Dose as a function of liver volume and
planning target volume in helical
intensity-modulated radiation therapy–based stereotactic body
radiation therapy for hepatic metastasis
Stereotactic body radiation therapy (SBRT) has been shown to be an effective, well-tolerated treatment for local control of tumors metastatic to the liver. Multi-institutional Phase II trials are examining 60 Gy in 3 fractions delivered by linac-based, 3D-conformal IMRT. HiArt Helical TomoTherapy is a treatment unit that delivers co-planar helical IMRT that is capable of image-guided SBRT. We hypothesized that the maximum tolerable dose (MTD) delivered to a lesion by Helical TomoTherapy-based SBRT could be predicted based on the planning target volume (PTV) and liver volume.
Methods and Materials: To test this, we performed inverse treatment planning and analyzed the dosimetry for multiple hypothetical liver gross tumor volumes (GTV) with conventional PTV expansions. Inverse planning was carried out to find the maximum tolerated SBRT dose up to 60 Gy to be delivered in 3 fractions based on the dose constraint that 700 cc of normal liver would receive less than 15 Gy. Plans were run with the goal of delivering the MTD between 30 and 60 Gy while meeting the normal tissue constraints, as per the Colorado Multiple Institutional Review Board (COMIRB) Phase I protocol. For the liver, this constraint was 700 cc of normal liver to receive less than 15 Gy. For the heart, stomach, and small bowel, this was 30 Gy maximum point dose. For the spinal cord, this was 18 Gy maximum point dose. Doses were increased or decreased by multiples of 3 Gy until the MTD was achieved. The PTV doses were prescribed to cover 95% of the PTV with no greater than a 140% maximum point dose. For all plans, a 2.5-cm jaw width was used.
Results: Regression analysis indicated a linear relationship between the MTD, the PTV and the liver volume, supporting our hypothesis. A predictive equation was generated, which was found to have an accuracy of ±3 Gy. In addition, dose constraints based on proximity to other normal tissues were tested. Inverse planning for PTVs located at varying distances from the heart, small bowel, and spinal cord revealed a predictable decrease in the MTD as the PTV increased in size or approached normal organs.
Conclusions: These data provide a framework for predicting the likely MTD for patients considered for Helical TomoTherapy liver SBRT.
Liver metastasis represents a significant clinical area in need of improved treatment. The prevalence of liver metastasis is very high, particularly among patients with colorectal, lung, and breast cancer. It has been shown that up to 50% of patients with Stage II and III colorectal cancer will develop liver metastases within 5 years of diagnosis. This is approximately 50,000 patients per year, with half of these presenting with liver as the only site of known metastatic disease. The 5-year survival rate for patients with liver metastasis treated with only systemic therapies remains less than 5%. Surgical resection for selected patients with liver metastasis provides survival benefit, resulting in a 5-year survival of up to 30%; however, operative/perioperative mortality is reportedly as high as 6%.
Radiotherapy has also been explored for liver metastasis. Traditional treatment of 30 Gy in 10 to 12 fractions results in palliative benefit but no substantial survival benefit. Dose escalation with fractionated regimens have shown some benefit, and protocols are currently accruing patients to hypofractionated regimens with 30 to 50 Gy in 10 fractions. SBRT has been used to deliver high doses, from 30 to 60 Gy in 3 to 5 fractions via linear accelerator-based therapies. The results from this approach have been promising, showing 1- to 2-year local control rates from 65% to 95%.
Various techniques have been used for hepatic SBRT. Three-dimensional conformal radiation therapy (3D-CRT) with linear accelerators has dominated the published experiences. These reports indicate the treatments are well tolerated and often allow retreatment of the same patient for out-of-field recurrence. Side effects reported include mild radiation-induced liver disease (RILD), characterized by transient transaminase elevation with or without fever and rare hemorrhagic gastritis. More recently, a Phase I dose escalation study established the safety of delivering 60 Gy in 3 fractions.
The goals of the current dosimetric study were to determine if Helical TomoTherapy is a suitable tool for hepatic SBRT and examine the relationship between target dose and normal tissue dose. We therefore examined the effect of the PTV and liver volume on MTD using hypothetical cases planned for treatment with tomotherapy. We hypothesized that the MTD to be delivered in 3 fractions would follow a linear function dependent on both PTV and liver volume. We also wanted to determine whether a simple patient eligibility formula for liver SBRT could be derived to screen potential candidates, especially patients with smaller total liver volumes after partial liver resection for previous metastases, based on easily measurable uninvolved liver and tumor volumes from diagnostic CT or MRI scans. In addition, the effect of PTV location relative to the heart, small bowel, and spinal cord on MTD was explored.
The integration of the limits on dose found in this study is the crucial step in determining the likely deliverable dose to a liver lesion with Helical TomoTherapy-based SBRT given the dosimetric constraints used. Although it was not tested, the results likely apply to other forms of coplanar IMRT or 3D conformal RT–based SBRT for liver metastasis, as the basic constraints of liver volume, PTV, and prescribed dose are the same. The many degrees of freedom for tomotherapy with beamlets allowed from any angle result in highly conformal coplanar plans. Noncoplanar beam arrangements may have a dosimetric advantage in certain instances although this was not tested in this study. The predictive equation presented provides an initial dose based on the optimal placement of the GTV in reference to constraining normal tissues. Although the predictive value for this linear function is high as indicated by the standard error, the data set had the worst fit at both ends of the PTVs tested. Thus, the dose found using the equation may be a less reliable prediction for extremes of PTV. In addition, the dose may be limited by the non–liver-related normal tissue constraints as shown. Dose constraints caused by proximity to the heart are limited by 2 factors: (1) the craniocaudal expansion of the PTV, and (2) the craniocaudal penumbra from the photon beam. The PTV expansion for this study was 10 mm in the craniocaudad dimension and 5 mm in the radial. This was a generic expansion used for calculation purposes only. Evaluation of tumor motion secondary to respiration with 4D-CT imaging, real-time fluoroscopy, or dynamic MRI in conjunction with abdominal compression allows for patient-specific PTV expansions. This could have a large effect on the PTV placement, thereby influencing the prescribed dose. Helical TomoTherapy may reduce the PTV component because of patient setup uncertainty by providing daily image-guided radiotherapy (IGRT) compared with systems without IGRT. This, in conjunction with adequate immobilization and abdominal compression to minimize intrafraction patient respiratory motion (BodyFix vacuum immobilization, for example; Medical Intelligence Corp., Schwabmünchen, Germany) may allow minimal PTV expansions. In reference to the penumbra, the use of smaller jaw size (primary collimator widths) may significantly decrease the penumbra in the direction of the heart while maintaining adequate PTV coverage, although smaller jaw sizes result in substantially increased treatment time. Additional constraints in relation to proximity to the spinal cord were tested. The proximity between the cord and the GTV were found to have no constraining effect on the tolerated dose, even for lesions up to 6 cm in diameter (data not shown). Normal tissue constraints as presented were tested as separate limitations. The interaction of these constraints is most pertinent in the left lobe of the liver, where proximity to stomach, small bowel, and heart can be simultaneous limiting factors. In this scenario, the utility of the plans as presented may be limited in predicting a tolerable dose.
The dose constraints used in this study were taken from a current multi-institutional Phase II SBRT protocol that is accruing patients nationally (Personal Communication with B. Kavanagh, University of Colorado, 2005). This protocol prescribes 60 Gy in 3 fractions. Helical tomotherapy SBRT for liver lesions based on the results of this publication should take this into consideration and should prescribe the MTD based on PTV and liver volume in 3 equal fractions. The dose constraints for normal organs in this trial were extrapolated from the critical volume model as well as the known constraints on partial liver resection. The constraint of 700 cc of normal liver to receive less than 15 Gy may be overly conservative for individuals with normal liver function, as no significant instances of RILD have been seen with this constraint . Another model used to predict liver tolerance, the NTCP model, has shown that the mean liver dose is the most significant predictor of RILD, with a threshold dose of 31 Gy. This has been supported by several reports of clinical experience. All plans in this study also met this constraint, with mean liver doses ranging from 5.3 Gy to 27.7 Gy.
Use of SBRT with helical tomotherapy for liver metastases is likely to be increasingly common in the future. The predictive equation reported in this work is intended to add value to treatment planning by providing a likely MTD based on patient-specific normal liver and tumor volumes. There are two main limitations to this model that must be addressed. First, the application of this equation is based on dose constraints that may prove to be overly conservative. Reports from the Phase I study by Schefter, et al. indicated that treatment with up to 60 Gy in 3 fractions using 3D-CRT was uniformly well tolerated. This study included some patients who would have fallen into the liver volume range included in the present study (uninvolved liver range, 875–2806 cc; median, 1783 cc). However, given the small number of patients treated, it seems unlikely that many (if any) patients were treated to a dose that would test the limitations of the predictive equation presented here. Indeed, the authors conclude that none of the patients treated were likely to experience RILD, based on mean liver dose. The second factor limiting the usefulness of this model is the lack of precise measurements of normal liver and target volumes before treatment planning. The uninvolved liver and tumor volumes can be measured from diagnostic MRI and CT images resulting in a rapid screen for potential candidates, especially those with small liver volumes after partial liver resection, which is not uncommon in this patient population. Before the simulation, normal liver volume can be predicted by several formulas. Liver volume can be predicted by body surface area (BSA). The model developed based on adults in the western hemisphere is as follows: total liver volume (TLV) = 1267.28 × BSA − 794.41. Vauthey et al. also presented a predictive equation based solely on patient weight, with similar predictive utility to the BSA-based equation. This equation is TLV = 18.51 × weight (kg) + 191.8. Normal liver volume can be measured or predicted from diagnostic CT images as well. PTV prediction may be difficult because of the irregular shape of many tumors as well as the required PTV expansion. This may be easily accomplished for spherical tumors using simple geometric formulas. This study was performed using cylindrical GTV structures with appropriate PTV expansions and therefore may represent a larger than average PTV for the maximum GTV diameter. However the asymmetric expansion of a spherical GTV because of the greatest motion in the craniocaudad direction secondary to respiratory motion will result in a more cylindrical than spherical PTV. These exercises in estimating PTV, liver volumes and subsequently dose may be useful in assessing the feasibility of liver SBRT before treatment simulation but will not substitute for careful CT planning with measurement of respiratory motion.
The HiArt Helical TomoTherapy system (TomoTherapy Inc., Middleton, WI) is capable of performing SBRT for liver lesions that meet the specified target and normal organ constraints as described above. This study provides broad initial screening eligibility criteria for patients with hepatic lesions who may potentially be suitable for tomotherapy or other linac-based SBRT with coplanar beams. These guidelines provide a likely acceptable SBRT dose to start with for planning purposes based on: GTV volume, total liver volume, radial proximity to stomach, small bowel and spinal cord; and craniocaudal proximity to the heart.