Surgery versus stereotactic radiotherapy for treatment of pulmonary metastases. A systematic review of literature

It is not clear as to which is the best treatment among surgery and stereotactic radiotherapy (SBRT) for lung oligometastases. A systematic review of literature with a priori selection criteria was conducted on articles on the treatment of pulmonary metastases with surgery or SBRT. Only original articles with a population of patients of more than 50 were selected. After final selection, 61 articles on surgical treatment and 18 on SBRT were included. No difference was encountered in short-term survival between pulmonary metastasectomy and SBRT. In the long-term surgery seems to guarantee better survival rates. Mortality and morbidity after treatment are 0–4.7% and 0–23% for surgery, and 0–2% and 4–31% for SBRT. Surgical metastasectomy remains the treatment of choice for pulmonary oligometastases.

Original search: "Lung metastases" surgery or "stereotactic radiotherapy" Titles screening: -Only original articles, no reviews, no case-reports -Treatment of pulmonary metastases surgery or SBRT Full-text selection: -11 full-text not available -56 articles excluded after reading -+1 article found from reference list cryoablation or other local treatments); safety or feasibility studies on new techniques were also excluded. If a manuscript reported on comparative results between surgical metastasectomy and SBRT, but only one of the two populations of study was greater than 50 patients, data were collected only on the arm of treatment meeting the minimal requirements on sample size. Last selection step was performed on full text analysis. One study was identified after checking references lists and included in the review. After final selection, a total of 79 studies were included and: of these, 61 were reports on surgical case series and 18 reported on the results of SBRT.

Bias
The majority of surgical studies (50.8%) do not report the size of resected lesions, which is a well-known prognostic factor and an important information regarding the burden of disease. Moreover, a few included articles reported the number of metastases ranging over five, which is the original uplimit for the definition of an oligometastatic state. Conversely, a minority of studies reported information regarding a history of previous local ablative treatments for oligometastases or the synchronous/metachronous development of metastases. Considering the relevant impact of these aspects on the patients' prognosis [10], this undoubtedly represents a huge limitation of the analysis. We observed how a relevant proportion of publications on surgical series report percentages of incomplete resections ranging between 3 and 21.3%: this represents evidence of involved margins of resection in most cases, but some studies reported percentages of macroscopic residual disease. Including this subset of patients in survival analysis may have a considerable impact on survival rates of the whole population, considering the known detrimental effect of incomplete resection on prognosis [11].
None of the included studies on SBRT reported that a histological confirmation of the metastatic nature of nodules was carried out before or after treatment. This might suggest that a proportion of the irradiated nodules were not neoplastic, which has potential impact on the survival and recurrence analyses.
Although a statistical analysis on the difference of populations' age between surgical and SBRT series could not be performed, median/mean age tend to be higher in SBRT studies. This point should be considered when comparing results of both treatments, since populations might be different in terms of baseline conditions, with relevant effects on outcomes.

Patient population & tumor characteristics Surgery
Results concerning study populations and the characteristics of tumors in surgical series are reported in Table 1. Among the 68 manuscripts on surgical series, only one was a prospective observational study, whereas the others were all retrospective. Some works report how their data were derived from prospectively maintained databases, but none of them clearly reported a prospective protocol for their study.
Population size ranged between 50 and 5206, with a median sample size of 114 patients (interquartile range [IQR]: . Apart from few articles reporting outcomes on the treatment of patients with metastases from sarcoma -where subjects were usually younger -mean or median values ranged uniformly between 48 and 66 years. With the obvious exception of the series of patients treated for metastases from breast cancer, the majority of works reported a marked prevalence of males in their populations. The larger proportion of studies pertained to populations of patients with metastases from colorectal cancer only (24 articles) or various primary tumor origin (19 articles). Seven articles illustrated results of patients treated for metastases from primary sarcoma. The large majority of articles (95.1%) report number of metastases, but the way this information is detailed vary quite extensively across studies: number of pulmonary deposits is referred as mean/patient, total number, single versus multiple or stratified per number with relative percentages/absolute numbers. The same observation can be made regarding size of pulmonary metastases, but in this case the information is provided in less than half of the manuscripts. A quite high conformity has been found in reported inclusion criteria. However, in 14 articles (20.6%), no reference was made to criteria for enrolling patients in the studies [12][13][14][15][16][17][18][19][20][21][22][23][24][25].

SBRT
Information regarding the study population and tumor characteristics of SBRT series are reported in Table 2. With the exception of one prospective study, all included papers regarding the results of SBRT were retrospective case series. Median sample population was 66 patients (IQR: 61-96, range: 50-700). Age has usually been reported as median or mean and range, whereas in three articles it has been stratified as 60 years the cutoff for subdivision. Only in one article median age was <60 years [26], whereas most populations' median/mean age ranged between 65 and 71. Male sex was prevalent in the majority of articles (88.9%) with a male/female ratio ranging between 0.93 and 6.12. Regarding tumor characteristics, 77.8% of articles included patients with pulmonary metastases arising from various primary tumors. Three articles (16.7%) were focused on results of irradiated metastases from colorectal cancer and only one on the recurrence of lung tumors. The number of lung deposits has been presented as absolute numbers, mean/patient or stratified with relative number/percentages. Tumor size has been variously described as diameter or volume, this latter both as gross tumor volume (GTV) or planning tumor volume (PTV). In 22.2% of papers, tumor size has been reported as stratified in classes, in a quite heterogeneous way across studies. Patients'   inclusion criteria were reported in 83.3% of articles, 72.2% also described definition criteria for an oligometastatic state and, in most of the publications, the maximum size of metastases accepted for irradiation <50 mm.

Treatment Surgery
The description of surgical approaches and length of follow-up are presented in Table 3. A common attitude was determined regarding surgical approach and extent of resection. The most common surgical approach reported was thoracotomy. Interestingly, a positive trend has been found in more recent publications in the reported percentages of resection performed via thoracoscopy (video-assisted thoracic surgery [VATS]). Bilateral localizations of disease have been treated with different approaches: staged thoracotomies within a short interval of time represent the main proportion, while other approaches such as median sternotomy, synchronous bilateral thoracotomy, handassisted procedures and clamshell incision have been described in several papers. A large proportion of authors favored a lung-sparing resection, atypical resection, which was also the most frequent intervention performed. Wedge resections were reported as almost the totality of procedure performed in one paper employing a substernal hand-assisted approach, whereas they represent only 11% of resections performed in a series of patients with metastases from breast cancer. Apart from these cases, in many publications atypical resections have been employed in approximately 60-70% of surgical procedures. Extensive resections represent a minor proportion in the majority of works, but pneumonectomies constitute up to 19.5% of resections performed in some series. Several studies mention percentages of patients were residual disease could be assumed (microscopic, R1) or evident (macroscopic, R2) after metastasectomy. A R0 resection was achieved in more than 90% of cases in most articles, but rates of incomplete resection and even macroscopic residual disease could reach 20% in several reports [18]. There was wide heterogeneity in reporting follow-up time with 15 authors (24.6%) not mentioning it at all. Median follow-up was greater than 2 years in 67.2% and greater than 4 years in 18% of included articles.

SBRT
The description of treatments and length of follow-up is reported in Table 4. A wide range of treatment regimens were found within and across studies, in terms of prescribed doses of radiation and fractions of delivery. Several papers report percentage of isodose delivered at isocenter and PTV and biologically effective doses (BED 10 ). All papers reported follow-up period, usually expressed as median and range. Median follow-up time was higher than 2 years in 44.4% of included articles, however, none of them reported a follow-up time of more than 4 years.

Outcomes & Complications Surgery
A description of the prognostic outcomes of surgical treatments and postoperative complications is reported in Table 5. Prognostic outcomes were variously described across studies: a survival analysis has been performed in the vast majority of articles, but intervals for assessing survivors' prevalence ranged widely between short-term (1 and 2 years) and long-term lapses (5 and 10 years). Rates of survival at 1, 5 and 10 years ranged between 88-96%, 15-76% and 11-41%, respectively. Recurrence-free survival (RFS) and PFS were reported in a minor proportion of manuscripts (24.6%). PFS rates at 1 and 5 years were 45-87% and 5-64%, respectively. Postoperative mortality was reported in 73.8% of studies, ranging between 0 and 4.7%. In nearly half of the included studies, a clear mention of the prevalence of complications was not found or could not be derived from presented data. Where reported, overall complication rates ranged between 0 and 23%. A description or a list of such complications has been found in 59% of papers.

SBRT
Outcomes of treatment were reported in an extensive manner in most papers ( Table 6). The authors report rates of overall survival (OS) or RFS in almost all articles, with the majority of them (88.9%) mentioning percentages of LC, reported as rates of LC and progression, local failure-free survival or local response. Intervals for assessing survival and tumor progression were definitely shorter in SBRT papers: only five of the included articles (31.2%) reported percentages of LC and survival at 5 years. Rates of OS at 1, 2 and 5 years ranged between 74-94.5%, 31.2-74.6% and 21.8-58.3%, respectively. PFS at 1 and 2 years was 23.5-72% and 10-57%, respectively. Mortality rates have been reported in 83.3% of publications and in 72.2% of series no patients died as a consequence of ablative treatment. Post-treatment mortality ranged between 0 and 2%. Overall rates of adverse events were described in 27.8% of papers, whereas a detailed description of them, often with relative percentages or absolute numbers, could be encountered more frequently. Adverse events were almost always reported using the Common Terminology Criteria for Adverse Events [27] and their incidence ranged between 4 and 31%. In the majority of articles that reported adverse events were Grade 1-Grade 2 toxicity, without the need for further treatments or minor measures only.

Discussion
Surgical excision of pulmonary oligometastases seems to guarantee improved outcomes in terms of survival, and it is currently the first-choice treatment for this condition [28]. In contrast, SBRT has traditionally been reserved for patients unsuitable for surgical treatment. Nonetheless, recent improvements in technology and more precise protocols are broadening SBRT current indications [29]. It has been proposed that SBRT might induce not only tumor cell death, but also a tumor-specific response of the host immune system, inactivation of remnant micrometastasis Stephens et al.  and improved control of disease, in what is known as 'abscopal effect' [30]. Consequently, some authors evaluated SBRT as an effective and less invasive alternative to surgery for patients with lung oligometastases [31]. The aim of this study was to report the current evidence on these two approaches for treatment of pulmonary metastases and compare them in terms of effectiveness and safety.
The main findings of our study can be summarized as follows: • We failed to demonstrate any substantial difference between surgery and SBRT in terms of short-term survival results. • Data on long-term outcomes suggest an advantage for surgery in terms of survival.
• The incidence of adverse events was overall similar in the reported articles, although in the radiotherapy articles an overestimation of adverse events has been observed. In the selected studies, OS at 1 year was 71-96% in surgical series and 74-94.5% after radiation therapy. A recent retrospective study comparing the two treatments for metastases from colorectal cancer demonstrated similar rates of survival in the short term, without significant differences in the two groups [32]. These figures are expression of a good outcome after resection/irradiation in the short term, but it is not clear whether this is a consequence of local ablative treatments or part of the natural history of the disease [33]. Patients with an oligometastatic state are a particular subset of subjects with favorable prognostic factors [33] and therefore, short-term survival is probably not the best parameter to establish the comparative effectiveness of these two treatments. A more appropriate indicator of therapeutic efficacy in the short term might be the PFS. In the present review, PFS was slightly higher in the surgical studies. In a recent retrospective study comparing surgery and SBRT for lung metastases, Filippi et al. demonstrated a worse outcome in terms of PFS and local progression among patients in the SBRT group [32]. Since in the context of metastatic disease the main goal of treatment is not cure but delaying progression and prolonging survival [34], the most important point is performing a procedure that eliminates any possible source of further metastatic spread. The presence of residual tumor has been demonstrated to be the main negative prognostic factor in many large surgical studies, some of them published in the late 90s [11]. While surgical excision with evaluation of resection margins allows an immediate feedback of completeness of resection, the efficacy of SBRT is defined with the concept of LC and it can only be assessed at follow-up investigations. The primary goal of SBRT for the treatment of oligometastases is to achieve an excellent rate of LC [31]. Although the R parameter refers to the degree of completeness of surgical resection and LC is a late measure of effectiveness of SBRT treatment, there is an evident analogy between the two factors in terms of potential eradication of disease. In the reported studies, slightly better rates of complete resection in the surgical series have been described when compared with LC at 1 year. This may have a significant impact on recurrence and PFS and therefore, on long-term prognosis.
Only a minority of reports on SBRT have a follow-up time long enough to describe long-term outcomes data, but from the included studies OS at 5 years tends to be higher in the surgical series. However, these results should be interpreted with caution, since SBRT is currently still reserved for patients unfit or unwilling to undergo a surgical treatment and this might have a considerable impact on the baseline conditions and survival rates of SBRT series populations. In SBRT studies the best outcomes seem to be reached with high doses of radiation: in the study from Kinj et al. [35]. 5-years OS reached 58.3% and in the series of Oh et al., this was 56% [36]. This confirms that high doses of radiation conveyed to the target nodule, expressed as biological effective dose (BED 10 ), result in better ablative power and control of disease [34]. However, the effectiveness of SBRT seems to depend not only on the irradiation protocols but also on the pattern of development of the oligometastatic state: in a recent large Japanese nationwide series, patients with a metachronous oligometastatic state demonstrated higher OS rates compared with those with synchronous oligometastasis [10]. Nonetheless, increasing attention is being paid toward the evolution of the pattern of disease in response to systemic therapy or previous local ablative treatments and future investigations will likely address the role of repeat-oligometastasis, the differences between a genuine or a chemotherapy-induced oligometastatic state and their impact on the prognosis of patients [37]. Moreover, the rapid evolution of systemic treatments and the introduction of molecular-targeted therapies might change the role of local aggressive treatments in this setting [38].
In the surgical series from Smith et al., the authors report a low survival rate of 15% at 5 years and in the same study, complete resection resulted the most relevant factor affecting survival [18]. In this study the prevalence of R0 resections was low (78%), confirming the importance of a complete eradication of tumor deposits during surgical treatment. Apart from this exceptions, 5-years OS is usually greater in surgical studies and can be as high as 76%. Moreover, several surgical series describe rates of survival at 10 years reaching 40% [39][40][41][42][43], suggesting that a stabilization or eradication of disease can be obtained in some cases.
Complications and adverse events were reported with similar percentages between surgical and SBRT series, but this is probably the result of a reporting bias: there is an overestimation of adverse events after SBRT, due to reporting of complications that do not require any further therapeutic measure.
There are consistent differences between surgical excision of metastases and other ablative techniques: excision of lung nodules allows availability of tissue for confirmation of the metastatic nature of nodules, assessment of resection margin for complete resection and pathological study for existing or future targeted therapies [44].
SBRT is an ablative technique that does not allow tissue harvesting for confirmation of diagnosis. Thus, one may wonder whether all sites of irradiation effectively correspond to metastatic deposits. Indication for irradiation is usually given at multidisciplinary tumor boards, based on patients' past medical history, radiological features of the nodules and dimensional evolution. Nevertheless, even with high levels of suspicion, some benign nodules may mimic tumor features in radiological examinations. Indeed, several surgical investigations report how some excised nodules did not confirm their metastatic nature at histopathological analysis [45,46]. Therefore, including patients with benign nodules treated with SBRT might overestimate the survival benefits in the whole population.
Another element of concern could be that treating patients with SBRT means not having any information regarding the R0 status after treatment or possible nodal involvement. Moreover, a suboptimal sensitivity of preoperative imaging techniques (CT, CT-PET) has been demonstrated [47] and unexpected further lung nodules at thoracotomy are not an uncommon finding [46,47]. For SBRT, this might implicate that radiation is not delivered on a potential source of further tumor dissemination.
Finally, our investigation demonstrated that surgical series provide information based on larger populations and with longer follow-up times. This makes, in our opinion, results coming from surgical reports more reliable. As long as greater experience will mature in the use of SBRT and more solid evidence will emerge from large series, we still recommend surgery as the main therapeutic option for oligometastatic disease of the lung. A prospective randomized protocol comparing the two techniques is required to confirm our conclusions.

Limitations
There are several limitations in this study. The low number of articles directly comparing the two therapeutic options lead the authors to include series dealing with one of the two treatments only. Nonetheless, most of the articles reported incomplete information on outcomes and baseline conditions of patients and this may have reduced the reliability of results. The net prevalence of retrospective studies and the presence of bias within and across them may have limited the accuracy of reported evidence. Including studies with metastases from several primary tumors may have limited the significance of the outcome results, but this was determined by the fact that most SBRT series deal with metastases arising from multiple histology. SBRT is usually reserved to patients excluded from surgical indication due to anatomical considerations or general conditions: this may have an impact on prognostic outcomes. SBRT articles described adverse events in a more accurate manner, reporting even low-grade events where there was no need for any intervention: this undoubtedly overestimated the overall reported incidence of complications compared with surgical series. Nonetheless many surgical articles did not report the R category, which is an aspect of main importance when dealing with a therapeutic resection for a neoplastic state.

Conclusion
Surgical resection of pulmonary oligometastases is a safe and effective practice and allows a good control of disease and prolongation of life, in cases where a complete resection can be achieved. SBRT is an attractive option with lower invasiveness and side effects, but long-term follow-up data are still limited. Nonetheless, the heterogeneity of therapy protocols in SBRT and the intrinsic differences between the two treatments do not allow to draw a conclusion on the superiority of one option over the other.

Future perspective
The recent technological evolution led to a better understanding of the tumors' behavior and the introduction of new molecules changed the view of metastatic disease from being a condition of negative outcome to a stage where the tumor can be stabilized or eradicated. This draws interesting perspectives: • What will be the role of local treatments in the context of an 'induced oligometastatic state' after systemic therapy regimens administered on patients with an overt polimetastatic disease? • In this setting, it would be of great interest to assess the new roles of the different ablative techniques during the evolution of disease; • Nonetheless, a Phase III trial with long follow-up times and strict inclusion criteria would be recommended to compare the effectiveness of the different aggressive local treatments in the context of oligometastatic disease.

Executive summary
• An oligometastatic state is a condition with a limited number of metastases, where local aggressive treatments, such as surgical resection or stereotactic radiotherapy (SBRT), may play an important therapeutic role. • Aim of the study was to define which is the best local treatment for patients with an oligometastatic disease.
• The study was conducted according to the PRISMA guidelines for systematic review and meta-analysis and literature search was performed by defining a searching strategy using the PICOS method. • Only articles in English language, with populations of more than 50 patients and where the treatment was performed with curative intent were included. • Sixty-one surgical studies and 18 articles on SBRT were finally selected.
• Overall survival is comparable in the short term but tends to be higher in the long term, although there is a scarcity of long-term results for SBRT. • Progression-free survival at 1 and 2 years tends to be higher in surgical studies.
• Mortality is comparable between surgery and SBRT, while the incidence of significant adverse events is higher in patients undergoing surgical resection. • The intrinsic differences between the two techniques and the heterogeneity of reporting of the included articles do not consent to define which is the best treatment. • Future investigations should include a prospective trial to compare the two modalities of treatment and define the role of the different ablative techniques in the era of new chemotherapy regimens.

Author contributions
F Londero contributed to the study conception and design, articles research and selection, data collection, manuscript draft and revision. W Grossi was responsible for the study design, articles research and selection, data collection, manuscript draft. A Morelli is also responsible for the study design, draft and revision of manuscript. O Parise performed data collection and organization, manuscript draft. G Masullo contributed to the study design, manuscript draft. C Tetta prepared the manuscript draft and revision. U Livi was responsible for the articles eligibility assessment, assessment of risk of bias, manuscript revision. G Maessen performed articles eligibility assessment, assessment of risk of bias, manuscript revision. S Gelsomino was responsible for the study conception and design, review process supervision, manuscript draft and revision.