or passive scattering proton beam delivers a uniform dose to patients, it is possible to adjust the range and/or modulation for a proton beam to shift the depth of the spread out Bragg peak (SOBP) region so that the adjusted beam would conform to the target after the water equivalent thickness (WET) changes due to anatomy change. For uniform scanning proton beams, such parameter change is very easy and can be made for the TPS and R&V in minutes plus some additional work on documentation. **Figure 6** shows an example of such case that patient developed pleural effusion at the 25th fraction. After simply increasing the proton

**Figure 5.** An example case of re‐planning with new patient specific device. The patient has small cell Stage IIIA lung cancer with COPD. A 66 CGE was delivered at 33 fractions using uniform scanning proton beams. (a) Original plan; (b) QA plan; (c) adapted plan based on the new CT data.

range by 2.2 cm, the target became fully covered while the normal tissues of lung and heart were still well protected. This simple approach can be highly desirable for certain anatomy changes such as patient weight change which pulls back or increase the range relatively uni‐ form, and/or a quick plan adaptation is needed due to concern on treatment breaks. Please note that such approach is unique in uniform scanning and may not be available in PBS or passive scattering PT.

Other strategies of plan change for USPT could be beam weight change, for example, decrease the weight of beam(s) that is adversely affected by the anatomy change, and increase the weight of beam(s) that is least affected. In addition, a hybrid approach, such as re‐planning for one beam and range adjustment for another, can also be used as appropriate.

The strategy used for plan adaptation depends largely on the institutional practice and the beam delivery technique used for lung cancer treatment. For lung cancer treatment with PSPT, Koay *et al.* reported that 20.5% of patients underwent adaptive planning using re‐planning with new patient‐specific hardware [21]. For USPT, Zheng *et al.* reported that 18.8% of lung cancer patients underwent adaptive planning, using various strategies including range change only (10.9%), range and modulation change (1.8%), range, modulation, and beam weight change (1.2%), and re‐planning with new hardware (5.5%) [22]. For PBS or IMPT, Chang *et al.* reported that 26.5% patients were re‐planed [6]. A brief summary of adaptive proton therapy literature discussed here is listed in **Table 2**.

**Figure 6.** The dose distribution from the right posterior oblique beam normalized at the isocenter for a lung cancer patient undergoing adaptive proton therapy using parameter adjustment. (a) Initial plan; (b) QA plan; (c) adapted plan with a 2.2 cm range increase. The patient had a right hilar mass and was treated with three proton beams for a total dose of 74 CGE. Fluid buildup was observed on a repeated CT scan after the 25th fraction.


Abbreviations: MDACC: M. D. Anderson Cancer Center; STPTC: Southern Tohoku Proton Therapy Center; UFTPI: University of Florida Proton Therapy Institute; NCCHE: National Cancer Center Hospital East; UHE: University Hospital Essen.

Note: Reports from the literature.

**Table 2.** Adaptive proton therapy.

#### **5. Practical considerations**

While adaptive planning can potentially improve the dose distribution and clinical outcome, there are also many pitfalls and limitations in the current adaptive planning process. An opti‐ mal adaptive planning process should be developed based on both practical considerations and theoretical dosimetric and clinical gains.

#### **5.1. Frequency of repeated CT scanning and QA planning**

The frequency of repeated CT scans depends on facility‐specific protocol or individual patient need. Chang reported that 4D CT scans were repeated during week 3 or 4 of treatment or as clinically indicated by the treating physician for lung cancer patients undergoing PSPT [23], and weekly or every 2 to 3 weeks for those receiving intensity modulated proton therapy at MD Anderson Cancer Center [6]. At our center, 4D CT is generally repeated after 7, 15, and 25 fractions of treatment. However, for special cases, more repeated CT scans may be needed such as when patients have a pleural effusion or large weight change. In addition, if hypo‐fraction‐ ated or hyper‐fractionated treatment is used, more frequent monitoring should be considered. Daily imaging has becoming available with the introduction of in room CT like CBCT into proton therapy; however, its clinical implementation may be limited due to the extra treatment time and human effort as well as concerns on the increased imaging dose to patients.

#### **5.2. Limitations of image registration and QA planning**

One key component of adaptive planning is the image registration. Accurate imaging regis‐ tration can be challenging, especially for lung adaptive planning where considerable anatomy change may be observed due to disease progression, tumor response to therapy and respira‐ tory motion [24, 25]. It is important to setup and immobilize the patient for repeated CT as close as possible to the initial CT scan that is used for the treatment planning as large patient setup variation could lead to difficulty on image registration and anatomy change evaluation. The accuracy of image registration needs to be carefully evaluated. In addition, there can be limitation on how a QA plan is generated. For example, our treatment plan system does not account for the patient pitch and roll when a QA plan is applied to the new CT data, although our image registration software does. Another issue is that there could be human errors asso‐ ciated with the image registration and QA plan process, such as beams may be placed with an incorrect isocenter in a QA plan. Limitations or errors in the image registration and QA plan process could result in artificial dose deviation unrelated to anatomy change and poten‐ tial errors in decision‐making of plan adaptation. Therefore, it is critical to understand these limitations and evaluate the accuracy of image registration and QA planning to avoid errors in decision‐making that may lead to unnecessary plan change and potential mistreatment. Our guideline is, in addition to review the QA plan and dose distribution beam by beam, we also analyze the anatomy change and the correlation between the dose change and anatomy change. Any noticeable dose change in the QA plan should be correlated to either patient anatomy/motion change or setup variation; otherwise, the dose change may be artificial as a result of software limitations or human errors, and further investigation should be warranted.

#### **5.3. Correlation between dose change and anatomy variation**

**5. Practical considerations**

Note: Reports from the literature.

**Table 2.** Adaptive proton therapy.

Hospital Essen.

54 Radiotherapy

**References Year Institution Treatment** 

**technique**

of 74 CGE. Fluid buildup was observed on a repeated CT scan after the 25th fraction.

Zheng *et al.* [22] 2015 ProCure USPT 165 18.8% After 7, 15, and

**No. of patients**

**Figure 6.** The dose distribution from the right posterior oblique beam normalized at the isocenter for a lung cancer patient undergoing adaptive proton therapy using parameter adjustment. (a) Initial plan; (b) QA plan; (c) adapted plan with a 2.2 cm range increase. The patient had a right hilar mass and was treated with three proton beams for a total dose

Koay *et al.* [21] 2012 MDACC PSPT 44 20.5% At week 3 or 4 At 24 fractions Chang *et al*. [6] 2014 MDACC PBS/IMPT 34 26.5% Every 2 weeks After 10

Abbreviations: MDACC: M. D. Anderson Cancer Center; STPTC: Southern Tohoku Proton Therapy Center; UFTPI: University of Florida Proton Therapy Institute; NCCHE: National Cancer Center Hospital East; UHE: University

**Adaptation percentage**

**Repeated CT scanning**

25 fractions

**Median time for plan change**

fractions

After 18 fractions

and theoretical dosimetric and clinical gains.

While adaptive planning can potentially improve the dose distribution and clinical outcome, there are also many pitfalls and limitations in the current adaptive planning process. An opti‐ mal adaptive planning process should be developed based on both practical considerations The penetration depth of a proton beam is a function of the proton energy and the WET of the materials it passes through. Therefore, for a proton beam of given energy, the depth of the dose falloff is directly correlated to the WET associated with the anatomy in the beam path. Common changes in anatomy that could lead to plan adaptation include patient weight gain or loss, tumor shrinkage or growth, pleural effusion, atelectasis, and so on. For example, when a patient gains weight, the WET in beam path will increase, leading to a range pull back. The effect of patient weight change typically is more noticeable for anterior proton beams, and may be addressed by simply adjusting the range and/or modulation as the WET change is relatively uniform within the field. Similarly, tumor shrinkage will result in a decrease in WET and beam overshoot, which could lead to more dose to normal tissues such as lung and cord. Target coverage is generally not an issue when a tumor shrinks but can be severely compro‐ mised when a tumor progresses and increases in volume. Tumor shrinkage or progression will have an effect on the dose distribution from all beams and is likely to lead a re‐planning if the tumor volume change is considerable. About half of cancer patients develop a plural effusion, which is a buildup of extra fluid in space between lung and chest cavity. Clearly, any change in pleural effusion would lead to change in WET and dose deviation from the beam passing through the fluid buildup. If the tumor is far away from the fluid buildup and no beam passes through it, the effect of pleural effusion could be negligible on the dose distribution and no plan adaptation is needed. In most cases, one would only need to make adjustments for the beam(s) that passes the fluid, by either changing the range and/or modulation or re‐planning the beam with a new compensator.

Please note that for the composite dose distribution from several proton beams, the correlation to the anatomy may not be straightforward. Anatomy and WET changes will lead to visible dose change for one beam; their effect may not show up well on the overall dose distribution and the DVH. For example, the volume of the target receiving at least 95% of the prescription dose (V95) may show minimal change, while there is a clear under coverage due to a range pullback from a certain beam and plan adaptation should be used. Therefore, beam‐by‐beam analysis is strongly recommended to evaluate the dose correlation due to the anatomy change.

#### **5.4. Patient motion and motion management**

Given the sensitivity of proton beam to anatomy variation, accurate evaluation and appropriate management of motion are very important in lung cancer therapy. For PSPT and USPT, the patient motion is typically accounted for during the initial treatment planning using tech‐ niques such as target expansion (ITV), range smearing, and stopping power ratio override [7, 26]. In addition, motion can be managed using respiratory gated system [27]. From our experience, the effect of motion variation in the QA plan based on repeated 4D CT scan seems to be relatively low, and the original plan is typically robust enough to adequately cover the target as long as no anatomy change is present. For PBS, the interplay of patient motion and dynamic beam delivery could result in dose heterogeneity in target and potential under cov‐ erage. To mitigate the interplay effect, the motion magnitude for patients treated with PBS is often restricted, such as at a maximum of 5 mm. In addition, several techniques have been used or proposed to mitigate the interplay effect, such as layer repainting, large beam spot, respiratory gated beam delivery, robust planning optimization accounting for the motion, and tumor tracking [6, 12, 28–31]. It has also been reported that the interplay effect may be aver‐ aged out during fractionated treatment [32]. However, to fully achieve the potential of IMPT, it may be necessary to routinely evaluate motion change and adapt treatment accordingly.

#### **5.5. Resource constraints and potential risk associated with plan change**

When re‐planning is used in plan adaptation, new patient specific apertures and compen‐ sators may need to be manufactured for both PSPT and USPT. The manufacturing process usually takes hours or more to complete, depending on field size and shape as well as the queuing status of other hardware. If no machine shop is available onsite, the hardware needs to be manufactured by other contracting companies which may take 1–2 days to become available. Furthermore, additional time is needed for the following QA process for the hard‐ ware and output measurement. While no hardware is needed for PBS, the robust treatment planning and optimization and the consequent QA process can be very time and effort con‐ suming. In addition, the plan change can lead to unexpected consequences and increased risk of treatment errors, especially when it is not communicated well. Therefore, we have to take the associated cost and risk into account in addition to the dosimetric and clinical gain when deciding whether plan change is necessary.

#### **5.6. Treatment volume with tumor shrinkage**

It is still unclear on whether the clinical target volume should be reduced accordingly when a tumor shrinks during the treatment course. Siker *et al.* cautioned field reductions for tumor shrinkage during radiotherapy, questioning the significance of tumor regression because his‐ tologic tumor clearance was hard to document [33]. However, Guckenberger *et al.* believed that adaptation of radiotherapy to the shrinking GTV did not compromise the dose coverage of volumes of subclinical microscopic disease [34]. In adaptive proton therapy for both USPT and PSPT, the treatment target volume is commonly kept the same as the initial plan and the same apertures are used, while the beam penetration is adjusted, that is, the range is adjusted or the compensator is recalculated, to account for the WET change associated with the tumor shrinkage. Exceptions can be made per physicians' discretion for cases that normal tissue sparing is critical, such as for patients with a very large initial tumor volume and normal dose can be close or exceed the tolerance with the initial plan. One proposal is to treat the initial target volume for at least 50 Gy, the standard dose for microscopic disease, and then treat the reduced volume to the full dose with a boost [35].

#### **5.7. Dose accumulation**

a patient gains weight, the WET in beam path will increase, leading to a range pull back. The effect of patient weight change typically is more noticeable for anterior proton beams, and may be addressed by simply adjusting the range and/or modulation as the WET change is relatively uniform within the field. Similarly, tumor shrinkage will result in a decrease in WET and beam overshoot, which could lead to more dose to normal tissues such as lung and cord. Target coverage is generally not an issue when a tumor shrinks but can be severely compro‐ mised when a tumor progresses and increases in volume. Tumor shrinkage or progression will have an effect on the dose distribution from all beams and is likely to lead a re‐planning if the tumor volume change is considerable. About half of cancer patients develop a plural effusion, which is a buildup of extra fluid in space between lung and chest cavity. Clearly, any change in pleural effusion would lead to change in WET and dose deviation from the beam passing through the fluid buildup. If the tumor is far away from the fluid buildup and no beam passes through it, the effect of pleural effusion could be negligible on the dose distribution and no plan adaptation is needed. In most cases, one would only need to make adjustments for the beam(s) that passes the fluid, by either changing the range and/or modulation or re‐planning

Please note that for the composite dose distribution from several proton beams, the correlation to the anatomy may not be straightforward. Anatomy and WET changes will lead to visible dose change for one beam; their effect may not show up well on the overall dose distribution and the DVH. For example, the volume of the target receiving at least 95% of the prescription dose (V95) may show minimal change, while there is a clear under coverage due to a range pullback from a certain beam and plan adaptation should be used. Therefore, beam‐by‐beam analysis is strongly recommended to evaluate the dose correlation due to the anatomy change.

Given the sensitivity of proton beam to anatomy variation, accurate evaluation and appropriate management of motion are very important in lung cancer therapy. For PSPT and USPT, the patient motion is typically accounted for during the initial treatment planning using tech‐ niques such as target expansion (ITV), range smearing, and stopping power ratio override [7, 26]. In addition, motion can be managed using respiratory gated system [27]. From our experience, the effect of motion variation in the QA plan based on repeated 4D CT scan seems to be relatively low, and the original plan is typically robust enough to adequately cover the target as long as no anatomy change is present. For PBS, the interplay of patient motion and dynamic beam delivery could result in dose heterogeneity in target and potential under cov‐ erage. To mitigate the interplay effect, the motion magnitude for patients treated with PBS is often restricted, such as at a maximum of 5 mm. In addition, several techniques have been used or proposed to mitigate the interplay effect, such as layer repainting, large beam spot, respiratory gated beam delivery, robust planning optimization accounting for the motion, and tumor tracking [6, 12, 28–31]. It has also been reported that the interplay effect may be aver‐ aged out during fractionated treatment [32]. However, to fully achieve the potential of IMPT, it may be necessary to routinely evaluate motion change and adapt treatment accordingly.

the beam with a new compensator.

56 Radiotherapy

**5.4. Patient motion and motion management**

Accurate accounting doses at the presence of anatomy change and plan adaptation is important to make informed decision on whether and how to adapt a plan. However, this can be chal‐ lenging due to limitation of image registration when large anatomic change or setup variation exist. In addition, CT scans are often repeated on a non‐daily basis, and the exact patient anat‐ omy between CT scans is unknown. To estimate the actual dose delivered between two image scans when daily patient anatomy information is not available, one may use a weighted sum‐ mation of the doses calculated on the two CT data sets, or interpolate patient anatomy between the two scans and calculate doses based on the interpolated data sets. The latter can be more realistic, but a good software tool for interpolation is needed.

#### **5.8. Criteria on plan change**

The criteria on when and how to adapt a plan can differ from institution to institution and depend on the attending physician and/or individual patient. There are several consider‐ ations during QA plan evaluations including: (1) Is there noticeable anatomic change? How will the anatomy change affect the dose? (2) How much does the PTV coverage change compared to the initial plan? Is the target coverage still acceptable? (3) How much does the normal tissue dose change? Is the normal tissue dose within tolerance? (4) How much is the dose deviation from the original plan? Will a re‐planning improve the dose distribution significantly? (5) How long does it take to have the revised plan ready for treatment? Will a treatment break be needed before the new plan becomes available? (6) How much are the cost and effort for a plan adaptation (e.g., whether new hardware fabrication was involved, or just some parameter change)? How many fractions are left? Is it worthwhile to make a plan change for the remainder of treatment? (8) Are there any special consideration for the patients, for example, does the patient need more sparing in lung due to pre‐existing lung function such as COPD?

Change *et al.* reported that the main criteria for plan adaptation was whether CTV or GTV receives <95% of dose and whether doses for normal tissues such as heart and cord dose were out of tolerance [6]. At our center, in addition to looking into dosimetric effect such as the tar‐ get coverage and normal tissue dose, we take into account the potential clinical gain as well as the cost and time associated with plan adaptation to decide on whether and how to adapt a plan. For example, if the patient is close to the end of treatment and the clinical impact of plan adaptation is low, we may use a simple adaptation strategy like range adjustment or no adaptation at all for the rest of treatment.
