Better survival of patients with oligo- compared with polymetastatic cancers: a systematic review and meta-analysis of 173 studies

Background: The modern concept of oligometastatic (OM) state has been initially developed to describe patients with a low burden of disease and with a potential for cure with local ablative treatments. We systematically assessed the risk of death and relapse of oligometastatic (OM) cancers compared to cancers with more diffuse metastatic spread, through a meta-analysis of published data. Methods: PubMed, the Cochrane Library, and EMBASE were searched for studies reporting prognosis of patients with OM solid tumors. Risk of death and relapse were extracted and pooled to provide an adjusted hazard ratio with a 95% confidence interval (HR 95%CI). The primary outcome of the study refers to overall mortality in OM vs. polymetastatic (PM) patients. Results. Mortality and relapse associated with OM state in patients with cancer were evaluated among 104,234 participants (n=173 studies). Progression-free survival was better in patients with OM disease (hazard ratio [HR] = 0.62, 95% CI 0.57–0.68; P <.001; n=69 studies). Also, OM cancers were associated with a better overall survival (OS) (HR = 0.65, 95% CI 0.62-0.68; P<.01; n=161 studies). In colorectal (CRC), breast, non-small cell lung cancer (NSCLC) and renal cell carcinoma (RCC) the reduction in the risk of death for OM patients were 35, 38, 30 and 42%, respectively. Biliary tract and cervical cancer do not significantly better in OM stage likely for paucity of data. Conclusions. Patients with OM cancers have a significantly better prognosis than those with more widespread stage IV tumors. In OM cancer patients a personalized approach should be pursued.


Introduction
The vast majority of metastatic solid tumors are incurable, and despite the evolution of treatments, patients ultimately die because of their disease. The modern concept of oligometastatic (OM) state was initially developed in 1995 1 to describe patients with a low burden of disease (e.g. 1 to 3-5 metastases) with a potential for cure with local ablative treatments. This assumption also relies on the hypothesis that metastatic spread follows a hierarchical pattern in time and number of localizations. 2 Large consensus on the definition and management of OM patients is currently lacking. Clinically, those cancers with a lower burden of metastatic disease have a favorable prognosis and they may be amenable of local treatment for the primary and distant tumors. Recently, infact, advances in imaging and local ablative therapies have permitted the treatment of these patients with additional locoregional treatment in addition to systemic therapies, and some patients may be cured and attain long term survival. 3 This scenario has been best elucidated in genitourinary, lung and melanomas. 4,5 In these settings oligometastatic cancers may be treated in oligoprogressive sites continuing systemic therapy that control the remaining disease. One of the first published trials proving benefit of an aggressive local treatment of oligometastases was published in Lancet during 2019. In the SABR-COMET randomized study median overall survival (OS) was 28 months (95% CI [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33] in the control group versus 41 months (26-not reached) in the stereotactic body radiotherapy to all metastases group (hazard ratio 0.57, 95% CI 0.30-1.10; P = .09). 6 The aim of this systematic review and meta-analysis was to investigate and establish the prognostic survival of OM compared to non-OM solid tumors. In particular, we evaluated if patients with oligometastatic solid tumors do better than patients with non-oligometastatic tumors.

Methods
This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines.

Search strategy and inclusion criteria
A comprehensive search was performed with the following terms: (advanced or metastatic or recurrent or relapsed or synchronous or metachronous) and (site or oligo* or "oligometastastic" or oligorecurrence or oligoprogression or single or multiple or 1-3 or >3 or >4 or >5 or 1-2 or 1-3 or 1-5 or number) and (synchronous or metachronous or metastases or relapse or recurrence or progression) and (tumor or tumour or cancer or carcinoma or melanoma or sarcoma) and ("hazard ratio") and (cox or multivariate or multivariable) and survival. We searched PubMed, the Cochrane Library and EMBASE for studies eligible for this meta-analysis published in English language from inception up to October 30 th , 2020. To be eligible, studies needed to have evaluated survival of patients with OM cancers (1 up to 3/5 metastases regardless of anatomic sites) regardless of line of therapy and to provide data of outcome according to the number of OM sites used by each author. Studies were excluded if they enrolled less than 10 patients, pediatric subjects, and hematological diseases. Commonly we define polymetastatic cancer as any disease with more than three or more than five metastases. Studies were searched and screened independently by three authors (FP, MG and GT).

Quality of studies and endpoints
The primary endpoint was overall survival (OS) and the secondary endpoint was progression-free survival (PFS). Quality assessment of the included studies was performed using the Newcastle-Ottawa Scale (NOS) for observational or retrospective studies (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp). With NOS scale, studies were defined as poor, sufficient or good quality if scores (the sum of points attributed to each domain) were <6, 6 or 7-9 points, respectively.

Data extraction and statistical analysis
The extracted data (from six reviewers) included the type of study, number of patients, cancer type, median age of included patients, performance status 0-1 (rate), treatment received, timing of oligometastases (synchronous or REVISED Amendments from Version 3 We have updated the table data as requested. We have changed a sentence in the final abstract section. We modified introduction by shortening the length. We also improved discussion by discussing the emergent problem of oligometastatic disease in light of the new imaging modalities. We discussed also the potential curative role of targeted local therapies (e.g. RT) in some disease (lung, renal carcinoma) supported by recent clinical trials. Some requests of the reviewer are not satisfied because data lack into included studies (exact burden of disease, treatments received in oligo vs polymetastatic subgroups). We extracted only data about timing of metastases (synchronous vs metachronous) that is reported in the table.
Any further responses from the reviewers can be found at the end of the article metachronous), number of OM sites used for comparison, and median follow up. Hazard ratios (HR) for OS and PFS with their 95% CIs, were extracted preferentially from multivariate analyses where available. The heterogeneity in the included studies was evaluated by the Chi-square-based Q-test and I 2 (I 2 = 0% to 25%, no heterogeneity; I 2 = 25% to 50%, moderate heterogeneity; I 2 = 50% to 75%, high heterogeneity; I 2 = 75% to 100%, extreme heterogeneity). When I 2 was larger than 50%, a random effects model was used; otherwise, the fixed effects model was used. Sensitivity analyses for OS were performed according to type of cancer, timing and number of oligometastases to find the potential heterogeneity among the included studies. If the number of studies was less than or equal to one, we did not carry out the subgroup analysis. The possibility of publication bias was explored by the Egger's and Begg's tests and Trim and Fill method. 7, 8 Begg's test explores bias with a funnel plot, conversely Egger's test is a linear regression of the effect estimates (OS) on their standard errors weighted by their inverse variance. The trim-and-fill method aims at estimating potentially missing studies due to publication bias in the funnel plot and adjusting the overall effect estimate. All analyses were performed using RevMan v.3 software. 9

Results
Among the publications retrieved using electronic search (n = 7510), 173 studies were eligible for meta-analysis, for a total of 104,234 patients 10 ( Figure 1). Baseline characteristics of the included studies and treatments received are presented in Table 1.

Records identified from:
Pubmed (              Progression-free survival was better in patients with OM disease (HR = 0.62, 95% CI 0.57-0.68; P < .01; n = 69 studies; Figure 2). Additionally, in the OS analysis, OM cancers were associated with a better OS (HR = 0.65, 95% CI 0.62-0.68; P < .01; n = 161 studies; Figure 3). Results were significant for all analyzed disease subgroups except biliary tract cancer and cervical cancer (only three studies included). In colorectal (CRC), breast, non-small cell lung cancer (NSCLC) and renal cell carcinoma (RCC), which constituted the more representative series, the reduction in the risk of death for OM patients were 35, 38, 30 and 42%, respectively ( Figure 3). Timing of onset (synchronous vs metacronous disease) did not influence the risk of death. Most studies reported OS analysis for up to three metastases (152 out of 161 studies).
After exclusion of eight studies that reported outcomes for up to five metastases the final results remained unchanged (HR = 0.64, 95%CI 0.61-0.67; P < .01). No cut-off was associated with a better outcome (1 vs 2 vs 1-2 vs 1-3 metastases).
Risk of bias through Begg's funnel plot was not significant for the OS analysis. Conversely, Egger's test showed evidence of bias (P < .01) (Figure 4). Trim and Fill analysis incorporated 29 missing studies. The overall effect measure (95% CI) based on this analysis was 0.7 (0.67-0.73), which became slightly weaker compared to the originally reported overall effect measure. Compared with cancers with more than three to five metastases, high-certainty evidence indicates OM tumors are associated with better prognosis in particular for CRC, breast, NSCLC and RCC. Despite the subgroup difference is not significant likely for less studies included in other groups, the results for these 4 cancers remain robust.

Discussion
The definition of oligometastatic refers to malignancies with a limited metastatic spread which may be amenable of radical treatment for both primary and each distant site, and that generally have a better prognosis compared to polymetastatic cancers. A very recently published paper clearly explains the timely clonal evolution of somatic mutations and consequently the metastatic process of many cancer types. 11 It may be hypothesized that OM cancer is associated with a more indolent spread and therefore may represent a less fatal disease. With the expansion of the oncological armamentarium, many efforts have been made over the years to improve outcomes of patients with minimal metastatic burden. Advance in imaging may also have improved in the last years the diagnosis of oligometastases with the possibility of a more targeted approach toward primary tumor and every single oligometastatic site. This may have created a bias compared to older series, where less accurate imaging modalities were available and more frequent cases of oligometastases could now be overdiagnosed.
We have performed the most exhaustive systematic review of the literature to quantify the prognostic value of OM stage in various cancers. Overall, OM cancer patients have a risk of death and progression that is a third less than the polymetastatic counterpart. The OM state is frequently calculated as an independent favorable prognostic variable, which means that these patients do well independent from other clinical-pathological characteristics. The effect size was calculated from 173 studies including more than 100,000 patients. The final results were similar in all the most frequent histologies including CRC, breast cancer, NSCLC, RCC and sarcoma with inferior survival in OM gastric, melanoma and head and neck cancers.
Prognosis of OM cancer may be also different according to site of oligometastases. For example in CRC, breast and RCC lung metastases have a generally more favourable outcome than liver (or peritoneal ones in CRC). In our series, sites of oligometasteses were mixed or not described at all so a subgroup analysis was not performed. 12 There is also evidence from randomized clinical trials 13-15 that ablative therapies improve survival in patients with OM cancer. For example, in some cancers small randomized studies 13-21 already provide evidence of survival improvement in patients that received both systemic and local therapies compared to those that received systemic therapies alone. As a matter of fact, resection of colorectal cancer liver metastases nowadays represents an essential curative option and a primary endpoint in multiple clinical trails. 13 Furthermore, Gomez et al. 14 found that in OM NSCLCs, adding local consolidative therapy to active oligometastases and to primary disease may improve OS from 17 to 41 months. Also, in RCC the treatment of indolent lung metastases may permit delaying the start of systemic treatment and obtain an excellent control. 15 A large burden of evidence now supports local therapy for minimal oligoprogressive cancers treated with targeted therapies or immunotherapy. Here, metastases-directed therapy could delay the switch of systematic therapy by radical local treatment of all progressive metastatic sites. 16,17 With the advent of immunotherapy, the combination of immune check point inhibitors and radiotherapy to single OM lesions may facilitate a potentiation of the immune response, increasing the chances of achieving an abscopal effect. This term describes an event in which focalized radiotherapy discharge systemic anti-tumoral action that can result in distant responses. 18 For example, in lung cancer the combination has a good safety profile and achieves high rates of local control and greater chances of obtaining abscopal responses than radiotherapy alone, with a relevant impact on outcome. 19 Oligometastatic cancers can also regarded as extended locoregional disease if, after proper conversion therapy, all sites of metastases and primary tumor may be radically resected with curative purposes. Such a strategy has been employed in largely incurable cancers as gastric and pancreatic carcinomas where selected cases with small liver-limited recurrences were managed with surgery. 20,21 Melanoma and head and neck OM cancers are also associated with better prognosis. In these settings isolated recurrences (lymph nodes, lung nodules or brain metastases) may be radically treated with surgery or radiotherapy.
This meta-analysis has several limitations. First, our review does not evaluate the absolute benefit of any local treatment and the prognosis and management of oligoprogressive disease or down staged polymetastases to an OM state. Second, the literature search covered a large lifetime span and may include older series where radiological evaluation did not include more advanced modalities that can now increase the accuracy of oligometastases detection. Third, most of studies have an observational design and outcome was retrospectively analysed. Likely publication bias may influenced the prognosis of this population. Finally, the optimal number of lesions defining the OM state cannot be defined in this paper.
A consensus paper of EORTC and ESTRO societies attempted to provide definitions of various OM conditions either naïve or attained after therapy and either synchronous or metachronous. 22 Some large, randomized studies have included local therapies for OM cancers. An NRG Oncology randomized phase II/III trial study compares therapy with stereotactic radiosurgery and/or surgery with standard of care therapy alone in treating patients with breast cancer that has one or two locations in the body (limited metastatic) that are previously untreated. The PREST study will assess the efficacy of ablative radiotherapy (stereotactic body radiotherapy applied to all oligometastases) administered to all tumor sites (metastases and prostate if applicable), in oligometastatic hormonesensitive prostate cancer patients. Finally, an ECOG-ACRIn phase III study compared standard chemotherapy to consolidative radiotherapy in patients with oligometastatic HER2 negative esophageal and gastric adenocarcinoma (https://clinicaltrials.gov/ct2/show/NCT02364557; https://clinicaltrials.gov/ct2/show/NCT04115007; https://clinicaltrials.gov/ct2/show/NCT04248452). In all ongoing studies the aim is the optimal timing (after a good shrinkage during systemic therapy) and integration of systemic medical therapy and local ablation/resection with the scope of improving long-term survivals.

Conclusions
In conclusion, this meta-analysis tried to quantify the prognosis associated with OM compared to cancers with more extensive diffusion. Based on our findings, we suggest that every metastatic patient should be accurately evaluated for the number of distant sites of disease, and a treatment strategy that involves both the primary and the metastases should be carefully considered. Patients could be reassured about their life expectancy and about the possibility of integrate both systemic and local therapy with the hope, in certain cases, for definitive cure. In others, focal treatment on the metastases may delay the immediate use of more toxic drugs (for example in elderly or indolent diseases). Also, we propose that these patients should be stratified when included in clinical trials and dedicated studies should be designed.

Data availability
Extended data Mendeley Data: Extended data for 'Better survival of patients with oligo-compared with polymetastatic cancers: a systematic review and meta-analysis of 173 studies'.

Open Peer Review
The authors of this report deserve praise for their extensive work. The data presented are intriguing and raise some intriguing questions.
For the included studies, it would be interesting to know the comparator group and its tumour burden. This information would help to appreciate the magnitude of the differences in outcome.

○
Additionally, one of the original questions raised with the introduction of the oligometastatic concept relates to the feasibility of local or locoregional treatment in a subset of patients with indolent disease. In this regard, I would present this information as a separate column in Table 1. Table 1, the columns on OS and PFS present rather generic information. Therefore, I would suggest including the specific outcomes. Also, the column on Age and PS should be split. Finally, the content of the column "Type of study" should be homogenised. ○ Further, the authors need to consider the time bias because modern imaging technologies increase the number of patients labelled as oligometastatic.

Always in
○ It would be essential to distinguish between different types of oligometastatic disease (e.g., indolent progressive and minimal residual disease after previous treatments). In this regard, in Table 1, the column "De novo or metachronous" seems to provide this information, but it is not entirely clear. Table 1 in the reference list.

Introduction
Please revise the language and, wherever possible, shorten the text (e.g. the first sentence is superfluous in this context). Also, please check some definitions such as "prognostic survival" and "with up to three to five metastatic sites."

Methods
Please adjust the definition of polymetastatic accordingly. Figure 1: More than 2,000 reports were excluded from the analysis. The reason needs to be clarified. Table 1 should indicate more clearly the prevalence of patients with oligometastatic disease.

Results
Page 14: "Timing of onset did not influence the risk of death". The authors should better explain this finding.

Discussion
The discussion could be improved by discussing some general issues first (challenges in the definition of OM, changing scenario in terms of diagnostic tools and available treatments) and then presenting some reflections on the cancer types where the effect of OM on OS was more prominent. For instance, the criterium of OM disease has been long applied in surgical oncology for selecting patients with lung metastases for surgical resection or patients with peritoneal carcinomatosis for cytoreduction and intraperitoneal chemotherapy.
intriguing and raise some intriguing questions.
For the included studies, it would be interesting to know the comparator group and its tumour burden. This information would help to appreciate the magnitude of the differences in outcome.
Data not available (comparator is the non-oligometastatic group but is not known site and number of metastases, for definition > 3-5 metastases).
○ Additionally, one of the original questions raised with the introduction of the oligometastatic concept relates to the feasibility of local or locoregional treatment in a subset of patients with indolent disease. In this regard, I would present this information as a separate column in Table 1.
Data not available. Table 1, the columns on OS and PFS present rather generic information. Therefore, I would suggest including the specific outcomes. Also, the column on Age and PS should be split. Finally, the content of the column "Type of study" should be homogenised.

OS and PFS are not generis but the exact outcomes (what is the meaning of specific outcomes?). Age and PS were split.
○ Further, the authors need to consider the time bias because modern imaging technologies increase the number of patients labelled as oligometastatic. Sentence added in discussion.

○
It would be essential to distinguish between different types of oligometastatic disease (e.g., indolent progressive and minimal residual disease after previous treatments). In this regard, in Table 1, the column "De novo or metachronous" seems to provide this information, but it is not entirely clear. Data were not available. Only the information reported were extractable.

Introduction
Please revise the language and, wherever possible, shorten the text (e.g. the first sentence is superfluous in this context). Also, please check some definitions such as "prognostic survival" and "with up to three to five metastatic sites." OK sentence modified. Sentences cancelled.

Methods
Please adjust the definition of polymetastatic accordingly. OK, sentence modified. Figure 1: More than 2,000 reports were excluded from the analysis. The reason needs to be clarified. OK reason included. Page 14: "Timing of onset did not influence the risk of death". The authors should better explain this finding. OK, sentence modified.

Discussion
The discussion could be improved by discussing some general issues first (challenges in the definition of OM, changing scenario in terms of diagnostic tools and available treatments) and then presenting some reflections on the cancer types where the effect of OM on OS was more prominent. For instance, the criterium of OM disease has been long applied in surgical oncology for selecting patients with lung metastases for surgical resection or patients with peritoneal carcinomatosis for cytoreduction and intraperitoneal chemotherapy. OK sentences added. The authors present the results of their systematic review and meta-analysis assessing the prognostic impact of oligometastatic disease on adult patients with solid tumors, as compared with a more diffuse metastatic spread. Overall and progression-free survival were significantly longer in patients with 3-to-5 metastatic lesions, irrespective of anatomic site. This may sound quite obvious in modern oncology, but the authors were able to provide a large amount of clinical data to support such an assumption.

Abstract:
In the Introduction, the main topic of this literature review was described concisely but exhaustively.

○
In the Conclusions, please use the term "oligometastatic disease (or OM)" instead of "oligometastases".

Methods:
The methodology of literature search and data extraction, paper selection criteria, and statistical analyses are thoroughly described. The review was carried out according to international guidelines (PRISMA). Please, clarify if papers not in English language studies were included.
○ Also, the Newcastle-Ottawa Scale (NOS) might be briefly described, as a number of readers may be not familiar with it.

Results:
In the Results section, the authors state that the reduction in the risk of death for oligometastatic patients was 35%, 38%, 30%, and 42% for colorectal, breast, non-small cell lung cancer, and renal cell carcinoma (RCC), respectively. In another part of this section, they state that compared with cancers with more than three to five metastases, "high-certainty evidence indicates OM tumors are associated with better prognosis in particular for CRC, breast, NSCLC and RCC". However, was such a difference significant? In agreement with Reviewer 1, I would suggest to group studies according to histology, and to graphically depict the risk of oligometastatic vs. more advanced disease for each of the four tumors mentioned above.

Discussion:
The Discussion was improved according to the suggestions of Reviewer 1, resulting in a stronger manuscript. There is an additional concept that I would address in the paper: the fact that the site of metastatic disease may affect patient prognosis, in addition to the number of metastatic lesions. In colorectal cancer, peritoneal metastases are associated with worse prognosis as compared with liver metastases, and lung metastases are associated with better prognosis. Furthermore, specific areas within the same organ may be related to a worse prognosis, e.g. a metastasis involving the hepatic hilum may be worse than a subcapsular liver metastasis.

Is the statistical analysis and its interpretation appropriate? Yes
Are the conclusions drawn adequately supported by the results presented in the review? Yes I included a statement in the Methods section about the exclusion of non-English language papers and the NOS scale definition. The studies could be regrouped according to the histology. The same could apply to Figure 2 and Figure 3.

○
The prognosis of patients with gastric cancer, melanoma, and head and neck cancer should be discussed in light of the results presented.

○
In the Discussion, it is not entirely clear if the authors consider the OM status an opportunity to spare patients from systemic treatment or an opportunity to pursue combined treatment. Again, this should be discussed in light of the results presented.

○
In the Discussion, the last paragraph seems more like a list of ongoing trials, including some form of local therapies over standard systemic treatment. How does this relate to the findings of the present study? Please discuss.
○ Are the rationale for, and objectives of, the Systematic Review clearly stated? Partly Are sufficient details of the methods and analysis provided to allow replication by others? Yes

Is the statistical analysis and its interpretation appropriate? Yes
Are the conclusions drawn adequately supported by the results presented in the review? Partly Here you can find some comments that you may find useful to improve this review: In the Abstract, I would mention the histotypes in which the OM status do not correlate with patient outcome. Also, in the Conclusions part, second sentence: this seems to be unrelated to the results presented and anyway not applicable in all cases (consider rephrasing/changing). Author response: OK -requests accepted.  Table 1, 8th column: some of the included studies have "various" sites of OM. I think this information should be specified in order to be consistent with the inclusion criteria. Author response: "Various" means that in those articles, sites of metastases were not specific. Only when explicitly reported they are included (e.g liver or lung). Specific comment in inclusion criteria added.

○ ○
The studies could be regrouped according to the histology. The same could apply to Figure 2 and We have improved the Introduction and criteria for search.

○
We have arranged in the Discussion section a specific discussion about particular settings of patients analysed and the main limitation of the paper (retrospective nature of studies).

○
We also discussed the main meaning of the results: improved prognosis and treatment opportunities with locoregional therapies in an oligometastatic setting.