Discussion
To date, the role of SBRT as ablative therapy of liver metastases is not consolidated and only recently, ESMO guidelines [
28] added SBRT in the toolbox of local therapy for selected patients with CRC liver metastases unsuitable for surgery or RFA/MWA. All this is mainly related to the lack of long-term results from prospective and randomized trials. This report summarized the results of a phase II trial with more than 5-year follow-up for patients treated with high dose SBRT. The rational and the early results for the very high dose regimen proposed are summarized in our own earlier works [
16,
17,
29].
Concerning treatment margins, the present study described the data from the original protocols which required relative large expansions from the CTV or ITV. With the current practice and with the knowledge that well-done abdominal compression, eventually associated to some additional breath control (e.g. breath hold) might allow to further reduce the margins and, consequently, to further protect the healthy fraction of the liver. This option was not part of the current investigation.
Table
4 summarizes the long-term outcome data of the few published SBRT trials with more than 40 enrolled patients and with a median follow-up longer than 2 years. The study by McPartlin [
30], reported a 2 and 4-year LC rates of 49.8 and 26.2%, respectively for patients treated with 33 to 57Gy in 6 fractions. These results are worse than surgery and thermo-ablation and dose escalation is needed to improve sustained local control, as stated by the same authors. Similar conclusions were confirmed by Klement and by Joo, asserting that the use of ablative prescription dose is a crucial starting point to make SBRT a valid therapeutic option [
31‐
33]. Joo [
31] reported 2-year LC rates of 52, 83, and 89% using a biological equivalent dose (BED) ≤80Gy (group 1), 100-112Gy (group 2), and ≥ 132Gy (group 3), respectively. The authors concluded that better LC can be expected if higher doses are applied.
Table 4
Long term results of current studies on SBRT for liver metastases
Hoyer, [ 15] Phase II, 2006 | 44 | 45Gy/3fr | 4.3 years | 79% at 2 years | – | 67% | 13% | 48% |
Fode, [ 18] Retrospective. 2015 | 212 (321 tot oligometastatic patient) | BED10 = 72–195 Gy | 5 years | 91% | – | 80% | 23% | 4% |
(Not specified for liver metastases) | (Not specified for liver metastases) |
Goodman, [ 39] Retrospective 2016 | 81 | 32–60/3–5 fr | 2.25 years | 96% | 91% at 4 years | 89.9% | 28% at 4 years | 4.9% |
McPartlin, [ 28] Phase I-II, 2017 | 51 | 22.7–62.1 Gy /6 fr | 2.3 years | 49.8% | 26% at 4 years | 63% | 9% at 4 years | 3% |
Joo, [ 29] Retrospective 2017 | 70 | 45-60Gy/3–4fr | 2.8 years | 93% | 68% at 3 years | 75% at 2 years | 0% |
Mendez Romero [ 40] Retrospective, 2017 | 40 | 37.5 Gy / 3fr | 2.2 years | 96% | 66% at 3 years | 95% | 48% at 3 years | 7.5% |
50.25 Gy / 3 fr | 90% | 81% at 3 years | 94% | 65% at 3 years |
Present study | 61 | 75Gy/3 fr (82% of lesions) 67.5–52.5 Gy/3fr (18% of lesions) | 6.1 years | 94% | 78% | 85.2% | 18% | 1% |
Klement [
32,
33] suggested the use of doses greater than 3x17Gy and suggested to deliver a BED of 209 ± 67Gy to achieve a 2-year LC rate of 90% in patients with no prior chemotherapy and 286 ± 78Gy when chemotherapy was administered. The authors confirmed also the strong influence of histology and pre-treatment chemotherapy on LC, concluding that metastases from breast cancer respond better to SBRT compared to other histology types and that increased radiation resistance of CRC metastases may rather reflect the higher resistance after pre-SBRT chemotherapy. These innovative results could improve patient selection to liver SBRT and its efficacy, even though more prospective long-term data are needed.
The current prospective phase II study could therefore contribute to this search for evidence.
The fractionation employed in our trial escalates the dose with a BED ranging from 144.38Gy to 262.5Gy. Our results confirmed the safety and efficacy of ablative doses with a very high and sustained LC rate of 78% and with a very low toxicity profile after more than 5 years of follow up. Considering the suggestions of previously cited studies, our results showed also the effectiveness in disease progression control even in those unfavorable patients with bigger lesions (diameter > 3 cm), radio-resistant histology such as CRC and/or heavily pre-treated with systemic and local therapies.
The size of the lesions is a relevant factor affecting local control of ablative radiotherapy. Correlation between LC and lesion size higher than 3 cm was demonstrated by Rusthoven in a phase II trial on liver SBRT, using a dose prescription of 60Gy in 3 fractions [
12]. Dose-response and size-response relationships were confirmed also in the recent report by McPartlin [
30], with a 4-year local control rate of 26%, correlated with the GTV. Long-term results using the fractionation of 75Gy in 3 fractions showed that a promising LC can be also obtained in bigger lesions with a diameter ranging from 3 to 6 cm, with a 5-year LC rate of 81% as shown in the current study.
The high local response rate was not significantly influenced by any factors in our experience and CRC metastases presented a competitive LC rate at 5 years of 75%, although most of these patients were unfavorably selected. More than 50% of patients, indeed, received liver-directed therapy prior to SBRT and 86% of CRC patients were heavily pre-treated with systemic therapies before radiation treatment.
While the role of previous systemic treatments in inducing radio-resistance has been widely explained by Klement [
32], the effect of previous local therapies is being studied.
In the last decades, the growth of cancer stem cells (CSC) in the site of previously local-treated tumors has been demonstrated. CSC seems to have the ability to self-renewal, to generate heterogeneous progeny and to self-divide unlimitedly. Considering these characteristics, CSCs seem to correlate to chemo- and radio-resistance development [
35]. In our experience previous local ablative therapies were borderline significant for local control. This result could be explained by considering the limited number of enrolled patients in this trial and future studies on larger series could be needed to confirm this correlation.
Critical analysis of LC data confirmed the suggestion from Klement about a better response to SBRT of breast cancer liver metastases, with an optimal 5-year LC rate of 87%.
Primary tumor and previous therapies are crucial factors influencing the OS also. The efficacy of local ablative therapies as potentially curative for oligometastatic patients is consolidated for surgery of CRC liver disease [
25,
34,
36‐
38], but it is debated for non-CRC liver metastases [
39‐
42].
In a multicenter phase I/II study on patients treated with SBRT for liver metastases, Rusthoven [
12] showed a 2-year OS of 30%. In this prospective experience, OS was better in patients with favorable cancer types (breast, CRC, sarcomas and renal cancer) than in patients with less favorable tumor types (lung, ovarian, and non-CRC gastrointestinal cancers), with a median OS of 32 months versus 12 months, respectively. These results promoted the efficacy of hepatic metastases ablation also for selected non-CRC liver metastases.
Although our study is not comparative and the primary endpoint is not OS, results show a benefit for the use of SBRT in the multidisciplinary management of oligometastatic patients with inoperable liver metastases not only from CRC, but also from other favorable primary disease, such as breast and gynecological cancers. In CRC liver metastases, indeed, our fractionation allowed an encouraging OS rate of 41% at 3 year and 27% at 5 year. Similar positive results were also achieved for patients with gynecological cancer-derived liver disease, with 3 and 5 year OS of 57 and 28%, respectively. Patients irradiated with breast cancer derived liver metastases showed a lower survival with a 3–5 year OS rates of 33–20%, respectively.
This apparently negative outcome could be related to an unfavorable selection of these patients. Indeed, patients with liver metastases from breast cancer, were pretreated with more systemic therapy before SBRT than those patients with other histological types. All 11 (100%) breast cancer patients received CHT and 8 (73%) patients were pretreated with more than 3 lines of systemic therapy. In this subgroup of patients, therefore, SBRT was more delayed.
These data confirmed that liver oligo-metastatic patients are often heavily pretreated and are unsuitable for the following systemic or local therapy at the time of referral for SBRT. As shown by Fode [
19], the use of SBRT allows to decrease tumor progression in this setting of oligo-metastatic patients, with an encouraging 5-year OS rate of 23%, comparable to our result.
In our trial, all patients were unsuitable for surgery and RFA, 87% of patients received chemotherapy pre-SBRT and about 40% were heavily pre-treated with more than 3 lines of systemic therapy before SBRT. Supportive care remained the only therapeutic option for most of these patients, with an expected OS rate likely lower than 23% at 5-year [
19].
Analyzing the current literature data on oligometastatic patients treated with ablative therapy, other prognostic factors for OS have been also identified beyond those of our study. A possible reason for this data could be given by patient selection.
All patients were enrolled in our trial according to restrictive inclusion criteria. They were defined by considering the importance of same prognostic factors suggested by surgical experiences. Since 1999, Fong showed the significant correlation between OS and specific variables, such as small number of lesions, size < 5 cm, absent or controlled extrahepatic disease, in patients with CRC liver metastases, treated with hepatic surgery [
36]. Selection of patients at the time of enrollment, therefore, could justify why these variables do not improve OS in our experience. This remark seems to be confirmed by Chang in a pooled analysis on SBRT for CRC liver metastases [
17]. In this experience, which employed inclusion criteria similar to our trial, number of lesions (1 vs 2–4), lesion size and active non-liver disease were not correlated with OS, as well.
The need to identify simple predictors of outcome, an important task of state-of-the-art research, was investigated by Mazzola [
43]. The authors proposed the use of diagnostic signatures from PET imaging as predictors of local response in liver metastases. These included pre-SBRT SUV-max and SUV-mean. Considering the relatively low sample size, our long term results suggest that patients selected by favorable primary disease, number, size and site of liver metastases and controlled extrahepatic disease, could benefit from high dose of SBRT also in terms of OS.
The data presented in our report refer to linac based SBRT. The use of alternative delivery modalities, like, e.g., robotic devices as CyberKnife has been investigated. A review from Ihnat [
44] showed that the most frequently treated are for less than 5 metastases with a maximum of 6 cm and no extrahepatic disease. In general, severe toxicity is rare and local control range from 70 to 100% at 1 year and from 60 to 90% at 2 years.
As a final remark, it should be advised that, thy type of very high-dose stereotactic regime should be carried out in appropriately qualitied and experienced centers and that sufficient quality assurance procedure are enforced to guarantee the safety of the treatments.