Final results of a phase II trial for stereotactic body radiation therapy for patients with inoperable liver metastases from colorectal cancer

Abstract

Purpose

To evaluate the feasibility and efficacy of stereotactic body radiation therapy (SBRT) in the treatment of colorectal liver metastases.

Methods

Forty-two patients with inoperable colorectal liver metastases not amenable to radiofrequency ablation (RFA) were treated with SBRT for a total number of 52 lesions. All patients received a total dose of 75 Gy in 3 consecutive fractions. Mean size of the lesions was 3.5 cm (range 1.1–5.4). Toxicity was classified according to the Common Toxicity Criteria version 3.0.

Results

Median follow-up was 24 (range 4–47) months. The progression in field was observed in 5 lesions. Twenty-four months actuarial local control (LC) rate was 91 %. Median overall survival (OS) was 29.2 ± 3.7 months. Actuarial OS rate at 24 months was 65 %. Median progression-free survival was 12.0 ± 4.2 months; 24 months actuarial rate was 35 %. No patients experienced radiation-induced liver disease or grade ≥3 toxicity.

Conclusions

SBRT represents a feasible alternative for the treatment of colorectal liver metastases not amenable to surgery or other ablative treatments in selected patients, showing optimal LC and promising survival rate.

Introduction

The incidence rate of colorectal cancer, compared to all other cancer types, in Europe and USA, is about 13 and 9 %, respectively (Ferlay et al. 2013). Thirty to 70 % of these patients will develop liver metastases, often isolated or associated with limited metastatic foci of disease (Yoon and Tanabe 1999; Scheele et al. 1990). In this oligometastatic condition, secondary liver involvement might represent an important worsening prognostic factor (Bozzetti et al. 1993).

The introduction of modern chemotherapy regimens in the current treatment for colorectal liver metastases has improved median overall survival (OS), although local control (LC) rates are still unsatisfactory (Biasco et al. 2006; Saltz 2005; Adam et al. 2009; Quan et al. 2012). Surgical resection also improves OS, even though most of patients suffering from colorectal liver metastases (70–90 %) are not good candidates to surgical treatment (Tomlinson et al. 2007; Adam et al. 2012; Clancy et al. 2014; Fong et al. 1999), and only 10–30 % can be converted to a resectable state with chemotherapy (Adam et al. 2009; Quan et al. 2012; Leonard et al. 2005). In total, about 10–60 % of patients with liver metastases are eligible for surgery.

Local approaches as radiofrequency ablation (RFA) can be a valid alternative to surgery, which is the treatment of choice, but presents some limitations in the presence of lesions larger than 3 cm of diameter or in proximity of major blood vessels, the main biliary tract or the gallbladder, or just beneath the diaphragm (Cirocchi et al. 2012; Gillams and Lees 2009; Siperstein et al. 2007).

In recent years, stereotactic body radiation therapy (SBRT) has developed as a system for delivering a conformal high dose of radiation to the tumour and a minimal dose to surrounding critical tissues, with a hypofractionation schedule, and has been used in several sites (Timmerman et al. 2003; Alongi et al. 2012).

This technique allows to overcome liver tissue low tolerance to irradiation and to avoid the radiation-induced liver disease (RILD), which can result in liver failure and death (Dawson et al. 2002). Several studies have investigated the efficacy of SBRT in the treatment of liver metastases from various primary tumours, with optimal LC rates and toxicity profile (Dawson and Lawrence 2004; Herfarth and Debus 2005; Katz et al. 2007; Lee et al. 2009; Scorsetti et al. 2013; Rusthoven et al. 2009; Corbin et al. 2013; Ambrosino et al. 2009). For patients with colorectal liver metastases not amenable to other locoregional treatments, SBRT can represent a therapeutic option for increasing LC and survival.

In this prospective study, we investigated the feasibility and efficacy of SBRT by means of volumetric modulated arc therapy (RapidArc, RA) in the treatment of liver metastases from colorectal cancer not amenable to surgery or other local treatments.

Materials and methods

Patients and study design

A prospective phase II trial, approved by the Institutional Ethical Review Board, on patients with inoperable liver metastases and/or not amenable to RFA, started in February 2010 at our Institution. Preliminary results were published elsewhere (Scorsetti et al. 2013). The protocol was amended to expand accrual to more patients and closed in October 2012. The present report addresses the final results for the subgroup of 42 patients affected by liver metastases from histologically proven colorectal cancer. The primary endpoint was to prove the efficacy of the SBRT approach to achieve LC. It was assumed as a failure a LC equal or inferior to 60 % and as a success, a LC ≧ 80 %.

According to the Fleming’s approach, to exclude an LC of 60 %, an α of 5 % and, to prove an LC of 80 % with a power of 90 %, at least 40 patients were needed with at least 33 cases of LC. Figure 1 shows the trial scheme.

Fig. 1.

Trial design

All patients were evaluated by a medical oncologist, a hepatobiliary surgeon, a radiologist, and a radiation oncologist in a multidisciplinary session.

Inclusion criteria were as follows: no evidence of progressive or untreated gross disease outside the liver, no concurrent chemotherapy, either within 14 days before SBRT or until the first follow-up, maximum tumour diameter less than 6 cm; no more than 3 liver lesions; normal liver volume larger than 1,000 cm³; no prior radiation therapy to the targeted area; adequate liver function, defined as total bilirubin <3 mg/dL, albumin >2.5 g/dL, normal prothrombin time (PT)/partial thromboplastin time (PTT) unless on anticoagulants, and serum levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) less than 3 times the upper limit of normality; no active connective tissue disorders; Karnofsky Performance Status of at least 70; minimum age of 18; and ability to provide a written informed consent.

Radiotherapy

Patients were immobilized with a thermoplastic body mask including a Styrofoam block for abdominal compression to minimize respiratory organ motion. A contrast-free and 3 phases contrast-enhanced computed tomography (CT) scan were acquired for all patients. In 11 patients (30.5 %), respiratory excursion was larger than 5 mm, and a 4-dimensional CT (4D-CT) imaging was necessary. Planning CT images were co-registered (with deformable registration methods) with magnetic resonance imaging (MRI) or positron emission tomography (from CT-PET) to better identify the gross tumour volume (GTV). The clinical target volume (CTV) was defined as equal to the GTV.

In all patients who underwent 4D-CT scan, an internal target volume (ITV) was defined as the envelope of all GTVs in the different respiratory phases. Dose plans were optimized on the average CT; this contributes to mitigate the residual organ motion not blocked by the abdominal compression. The planning target volume (PTV) was generated from either the GTV or the ITV by adding an overall isotropic margin of 5 mm from ITV or of 7–10 mm in the cranial–caudal axis and 4–6 mm in the anterior–posterior and lateral axes. A dose of 75 Gy in 3 consecutive fractions of 25 Gy prescribed as the mean dose to PTV; when full doses were not achievable because of organs at risk (OAR), then dose was compromised as described in (Scorsetti et al. 2013).

The plan objective was to cover at least 98 % of the CTV (ITV) volume with 98 % of the prescribed dose (V _98% = 98 %). Hot spots with doses higher than 107 % were accepted only if inside the target. Planning constraints for OAR were defined as follows: V _15Gy (volume receiving 15 Gy) < (total liver volume − 700 cm³) for healthy liver, $D_{0.1{\text{cm}}^3}$ for spinal cord <18 Gy (dose at a volume of 0.1 cm³ should be lower than 18 Gy), V _15Gy < 35 % for both kidneys, V _21Gy < 1 % for duodenum, small bowel, oesophagus, and stomach. In case of overlap between PTV and duodenum or stomach, the priority was given to the OAR. No specific constraints were applied to bile ducts, but as a general principle for SBRT treatments, the dose out of the PTV was “minimized” by applying strong fall-off requirements during the optimization.

SBRT was delivered using RA. The final dose distributions were computed with the Analytical Anisotropic Algorithm implemented in the Eclipse planning system. Patients were treated with 10-MV flattening filter-free (FFF) beams using the TrueBeam linac (Varian, Palo Alto, CA, USA) equipped with the millennium MLC with leaf dimension at isocentre of 5 mm. A maximum dose rate of 2,400 MU/min was used. Free breathing Cone Beam CT was performed daily to assess patient positioning. With this modality, a sort of “average” image dataset is acquired accounting for the residual organ motion not fully mitigated by the abdominal compression.

Assessment

Tumour response was defined using European Organization for Research and Treatment of Cancer Response Evaluation Criteria In Solid Tumours (EORTC-RECIST) criteria version 1.1 (Eisenhauer et al. 2009). Time to local progression was calculated as the time from the first day of SBRT to the day of first progressive disease of the irradiated tumour. Patients were observed for LC, even if distant or new liver metastases developed. Progression-free survival (PFS) included any intra- or extra-hepatic disease progression. After the end of SBRT, these examinations were requested 21 days after and then every 2 months. Imaging for follow-up included CT scans every 3 months and, with the same periodicity, PET-CT was also available for a subgroup of 54 % of patients. For all patients, tumour response was defined using RECIST criteria. For the subgroup of 23 patients with PET-CT, response was defined also using the PERCIST criteria. Acute and late toxicities were scored by the Common Terminology Criteria for Adverse Events (CTCAE version 3.0). Any increase in grade from baseline was considered toxicity related to the treatment. RILD was defined by Lawrence’s criteria.

Statistics

Kaplan–Meier method was used to generate the actuarial LC, OS, and PFS curves. Log-rank test was used to compare results between the group of lesions less than 3 cm and the group of lesions larger than 3 cm. Univariate analysis was used to correlate morphologic and clinical factors to OS and LC.

Results

Patients and treatments characteristics

Forty-two patients were enroled between February 2010 and October 2012 for a total of 52 liver metastases. Patients and treatment characteristics are shown in Table 1.

Table 1.

Patients characteristics

Patients number	42
Mean age and range (year)	67; (43–87)
Sex (M:F)	36:6
Primary site	Colon				Rectum
	30 (71 %)				12 (29 %)
TNM primary classification	T1	T2	T3	T4	N0	N1–2	M1
	2 (5 %)	9 (21 %)	28 (67 %)	3 (7 %)	21 (50 %)	21 (50 %)	17 (40 %)
Site of metastatic disease	Liver only				Liver and lung
	18 (88 %)				2 (12 %)
Timing of liver metastases	Synchronous (DFI < 12 months)				Metachronous (DFI > 12 months)
	20 (47.6 %)				22 (52.4 %)
Previous local treatments	Surgery				RFA or other
	17 (40 %)				4 (9.5 %)
Unsuitable to surgery	17 Patients with recurrence after previous surgery and unsuitable for a new hepatic resection 6 Patients with lesions close to vascular structures 19 Patients unsuitable for age and/or co-morbidities
Unsuitable to RFA	20 Patients with lesions >3 cm 4 Patients with recurrence after previous RFA 18 Patients with lesions close to vascular, biliary or gastrointestinal structures
Systemic treatments	Pre-SBRT chemotherapy				Post-SBRT chemotherapy
	42 (100 %)				6 (14 %)
Nbr of prior syst. treatments	0	1		2		3	>4
	2 (5 %)	12 (28 %)		7 (17 %)		9 (22 %)	12 (28 %)
Time of SBRT since diagnosis	≤12 months				>12 months
	3 (7 %)				39 (93 %)
Presence of stable extra-hepatic disease at SBRT time	Yes				No
	11 (26 %)				No (74 %)
Number of treated lesions	52
Nbr of lesions per patient	1		2		3
	34 (81 %)		5 (12 %)		3 (7 %)
Size of lesions	≤3 cm				>3 cm
	28 (55 %)				24 (45 %)
Mean volume (range) (cm3)	CTV				PTV
	18.6 ± 22.03 (1.8–134.3)				54.90 ± 41.90 (7.7–909.10)

Twenty patients (47.6 %) had synchronous metastatic disease at diagnosis, eighteen of which had only synchronous liver disease and two synchronous liver and lung metastases. Twenty-two patients (52.4 %) had metachronous liver metastases. For these patients, the median disease-free interval (DFI, time between diagnosis and liver metastasis) was 25 months (range 12–126 months). After diagnosis of metastatic disease, all patients received different chemotherapy regimens, 17 (40 %) patients received liver resection alone, while 4 (9.5 %) patients received RFA. Extra-hepatic metastatic disease was present in 11 patients (26 %) at the time of SBRT. The median time elapsed from diagnosis of metastatic disease to SBRT execution was 12 months. Number of treated lesions was 1 in 34 patients (81 %), 2 in 5 patients (12 %), and 3 in 3 patients (7 %). Mean lesion size was 3.5 cm (range 1.1–5.4). Median follow-up time from SBRT was 24 months with range from 4 to 47 months and a minimum of 18 months for the alive patients as per the protocol design. Six patients started new chemotherapy regimen after SBRT at the time of progression.

Dosimetric characteristics

Table 2 summarizes the numerical analysis performed on CTV, PTV, and OARs and based on dose volume histograms (DVH). Reported are the main parameters valuable for plan assessment, the corresponding planning objectives, the mean values of the findings (with 1 standard deviation uncertainty), and the observed range. As shown in the table, all objectives were met with one single patient exception for the near-to-maximum dose to the spinal cord. Figure 2 shows the average cumulative DVH computed for the whole cohort of patients (solid lines). The dashed lines represent the inter-patient variability expressed at ±1 standard deviation. For CTV and PTV, given the different prescriptions, dose is expressed in percentage; for OARs in Gy. Relatively low minimum doses were reported for PTV due to the strategy applied to mitigate the trade-off with OAR protection (reduction of dose prescription or acceptance of lower coverage).

Table 2.

Summary of the DVH analysis for the CTV, the PTV, and the OARs

Organ	Volume (cm3)	Parameter	Objective	Mean ± SD	Range
CTV		Mean (Gy)		74.3 ± 5.8	45.6–85.7
		(D 5% − D95%)/mean		0.04 ± 0.07	0.03–0.05
		D 1% (Gy)		76.0 ± 5.9	46.8–87.7
		D 99% (Gy)		72.0 ± 6.0	44.4–81.5
		V 95% (%)		99.6 ± 2.2	86.8–100.0
		V 107% (%)		3.0 ± 16.7	0.0–99.6
PTV		Mean (Gy)		71.4 ± 6.3	45.0–78.5
		(D 5% − D 95%)/mean		0.15 ± 0.09	0.14–0.16
		D 1% (Gy)		76.3 ± 5.8	46.6–87.4
		D 99% (Gy)		61.4 ± 12.6	16.2–74.4
		V 95% (%)		79.5 ± 25.9	29.1–100.0
		V 107% (%)		3.0 ± 16.7	0.0–99.6
Left kidney	182.2 ± 52.5 (28.5; 268.9)	Mean (Gy)		0.8 ± 0.7	0.1–2.2
		V 15Gy (%)	<35 %	0.0 ± 0.0	0.0–0.0
Right kidney	174.8 ± 41.7 (23.9; 254.3)	Mean (Gy)		2.7 ± 3.8	0.1–12.1
		V 15Gy (%)	<35 %	5.2 ± 10.1	0.0–35.2
Spinal cord	45.1 ± 26.6 (8.9; 108.4)	$D_{0.1{\text{cm}}^3}$ (Gy)	<18 Gy	9.3 ± 3.9	1.7–19.2
Stomach	232.1 ± 109.1 (106.1;544.7)	Mean (Gy		5.5 ± 2.8	1.3–10.7
		V 21Gy (%)	<1 %	0.0 ± 0.0	0.0–0.0
		D 1% (Gy)		12.1 ± 5.7	3.1–18.9
Duodenum	114.1 ± 104.5 (31.3; 358.7)	$D_{0.1{\text{cm}}^3}$ (Gy)		9.4 ± 8.8	0.7–25.9
		V 21Gy (%)	<1 %	0.0 ± 0.0	0.0–0.0
Left lung	1,350.9 ± 305.2 (555.6; 1,713.4)	Mean (Gy)		1.1 ± 1.4	0.1–5.7
Right lung	1,755.2 ± 436.2 (938.2; 2,643.6)	Mean (Gy)		2.3 ± 1.9	0.1–7.3
Liver-PTV	1,820.9 ± 1,780.3 (1,021.2; 10,623.6)	V 15Gy (cm3)	<Liver volume −700cm3	477.9 ± 228.6	228.6–477.9

Data are reported as average values plus or minus one standard deviation and range

St. Dev standard deviation, Dx % dose received by at least x % of the volume, Vx % volume receiving at least x % of the dose

Fig. 2.

Average dose volume histograms for CTV, PTV, and various OARs. Dashed lines correspond to 1 standard deviation inter-patient variability

Local control

Complete response was achieved in 22 (43 %) lesions, partial response in 17 (32 %), and stable disease in 9 (17 %) (Table 3). Figure 3 shows a complete response in a patient with two treated lesions. The mean LC was 43.5 ± 1.9 % (median not reached), and the actuarial 1-, 2-, and 3-year LC was 95 % (95 % CI 89–100 %), 91 % (95 % CI 82–99 %), and 85 % (95 % CI 73–97 %) as shown in Fig. 4a. Five patients for a total of 5 lesions developed in-field recurrence at 5, 10, 13, 14, and 29 months, with a median time to local progression of 17 months. A subgroup analysis for lesion diameter >30 mm compared with smaller metastases revealed no significantly increased risk of local recurrence (p = 0.92). No correlation was observed between LC and PTV or CTV coverage (also in the cases with quite low minimum doses to PTV); the minimum dose to PTV for the five patients with local recurrence ranged from 65.2 to 68.3 Gy (87–91 % of the prescription).

Table 3.

Patterns of progression and local response

Pattern of progression	Patients number (total 42) (%)
No progression after SBRT	17 (40)
Progressive disease after SBRT	25 (59)
Extra-field intra-hepatic progression only	7 (28)
Extra-hepatic progression only	10 (40)
Both intra-hepatic/extra-hepatic progression	8 (32)

Pattern of local response	Number of lesions treated (total 52) (%)
In-field response
Complete response (CR)	22 (43)
Partial response (PR)	17 (32)
Stable disease (SD)	9 (17)
Progressive disease (PD)	4 (8 %)

Fig. 3.

Patient treated with SBRT for two liver colorectal metastases. a–d Positron emission tomography (PET) pre-treatment image showing the lesions, defined by metal surgical clips. b–e Visualization of dose distribution on the planning target volume. c–f PET-CT image at 3 months after radiation therapy, showing complete metabolic response

Fig. 4.

a In-field local control after SBRT. b Progression-free survival for patients treated with SBRT. c Overall survival. d Overall survival for small and larger lesions

Progression-free survival (PFS) and overall survival (OS)

Twenty-five patients (59 %) presented with a disease progression. Patterns of progression are shown in Table 3. The median PFS was 12 ± 4.2 (95 % CI 3.8–20.2), and the actuarial PFS at 2 years was 48 % (95 % CI 32–64 %) (Fig. 4b). Twenty-seven patients (64 %) were alive at the time of analysis. Twelve patients (28 %) died for cancer-specific causes, whereas 3 (7 %) died of other causes. The median OS time was 29.0 ± 3.7 (95 % CI 21.8–36.2 months). The actuarial OS at 24 was 65 % (95 % CI 50–80 %) (Fig. 4c). Univariate analysis of the main prognostic factors affecting OS showed a statistically significant value only for the size of GTV with a decrease of OS for GTV larger than 3 cm (p = 0.01), as showed in Fig. 4d (see also Table 4).

Table 4.

Prognostic factors affecting OS rates on univariate analysis

Factors	Patients (no)	OS rates (%)		p value
		12 year	2 years
Synchronous liver metastases	17	81	55	0.91
Metachronous liver metastases	25	86	58
Disease-free interval (DFI) for metachronous disease
≤12 months	7	75	56	0.64
>12 months	18	87	70
Presence of extrahepatic disease at time of SBRT
Yes	11	70	55	0.91
No	32	81	60
Cumulative GTV
<3 cm	15	91	76	0.01
>3 cm	27	68	40
Number of lesion
One	35	80	58	0.94
2 or 3	7	71	55

Toxicity

Thirty-three patients (78 %) developed G2 acute toxicity. Specifically, the most frequent side effects were G2 fatigue (55 %) and G2-transient hepatic transaminase increase (25 %), normalized within the 3 months after SBRT. Five patients (12 %) with treated lesions in II and III hepatic segments experienced G2 nausea. No patients developed RILD. No grade ≥3 toxicity was observed. Despite the absence of explicit constraints, no evidence of bile ducts stenosis was observed yet consistently with the low expectation from the literature (Eriguchi et al. 2013) although with a relatively short follow-up, some late toxicity may not have yet manifested.

Discussion

Liver is a common site of progression for colorectal cancer, with an incidence of liver metastases of 30–70 %. The OS of patients with untreated colorectal liver metastases is about 20–30 % at 1 year, 8–10 % at 2 years, and 0–5 % at 5 years (Biasco et al. 2006; Saltz 2005) with a median OS ranging from 6 to 12 months (Saltz 2005; Adam et al. 2009; Quan et al. 2012; Tomlinson et al. 2007). The integration of chemotherapy and surgery is nowadays the standard of care, with acceptable rates of LC and survival (Saltz 2005). Chemotherapy in addition to local therapy might still be recommended. Surgical resection improves OS, with 1- and 5-year rates of 90–95 and 30–60 %, respectively, and with a median OS of 40–53 months (Tomlinson et al. 2007; Adam et al. 2012; Clancy et al. 2014; Fong et al. 1999), these results are in part an effect of patient selection.

Fong (Fong et al. 1999) defined favourable prognostic factors for 1,001 patients undergoing resection of the metastases including the absence of extra-hepatic disease, metachronous presentations with extended disease-free intervals (DFI) >12 months, single metastasis <5 cm, early-stage primary tumours, low carcinoembryonic antigen, and negative hepatic surgical margins. Tomlinson (Tomlinson et al. 2007) reported a 99 % long-term cancer-specific survival in 102 patients with colorectal liver metastases surviving ≥10 years after resection, suggesting the achievement of cure in this subset of oligometastatic patients.

About 40–90 % of metastatic patients, however, is not suitable for liver resection because of number and size of lesions as well as because of technical difficulties, unfavourable tumour factors, or patients co-morbidities (Adam et al. 2009; Quan et al. 2012; Tomlinson et al. 2007; Adam et al. 2012; Clancy et al. 2014; Fong et al. 1999; Leonard et al. 2005). RFA is an alternative to surgery, but its efficacy might depend on several factors. Our results have demonstrated that SBRT can provide good LC also for lesions larger than 3 cm in diameter, unlike the RFA data (Cirocchi et al. 2012; Siperstein et al. 2007).

In recent years, several studies have been performed about SBRT, mostly including patients with inoperable stage I non-small cell lung cancer and oligometastatic diseases (Timmerman et al. 2003; Alongi et al. 2012).

The rationale for oligometastatic ablation with SBRT represents a very complex net of several factors. Phenomenological elements, the patterns-of-failure concept, the theory of oligometastases as an intermediate and potentially curative state, the Norton–Simon hypothesis, and the impact of immune-modulation are some of the instruments applied in the rationale (Demaria and Formenti 2012).

In recent years, SBRT has also been investigated for the treatment of inoperable liver metastases, with encouraging results on LC and OS (Alongi et al. 2012; Dawson and Lawrence 2004; Herfarth and Debus 2005; Katz et al. 2007; Lee et al. 2009; Scorsetti et al. 2013; Rusthoven et al. 2009; Corbin et al. 2013; Ambrosino et al. 2009; Goodman et al. 2010; Kavanagh et al. 2008). The main series of liver SBRT for colorectal cancer metastases are summarized in Table 5. The three studies mentioned there include 44-, 20-, and 65-treated patients (Chang et al. 2011; Hoyer et al. 2006; van der Pool et al. 2010). Particularly, the best results were obtained by University of Texas Southwestern investigators who observed no local failures at 2-years after 60 Gy in 5 fractions delivered within 2.5 weeks (Rule et al. 2011) but on a cohort of miscellaneous primary tumours.

Table 5.

Results of current studies on SBRT for colorectal liver metastases

Author, (reference) design study	Patients with CRC	No. of lesions from CRC	Median lesion volume (range)	Dose (Gy/fr)	Follow-up (median)	Actuarial local control (%)		Actuarial overall survival (%)		Time to progression (median)	Toxicity≥G3 (%)
						1-year	2-years	1-year	2-years
Chang, phase II	65	102	30.1 ml (0.6–3,088)	22–60 Gy/1–6 fr	1.2 years	67	55	72	38	–	20a
Hoyer, phase II	44	–	35 mm (10–88)	45 Gy/3 fr	4.3 years	90	78	67	38	6.5 months	48a
Van der pool, phase I–II	20	31	23 mm (7–62)	37.5 Gy/3 fr 45 Gy/3 fr	26 months	100	74	100	83	11 months	10
Present study phase II	42	52	33 mm (11–54)	75 Gy/3 fr	18 months	95	91	80	70	12 months	0

^aHoyer et al. and Chang et al. reported progression to grade 2 and higher morbidity scores

In the present prospective study, 42 consecutive patients with a total of 52 (colorectal) liver metastases were treated with a dose of 75 Gy in 3 fractions. With a median follow-up of 24 months, LC rates were consistent with published data. Chang (Chang et al. 2011) treated 65 patients and 102 lesions with a LC rate of 76 % (55 %) at 1 (2) years. Hoyer (Hoyer et al. 2006) treated 44 patients with small colorectal liver metastases with 1- and 2-year LC rates of 90 and 78 %, respectively. Van der Pool (van der Pool et al. 2010) published data on 20 patients and 31 lesions with a LC rate at 1 and 2 years of 100 and 74 %, respectively. In these studies, median size of treated lesions was about 3 cm. In our study, mean lesion volume was slightly higher (3.3 cm), with a similar rate of local response. While Rusthoven (Rusthoven et al. 2009), with a prescription dose of 60 Gy in 3 fractions, showed a decrease of LC for lesions >3 cm in diameter, our results have shown that LC is not correlated with the size of the lesions (less or larger than 3 cm in diameter), when a higher prescription dose is administered. The role of dose escalation in the treatment of liver metastasis is confirmed by Chang et al. who showed a correlation between LC and a prescription dose higher than 48 Gy in 3 fractions and confirmed the strong correlation between LC and OS. These data are consistent with the knowledge that escalated dose of radiation can improve ablative effects expressed in terms of LC and survival. The rationale for the 3 × 25 Gy prescription to assess whether high LC was also achievable for larger lesions has been confirmed by our study and, regarding to individualized prescriptions, 18–20 Gy might be appropriate for the smaller while 25 Gy may be needed for the larger metastases.

In patients with more favourable prognostic factors, univariate analysis showed that OS is not significantly influenced either by the number of treated lesions or by extra-hepatic disease and DFI. This aspect seems to be related to the oligometastatic condition of the treated patients showing the benefit of SBRT in this setting. Cumulative GTV larger than 3 cm does not influence LC, suggesting the efficacy of our higher fractionation of radiation therapy, but influences OS. Large tumour size is an important prognostic parameter in treatment of CRC metastases, as showed by Fong in previous surgical series and Hoyer (Fong et al. 1999; Hoyer et al. 2006). We feel that it is important to perform SBRT in oligometastatic patients before a wider spreading of liver disease would appear.

One of the biases of the present study could be the follow-up period of 24 months that is relatively shorter than other data on surgery and local therapies. Wei showed an OS rate of 90–95 % at 1 year and 30–60 % at 5 years, with a median OS of 40–53 months after surgery. Gillams (Gillams and Lees 2009) and Siperstein (Siperstein et al. 2007) showed survival rates at 1–2 and 5 year of 87–70 and 34 %, respectively, and the median OS of 25 months after RFA. Our results seem to encourage the use of SBRT in the treatment of colorectal liver metastases in those patients not eligible for surgery and/or RFA because of co-morbidities, tumour size, and/or location.

To date, eligibility criteria and prognostic factors for patients with liver metastases candidate to SBRT are controversial, and a multidisciplinary tumour board discussion is recommended. According to the current literature, however, some selection criteria may be considered to identify the best candidates (Rule et al. 2011).

Conclusions

SBRT with a dose prescription of 75 Gy in 3 fractions is a non-invasive and effective therapeutic option for unresectable colorectal liver metastases and allows to achieve acceptable rates of LC and survival, both for small and larger than 3 cm lesions. With the current follow-up (minimum 18 months for survivors), no major toxicity was observed. Longer follow-up data and randomized phase III trials are needed to define the role of SBRT for unresectable colorectal liver metastases.