Correlation between Biological Effective Dose and Radiation-induced Liver Disease from Hypofractionated Radiotherapy.

BACKGROUND: The prevention of radiation-induced liver disease (RILD) is very significant in ensuring a safe radiation treatment and high quality of life. AIMS AND
OBJECTIVES: The purpose of this study is to investigate the correlation of physical and biological effective dose (BED) metrics with liver toxicity from hypo-fractionated liver radiotherapy.
MATERIALS AND METHODS: 41 hypo-fractionated patients in 2 groups were evaluated for classic radiation-induced liver disease (RILD) and chronic RILD, respectively. Patients were graded for effective toxicity (post-treatment minus pre-treatment) using the Common Terminology Criteria for Adverse Events (CTCAE) v4.0. Physical dose (PD) distributions were converted to BED. The V10Gy, V15Gy, V20Gy, V25Gy and V30Gy physical dose-volume metrics were used in the analysis together with their respective BED-converted metrics of V16.7Gy3, V30Gy3, V46.7Gy3, V66.7Gy3 and V90Gy3. All levels were normalized to their respective patient normal liver volumes (NLV) and evaluated for correlation to RILD. Results were measured quantitatively using R2 regression analysis.
RESULTS: The classic RILD group had median follow-up time of 1.9 months and the average PD-NLV normalized V10Gy, V15Gy, V20Gy, V25Gy and V30Gy metrics per grade were plotted against RILD yielding R2 correlations of 0.84, 0.72, 0.73, 0.65 and 0.70, respectively while the BED-volume metrics of V16.7Gy3, V30Gy3, V46.7Gy3, V66.7Gy3 and V90Gy3 resulted in correlation values of 0.84, 0.74, 0.66, 0.78 and 0.74, respectively. BED compared to PD showed a statistically significant (p=.03) increase in R2 for the classic RILD group. Chronic RILD group had median follow-up time of 12.3 months and the average PD-NLV normalized V10Gy, V15Gy, V20Gy, V25Gy and V30Gy metrics per grade were plotted against RILD grade yielding R2 correlations of 0.48, 0.92, 0.88, 0.90 and 0.99 while the BED-volume metrics of V16.7Gy3, V30Gy3, V46.7Gy3, V66.7Gy3 and V90Gy3 resulted in correlation values of 0.43, 0.94, 0.99, 0.21 and 0.00, respectively.
CONCLUSION: The strong correlations of the V10Gy and V15Gy PD-volume metrics as well as the V16.7Gy3 (BED of V10Gy) to both classic and chronic RILD imply the appropriateness of the current 15Gy evaluation level for liver toxicity with hypo-fractionated treatments. Copyright:

INTRODUCTION

In the modern-era clinic, the delivery of stereotactic body radiation therapy (SBRT) and hypofractionated treatments has become a commonplace in the management of primary[1] and metastatic liver tumors[23] with proven effectiveness in providing high local control rates.[4] The advancement of SBRT allows the precise delivery of high doses of radiation to targets while minimizing dose to surrounding critical structures. The central concern, however, is the possibility of radiation toxicity to the liver which in some cases can prove fatal.[567] This serves as a dose-limiting factor in the delivery of hypofractionated treatments.

As a consequence, the prevention of radiation-induced liver disease (RILD) becomes of paramount importance in ensuring a commendable standard of patient care, allowing safe radiation treatment and high quality of life. The use of the Common Terminology Criteria for Adverse Events (CTCAE) v4.0 allows the assessment of presence and magnitude of radiation-associated effects.[8] This assessment, however, becomes increasingly difficult with the use of hypofractionated treatments. The presently available data are limited to extrapolated toxicity from standard fractionated treatments and so the investigation of the relationship between RILD and absorbed dose becomes necessary.[19]

Typically, RILD occurs 4–8 weeks after termination of RT, but sometimes, it has occurred as early as 2 weeks or as late as 7 months post-RT.[10] A broad range of RILD incidence rates (~6%–66%) has been reported in the literature for hepatic radiation of 30–35 Gy. There are two types of RILD: classic RILD and nonclassic RILD. Patients with classic RILD usually have symptoms of fatigue, abdominal pain, increased abdominal girth, hepatomegaly, and anicteric ascites 0.5–4 months after liver RT, whereas patients who develop chronic RILD have underlying chronic hepatic diseases, such as cirrhosis and viral hepatitis, and show more dysregulated hepatic functions after 6 months post-RT.[5]

This issue is further complicated by the differing biological effects with varying fraction sizes.[11121314151617181920212223] However, the adoption of the linear-quadratic linear (LQ-L) biological effective dose (BED) formulation[24] for this study provides a basis, for which varying low and high dose per fraction treatments can be analyzed. This manuscript aims at determining the correlation of physical and BED dose metrics from hypofractionated treatments with liver toxicity as expressed by RILD. This information can be a very valuable input during treatment plan optimization where much of the effort is devoted to the reduction or even elimination of potential complications to the organs at risk.

MATERIALS AND METHODS

A total of 41 patients, who reported RILD, were enrolled in this study. The patient cohort was split into two subgroups according to time that they developed RILD (tfollow-up). In this way, they could be listed in the respective classic or chronic RILD groups. Thirty-six patients received an intensity-modulated radiation therapy (IMRT) course of 3 Gy in 8–10 fx (24–30 Gy) while five patients received intensity-modulated SBRT, three receiving 10 Gy in 5 fx (50 Gy), and two receiving 15 Gy in 3 fx (45 Gy). Group one, which corresponds to the classic RILD with tfollow-up<4 months, consisted of 25 patients (23 IMRT and 2 SBRT), whereas group two, which corresponds to the chronic RILD with tfollow-up >6 months consisted of 16 patients (13 IMRT and 3 SBRT). Patient's PD distributions were exported and converted to BED values using a constructed MATLAB 2010b (MathWorks, Boston, MA) graphical user interface conversion application. Conversions were made using the linear quadratic (LQ) model for doses <6 Gy per fraction and LQ-L model[22] for doses per fraction ≥6 Gy, which can be written as:

BEDn= D + [D2/(α/β)] for D < DT      (1)

and

BEDn= DT+ [DT 2/(α/β)] + [(γ/α)(D − DT)] for D ≥ DT      (2)

with

γ/α = 1+ [2DT/(α/β)]      (3)

where BED is the biological effective dose, n is the number of fractions, D is the dose per fraction, α/β is the point at which the linear and quadratic components of cell killing are equal, DT is the threshold dose for which the LQ model converts to the LQ-L model, and γ is the natural log cell kill per Gy in the high dose per fraction linear portion of the survival curve. Values of α/β = 3 Gy, DT = 6 Gy, n = 3, 5, and 10 fractions were used. γ/α was approximated using the slope of the line tangent to the LQ curve at the point DT= 6 Gy as done in Astrahan.[24] The tangent line, γ/α, expressed by the derivative reduces to 5 as follows: γ/α = 1 + 2DT/(α/β) = 1 + 2* (6 Gy/3 Gy) = 5.

This model was chosen due to the deviation of the LQ formulation prediction and experimental observations in many clonogenic cell-survival studies showing a dose–response relationship exhibiting an exponential decrease of survival at high dose which more closely approximates a straight line on a log-linear plot.[121319252627] Specifically, in the paper by Astrahan,[24] the survival response curves of a number of different tissues are presented as part of a very comprehensive analysis about the LQ and LQ-L models and the transition from one model to the other. The use of the exact value of 6 Gy is not critical in our study because the fractional doses of the SBRT cases are much higher than that of 10 and 15 Gy.

The patients were graded by a physician using the CTCAE v4.0.[8] Patients' pretreatment (initial) and posttreatment (follow-up) RILD grades were recorded. The effective RILD score was determined by the difference of those grades (follow-up grade minus initial grade). The normal liver volume (NLV) was defined as the total liver volume (TLV) minus the gross tumor volume (GTV).[28]

NLV = TLV − GTV      (4)

The mean physical dose (PD) together with the V10Gy, V15Gy, V20Gy, V25Gy, and V30Gy dose–volumes metrics to the NLV was calculated along with their respective BED values of V16.7Gy3, V30Gy3, V46.7Gy3, V66.7Gy3, and V90Gy3. The above dose–volumes were normalized for each patient by division by that patient's NLV. Based on the values of those parameters, the value of the averaged volume corrected dose () could be calculated by the following mathematical expression:

where V is the liver volume of the ith patient receiving at least X dose; NLV is the NLV of the ith patient; and N is the number of patients in the RILD grade category (acute or chromic).

The (Equation 5) was graphed against the effective RILD grade and evaluated via R-squared linear regression fitting to quantitatively determine the correlation between effective RILD grade, PD, and BED. Box-and-whisker plots were created to show the spread and distribution of averaged dose–volume fractions within each RILD grade.

RESULTS

The mean PD and corresponding BED values of all the patients examined in this study are shown in Table 1 for the classic and chronic RILD, respectively.

Table 1

Individual mean physical dose and corresponding biological effective dose values for the classic and chronic radiation-induced liver disease patient subgroups

Patient #	Classic RILD		Chronic RILD
	Mean PD (Gy)	Mean BED (Gy)	Mean PD (Gy)	Mean BED (Gy)
1	12.85	21.9	6.0	9.1
2	7.72	12.5	15.0	25.8
3	11.62	20.2	10.6	17.4
4	13.58	22	4.8	7.2
5	7.07	10.4	10.8	18.0
6	7.35	11.5	7.9	11.7
7	11.57	19.4	10.8	4.3
8	11.21	18.0	11.5	38.1
9	15.02	26.9	11.5	19.0
10	7.94	11.8	5.0	7.2
11	8.70	15.4	1.7	2.8
12	4.86	6.7	7.1	11.6
13	8.39	12.7	2.8	4.2
14	8.17	14.3	7.2	11.8
15	4.00	5.7	11.5	19.5
16	16.34	27.7	1.6	2.1
17	10.05	16.1
18	21.50	39.8
19	7.57	12.5
20	15.28	25.7
21	16.93	29.0
22	5.38	8.6
23	12.00	19.4
24	21.0	62.9
25	4.9	13.1

PD: Physical dose, BED: Biological effective dose, RILD: Radiation-induced liver disease

% for classic and chronic RILD is shown in Table 2. The classic RILD group had a median follow-up time of 1.9 months with the physical for V10Gy, V15Gy, V20Gy, V25Gy, and V30Gy per grade plotted against RILD yielding R2 correlations of 0.84, 0.72, 0.73, 0.65, and 0.70, respectively. The corresponding biological s of V16.7Gy3, V30Gy3, V46.7Gy3, V66.7Gy3, and V90Gy3 resulted in R2 correlations of 0.84, 0.74, 0.66, 0.78, and 0.74, respectively.

Table 2

Percent volume corrected dose averages for classic (tfollow-up <4 months) and chronic (tfollow-up >6 months) radiation-induced liver disease

	V10Gy	V15Gy	V20Gy	V25Gy	V30Gy	V16.7Gy3	V30Gy3	V46.7Gy3	V66.7Gy3	V90Gy3	n
Classic	43.3	27.8	18.8	12.3	7.6	38.4	21.3	13.3	4.7	1.2	25
Chronic	29.5	18.0	11.6	8.3	5.4	23.6	13.2	8.5	4.0	0.9	16
P	<0.05	<0.05	<0.05	0.13	0.22	<0.05	<0.05	0.11	0.71	0.87

Vd: Volume encompassing the d (Gy physical or Gy3 biological) dose level

The chronic RILD group had a median follow-up time of 12.3 months with the physical for V10Gy, V15Gy, V20Gy, V25Gy, and V30Gy per grade plotted against RILD grade yielding R2 correlations of 0.48, 0.92, 0.88, 0.90, and 0.99 while biological s of V16.7Gy3, V30Gy3, V46.7Gy3, V66.7Gy3, and V90Gy3 resulted in R2 correlations of 0.43, 0.94, 0.99, 0.21, and 0.00, respectively. Linear regression analysis is shown in Figure 1.

Figure 1

Averaged physical and biological dose–volumes per grade plotted against classic and chronic radiation-induced liver disease with the regression values also shown

The two closest correlated dose–volume levels of averaged PD and BED for both classic and chronic RILD were further analyzed by creation of box-and-whisker plots to display the distribution of data. This corresponded to V10Gy, V20Gy PD–volumes and V16.7Gy3, V66.7Gy3 BED–volumes for classic RILD which can be seen in Figure 2. For chronic RILD, this corresponded to the V15Gy, V30Gy PD–volumes and with BED–volumes of V30Gy3, V46.7Gy3 whose spread can be seen in Figure 3.

Figure 2

Box-and-whisker plots of the two most closely correlated dose-levels to classic radiation-induced liver disease corresponding to the V10Gy and V20Gy physical doses and the V16.7Gy3 and V66.7Gy3 biological effective dose values

Figure 3

Box-and-whisker plots of the two most closely correlated dose-levels to chronic radiation-induced liver disease corresponding to the V15Gy and V30Gy physical doses however, with the biological effective dose–volumes of V30Gy3 and V46.7Gy3

DISCUSSION

Ideally, balanced groups between IMRT and SBRT were planned to be used for this study. However, it was difficult to enroll patients and consistently collect clinical data for the SBRT arm of the study. Nevertheless, the scope of the study was not to compare the clinical effectiveness of IMRT vs. SBRT, but to identify the existence of correlations between dose–volume metrics with RILD. To accomplish our goal, we had to convert all the dose distributions to an equivalent fractionation scheme of 2 Gy fractional dose because fraction size has long since been known to be a dominant factor for toxicity risk.[29] The hypofractionation trend of increasing dose per fraction has shown clear evidence of increased local tumor control rates, however with an existing caveat of larger toxicity potential. In the case of the liver, the most severe effect is characterized as RILD which has been reported as one of the most serious treatment-related complications for patients with hepatic irradiation.[30]

The accuracy and validity of the liver PD constraint is questionable when delivering large doses per fraction to high total doses as is done in hypofractionation and especially SBRT.[1123] Table 3 highlights the current dose constraints for hypofractionated partial-liver radiotherapy as recommended by the quantitative analysis of normal tissue effects in the clinic (QUANTEC) group for radiation-associated liver injury.[31] One of the particular interests is the preservation of ≥700 mL of normal liver to receive a dose of ≤15 Gy. The use of this safety criterion has, to the date of the QUANTEC publication (2010), resulted in zero RILD or severe toxicity following SBRT.[32] While a serious complication rate of zero is applauded, this fact implies the restriction of potential improvements on overall survival rates attainable with dose escalation due to an overconservative dose–volume constraint.

Table 3

Current hypofractionated partial-liver radiotherapy dose-constraints as recommended by the quantitative analysis of normal tissue effects in the clinic group for radiation-induced liver injury

WLRT	RLRT	SBRT
≤30 Gy at 2 Gy/fx	<32 Gy at 2 Gy/fx	<15 Gy in 3 fx
21 Gy in 7 fx	>10% of normal liver spared	<20 Gy in 6 fx
		≥700 cc of normal liver receiving ≤15 Gy
		3-5 fx

Gy/fx: Gray per fraction, fx: Fraction, RT: Radiotherapy, RLRT: Partial-liver RT, QUANTEC: Quantitative analysis of normal tissue effects in the clinic, SBRT: Stereotactic body RT, WLRT: Whole liver RT

From Figures 2 and 3, we observe that very strong positive correlations (R2 = 0.70–0.99) exist with the lower dose levels of V10Gy and V15Gy for both classic and chronic RILD, corresponding to their biological equivalents of V16.7Gy3 and V30Gy3, where we also observe strong correlations. Furthermore, it is noteworthy to mention the near-equal correlations between PD and its respective BED in both RILD categories. Although the final results indicate that the PDs and their respective BED correlate with RILD very similarly, this is something that was not known at the beginning of this work. However, as a finding of this study, it is very interesting and it has significant clinical implications because it indicates that the PD constraints can be used during treatment plan optimization to reduce the risk for RILD post-RT.

It is very interesting to compare the present findings with results based on high dose rate (HDR) brachytherapy. A recent study tried to assess radiobiological restrictions and tolerance doses as well as other toxic effects derived from repeated applications of single-fraction HDR irradiation of small liver volumes.[33] The author reported that inactivation of liver parenchyma occurs at a BED of approximately 22–24 Gy corresponding to a single dose of ~10 Gy (alpha/beta ~5 Gy). This tolerance dose is consistent with the large potential to treat oligotopic and/or recurrent liver metastases by computed tomography (CT)-guided HDR brachytherapy without RILD. In another study, which tried to determine the safety and efficacy of CT-guided brachytherapy in hepatocellular carcinoma,[34] the authors performed 124 CT-guided brachytherapy sessions on 83 patients with one to three lesions per treatment. A high rate of local control was observed, regardless of applied dose in a range of 15–25 Gy. Although they reported nine complications requiring intervention, they found no evidence for RILD. However, there is a lack of studies in the literature, where dose–volume metrics and BED are correlated with RILD after HDR brachytherapy to make a direct comparison of the presented results.

In this study, the authors convert the different PDs to a common fractionation scheme to correlate them with RILD. However, this process is relying on the accuracy of the LQ and LQ-L models as well as on the accuracy by which the α/β value was determined. Even though the knowledge of those two factors is at a good level, the use of a higher number of patients where the different groups would equally be represented would give a more clear and reliable picture of the examined correlations.

CONCLUSION

This institutional retrospective review of hypofractionated treatments led to two main findings. The first is that the dose–volume metrics V10Gy and V15Gy are closely correlated with both classic and chronic RILD, which confirms the appropriateness of using the PD of 15 Gy as a constraint. A further investigation of these dose levels for RILD may yield their increased predictive power regarding immediate and lasting complications. The second finding is that the close correlation of BED with both RILD categories indicates its potential to be used as a treatment evaluator irrespective of the fractionation scheme applied. This study reaffirms that lower dose–volume metrics may prove prudent in biological relation in both immediate and lasting liver toxicities for hypofractionation and SBRT.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.