Introduction

At the time of diagnosis, approximately 20–25% of patients with colorectal cancer presented with synchronous metastases, which are unresectable in 75–90% of these patients [1,2,3]. In addition, patients may present with advanced localized disease that is unresectable due to extensive involvement of surrounding structures or due to involvement of vital structures. For patients with incurable colorectal cancer, an important question which remains unanswered to date is whether the best treatment strategy is primary tumor resection (PTR) with chemotherapy or immediate chemotherapy without PTR. Previous published comparative studies reported conflicting results on this issue, with some studies demonstrating improved survival with PTR compared to primary tumor intact (PTI) [2, 4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32], while other studies found no significant differences between the two groups [33,34,35,36,37,38,39,40], and other studies suggested systemic chemotherapy without resection of the primary tumor is the treatment strategy of choice for patients with incurable colorectal cancer [41,42,43,44,45,46,47,48,49,50]. The purpose of the present study was to perform a systematic review of the literature and employ meta-analytical techniques to compare survival and adverse events in patients undergoing PTR versus PTI, with or without chemotherapy, in order to determine whether PTR should be performed in patients with incurable localized or metastatic colorectal cancer.

Materials and Methods

Search Strategy

This systematic review and meta-analysis was based on a written protocol and was reported in line with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [51] and Assessing the methodological quality of systematic reviews (AMSTAR) guidelines [52]. A comprehensive literature search was performed of the following databases: PubMed, MEDLINE, Embase, Science Citation Index Expanded, and Cochrane Central Register of Controlled Trials (CENTRAL). Detailed search strategy is provided in the Supplementary Table 1. All abstracts, studies, and citations identified were reviewed, and the references of the identified studies were also searched. No restrictions were made based on language, publication year, or publication status. The latest date for this search was May 2, 2018.

Selection Criteria

Prospective or retrospective studies were considered for this meta-analysis if they met the following criteria:

  1. 1.

    Reported on patients with incurable metastatic colorectal cancer. Incurable metastatic colorectal cancer was defined as the presence of unresectable metastases, including liver metastases, lung metastases, intraperitoneal, and omental carcinomatosis, considered too extensive to achieve a complete or macroscopically curative resection.

  2. 2.

    Reported on patients with incurable advanced localized tumor that was unresectable due to extensive involvement of surrounding structures or due to involvement of vital structures.

  3. 3.

    Reported on survival between patients undergoing PTR versus PTI. The PTI group included patients who may have received chemotherapy, and/or undergone non-resectional surgery (NRS) such as construction of diverting stoma or bypass procedure without resection of the primary tumor, or received no treatment at all.

  4. 4.

    If two studies from the same institution or database reported the same outcomes of interest, only the most recent publication was included in the analysis, unless the studies were mutually exclusive or the outcome was measured at different time intervals.

Outcomes of Interest

Primary outcome:

  1. 1.

    Overall survival

Secondary outcomes:

  1. 1.

    Cancer-specific survival, progression-free survival.

  2. 2.

    Morbidity reported in detail and as major or minor adverse events. A major adverse event was defined as any event that is life-threatening, requires inpatient hospitalization, results in a single organ or multi-organ failure, or requires operative, endoscopic, or radiological intervention to treat it. Major adverse events correspond to Grade III and Grade IV of the Clavien-Dindo classification, and in cases where the authors did not specifically classify the severity of adverse events, this classification method was followed [53, 54].

Two review authors (CS and EK) independently determined the eligibility of all retrieved studies and extracted the required data from the included studies. The risk of bias of the included studies was assessed based on the following bias risk domains: allocation sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, and vested interest bias [55]. For each of these risk domains of bias, the studies were categorized as high, low, or uncertain risk.

Statistical Analysis

The mean overall survival in months and the proportion of adverse events, with 95% confidence intervals (CI), by treatment strategy was calculated using the random-effects model [56] in OpenMetaAnalyst [57]. Survival between PTR and PTI was compared as a time-to-event outcome in the form of a hazard ratio (HR) or as a difference in duration of survival in months in the form of mean difference (MD). If the HR was not reported in the publications and survival data were presented in the form of Kaplan–Meier curves, the survival rates at specified times were extracted from the Kaplan–Meier curves to reconstruct the HR estimate and its variance, using methods described by Parmar et al. [58]. The generic inverse variance random-effects model was used [56, 59] in RevMan (Review Manager) version 5.3 (The Nordic Cochrane Center, Copenhagen; The Cochrane Collaboration, 2008) [60]. Publication bias was assessed by graphical exploration with funnel plots, with the absence of publication bias indicated by data points forming a symmetric funnel-shaped distribution. Inter-study heterogeneity (HG) was assessed by graphical exploration with forest plots, by using the Chi2 (or χ2) test, by means of the inconsistency index (I2) to quantify HG, and by performing subgroup analyses [61,62,63].

Results

Eligible Studies

A total of 1416 references were identified through electronic searches of Science Citation Index Expanded (n = 435), EMBASE (n = 17), MEDLINE (n = 943), and CENTRAL (n = 21). Further 34 studies were identified from the references of the above studies. The duplicates between databases were 407 and were excluded. Further, 871 clearly irrelevant references were excluded through screening titles and reading abstracts. The 172 remaining studies were investigated in full text detail, and further 95 studies were excluded. Among these excluded studies, eight were excluded because of duplication of all their reported outcomes of interest in other publications from the same institution or database [9, 28, 29, 64,65,66,67,68]. Figure 1 shows the study flow diagram. Seventy-seven comparative studies fulfilled the inclusion criteria of this meta-analysis [2, 4,5,6,7,8, 10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27, 30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50, 69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100]. There were 159,991 patients for analysis, including 94,745 (59.2%) in the PTR group and 65,246 (40.8%) in the PTI group. The characteristics of the included studies are shown in Supplementary Table 2. The risk of bias in the included studies is summarized in Supplementary Figure 1 and the risk of bias for each included study is shown in Supplementary Figure 2. Most included studies were retrospective, and on quality assessment they were found to have high risk of bias in the domains of random sequence generation, allocation concealment, blinding of participants and personnel, and blinding of outcome assessors [55].

Fig. 1
figure 1

Study flow diagram

Full size image

Survival

Table 1 demonstrates the mean overall survival in months by treatment strategy. The calculated mean survival by treatment strategy, in order of increasing survival, was as follows: PTI without chemotherapy 4.02 months (95% confidence interval 2.81–5.23 months), PTR without chemotherapy 7.42 months (3.96–10.87), PTI and chemotherapy 14.30 months (12.56–16.05), PTI and chemotherapy with bevacizumab 17.27 months (15.61–18.94), PTR and chemotherapy 21.52 months (19.82–23.22), PTR and chemotherapy with bevacizumab 27.52 months (21.89–33.14).

Table 1 Mean overall survival in months by treatment strategy in order of increasing overall survival
Full size table

Table 2 shows the results of the overall meta-analysis and subgroup analysis for survival. Supplementary Table 3 includes additional information such as the results of both fixed-effect and random-effects models, and the tests for heterogeneity (I2 and χ2 test P values). Overall analysis revealed that patients treated with PTR had significantly increased overall survival (HR 0.58, P < 0.0001; Supplementary Figure 3) by 7.46 months (MD 7.46 months, P < 0.0001; Supplementary Figure 4) compared to patients that had PTI, with significant heterogeneity identified between studies (HG P < 0.0001). Similarly, PTR resulted in significantly longer cancer-specific survival (HR 0.44, P < 0.0001; MD 10.01, P < 0.0001), and progression-free survival (HR 0.76, P < 0.0001; MD 1.67, P < 0.0001) compared to PTI, with no significant heterogeneity between studies.

Table 2 Results of overall meta-analysis and subgroup analysis for survival
Full size table

Overall survival remained significantly improved after PTR compared to PTI during subgroup analyses of:

  • patients with metastatic (stage IV) disease (HR 0.60, P < 0.0001) by 7.23 months (MD 7.23, P < 0.0001)

  • studies recruiting patients from 2000 onwards (HR 0.62, P < 0.0001; MD 7.29, P < 0.0001)

  • asymptomatic patients (HR 0.69, P = 0.002; MD 3.86, P = 0.002)

  • studies which performed propensity score-matched analysis (HR 0.65, P = 0.003; MD 5.68, P = 0.0003)

  • elderly patients (aged 65 and older) (HR 0.46, P < 0.0001; MD 7.71, P < 0.0001)

  • colon cancer (HR 0.58, P = 0.01; MD 6.31, P = 0.0005)

  • rectal cancer (HR 0.54, P = 0.0009; MD 6.88, P < 0.0001)

  • comparison of patients after PTR with patients after NRS (HR 0.56, P < 0.0001; MD 8.72, P < 0.0001)

  • patients who did not receive chemotherapy (HR 0.63, P = 0.002), although no significant difference in the duration of survival was demonstrated (MD 3.52, P = 0.09)

  • patients receiving chemotherapy (HR 0.59, P < 0.0001; MD 6.81, P < 0.0001)

  • patients receiving bevacizumab (HR 0.59, P = 0.005; MD 10.56, P = 0.01)

Patients who have undergone PTR and chemotherapy had significantly longer overall survival compared to patients undergoing only PTR without chemotherapy (HR 0.54, P < 0.0001; MD 11.46, P < 0.0001). Similarly, patients in the PTI group who received chemotherapy had significantly improved survival compared to patients in the PTI group who did not receive chemotherapy (HR 0.59, P < 0.0001; MD 5.04, P = 0.001).

Adverse Events

Table 3 shows the proportion of adverse events for the patients in the PTI group and the perioperative adverse events in the NRS subgroup. The total morbidity of the patients in the PTI group was 21.7% (14.9–28.4%), and specifically the most common reported adverse events were: obstruction 14.4%, anemia 11.0%, hemorrhage 1.5%, perforation 0.6%, and fistula 0.3%. The proportion of patients in the PTI group requiring surgery due to adverse events was 15.8% (9.0–22.5%).

Table 3 Proportion of adverse events in patients with primary tumor intact (PTI) and for the subgroup of patients who underwent non-resectional surgery (NRS)
Full size table

Patients belonging to the NRS subgroup (i.e., patients with the primary tumor intact but had stoma diversion or bypass surgery) experienced a 30-day mortality rate of 10.6% (6.5–14.7%) and a morbidity rate of 21.7% (13.8–29.6%). In the same group, the major adverse events rate was 7.9% (2.4–13.4%) and 21.7% (16.2–27.2%) for minor adverse events. There was a 0.1% reoperation rate (0–2.4%). The most common perioperative adverse events in the NRS subgroup were respiratory 3.0%, hemorrhage 2.4%, cardiac 2.3%, ileus/bowel obstruction 1.9%, urinary 1.7%, and deep venous thrombosis/pulmonary embolism 1.0%.

Table 4 shows the proportion of perioperative adverse events of the patients in the PTR group. The 30-day mortality rate related to PTR was 4.5% (3.1–5.9%) and morbidity was 22.4% (17.9–26.8%). The major adverse events rate was 10.2% (7.4–13.0%) and 18.5% for minor adverse events (14.1–22.9). The reoperation rate was 2.5% (1.5–3.5%). The most common perioperative adverse events in the PTR group were wound infection 5.7%, ileus/bowel obstruction 4.0%, urinary 3.7%, respiratory 2.9%, intraabdominal collection/sepsis 2.2%, cardiac 1.9%, anastomotic leak 1.6%, hemorrhage 1.1%, wound dehiscence 0.7%, deep venous thrombosis/pulmonary embolism 0.6%, and cerebrovascular accident 0.3%.

Table 4 Proportion of perioperative adverse events in patients who underwent primary tumor resection (PTR)
Full size table

Discussion

This is the largest systematic review and meta-analysis published on this subject to date. Although the overall meta-analysis demonstrated a large effect of PTR in overall survival, due to the large variation in the effect of PTR in survival during subgroup analysis and the associated risk of bias of the included studies, the present review suggested only a limited improvement in survival and that PTR should be considered by the multidisciplinary team on an individual patient basis. Despite published study protocols [3, 101,102,103,104], no randomized controlled trials have been completed to date comparing PTR and PTI in patients with incurable colorectal cancer, and this is partly due to the difficulties in recruiting patients and in designing and performing such trials. Non-randomization of the included studies may have led to patient selection bias and confounders affecting the comparability between the groups at baseline. Concerns were raised that patients who underwent resection of the primary tumor had a more favorable performance status and better overall prognosis in terms of fewer metastatic sites involved, fewer liver-only metastases, and fewer rectal cancer primary tumors [105,106,107]. In addition, some studies included in their PTR groups symptomatic patients who presented with primary tumor-related symptoms or complications at initial presentation [21, 94, 108]. A meta-analysis can address the above limitations by allowing the evaluation of the effects in subsets of patients [109] through the use of prespecified subgroup or sensitivity analyses. In the current meta-analysis, the overall survival remained significantly improved with PTR compared to PTI during subgroup analysis of only patients with metastatic (stage IV) disease, elderly patients aged 65 and older, patients with primary colon cancer or primary rectal cancer, and more recent studies with patients recruited after 2000. To address selection bias, some of the included studies conducted propensity score-matched analysis accounting, among other factors, for number and site of metastases [14, 15, 18, 20, 21, 25, 32, 40, 41], and subgroup analysis of only these studies again demonstrated significant improvement in overall survival with resection of the primary tumor, albeit with a slightly lower survival advantage of 5.68 months. The survival benefit was found to be also reduced in the subgroup analysis of asymptomatic patients (3.86 months), suggesting decreased benefit in this subset of patients, which should be taken into consideration when deciding which patients should proceed with PTR.

With advances in chemotherapy, the response rates of patients with metastatic colorectal cancer to systemic chemotherapy have significantly improved. In contrast to the response rates of approximately 15% to fluorouracil (5-FU) with leucovorin (folinic acid), combinations with other chemotherapeutic agents such as oxaliplatin or irinotecan have yielded response rates of 35–56% [110,111,112,113]. The addition of targeted biotherapy agents targeting vascular endothelial growth factor (anti-VEGF, e.g., bevacizumab) or epidermal growth factor receptor (anti-EGFR in the setting of KRAS wild-type tumors, e.g., cetuximab) to the above combinations have increased response rates further, resulting in significantly improved overall survival and quality of life for patients with incurable metastatic colorectal cancer [111, 112, 114,115,116,117,118]. This meta-analysis has demonstrated that PTR resulted in significantly improved overall survival compared to PTI in the subgroup analysis of only patients that received chemotherapy with increase in overall survival of 7.27 months and especially in the subgroup analysis of only patients that received bevacizumab with increase in survival of 10.56 months. Interestingly, subgroup analysis of patients that did not receive chemotherapy demonstrated only 3.52 months increase in overall survival with PTR compared to PTI without statistical significance (P = 0.09), which suggests that the survival advantage of PTR becomes more pronounced with the use of chemotherapy, and even more so with the use of targeted biotherapy. It is noted that the different chemotherapy regimens and target agents used in combination with PTR or PTI may have a greater impact on oncologic outcomes and patient survival rather than the resection of the primary tumor itself.

Palliative PTR may be required due to adverse events linked to the primary tumor, such as obstruction, perforation or intractable bleeding, but in the setting of current effective chemotherapy regimens, the risk of primary tumor-related complications and the need of subsequent urgent intervention are lower than before. The current meta-analysis identified a 21.7% morbidity rate in the PTI group and a 15.8% risk of emergency surgery due to adverse events. These risks of PTI should be counterweighted against the risks of surgery. It has been suggested that PTR is associated with significant surgical trauma and perioperative mortality and morbidity which may preclude early initiation of systemic therapy [21, 30, 36, 37, 49, 81, 119] or result in a significant systemic inflammatory response and disturbance of homeostasis which may lead to immunosuppression and faster growth of metastases [120]. This study calculated the risk of perioperative mortality to be 4.5% and the risk of morbidity to be 22.4%. Studies which performed multivariate logistic regression analysis to determine independent prognostic variables associated with postoperative mortality and morbidity in patients with stage IV colorectal cancer suggested that baseline characteristics (age, performance status, comorbidity, ASA score), tumor burden (advanced local and metastatic disease), emergency surgery, and primary rectal cancer were related to postoperative morbidity and mortality [49, 105, 119, 121]. An alternative treatment strategy which would prevent complications related to the primary tumor and would theoretically allow the patient to proceed more safely and faster to chemotherapy would be non-resectional surgery (NRS) in the form of a diverting stoma or a bypass procedure. The morbidity related to this treatment strategy was found to be 21.7%. Comparison of the overall survival between PTR and NRS demonstrated that the improvement in survival with PTR remained significant, suggesting that the survival benefit of PTR is not only through the prevention of complications related to the primary tumor.

Although it is not clear why PTR is associated with better outcomes in patients with incurable colorectal cancer, the improvement in overall survival and especially cancer-specific survival following PTR may be attributed to a better response to chemotherapy after reduction of systemic tumor burden. This may explain the improved survival of PTR when combined with chemotherapy. Similar survival benefit has been demonstrated by resecting primary renal and ovarian tumors in the presence of metastatic disease [122,123,124]. Also, based on the ‘seed and soil theory’ which is used to explain the metastatic preference of cancer cells for specific organs, the primary tumor may induce in distant organs a prosperous environment to enhance the growth of metastatic deposits and progression from micro- to macrometastases [125, 126]. Van der Wal et al. suggested that in the presence of the primary tumor, the liver parenchyma adjacent to synchronous liver metastases provides an angiogenic prosperous environment for metastatic tumor growth [127]. Furthermore, Holzel et al. suggested that all distant metastases are initiated before removal of the primary tumor and that metastases do not metastasize again [128]. Based on the results of the current study, every patient with incurable colorectal cancer should be considered for resection of the primary tumor. Nevertheless, not every patient will be a candidate for surgery, and among patients, the benefits of surgery will be different. Studies which performed multivariate analysis to determine which factors were associated with survival in patients with unresectable colorectal cancer, in addition to PTR, identified the following independent prognostic variables: primary tumor differentiation grade [4, 5, 11, 21,22,23, 25, 28, 29, 48], number of metastatic sites [2, 12, 14, 18, 21, 22, 25, 47, 66], primary tumor location [4, 6, 12, 22, 23, 25, 28, 66], extent of metastatic liver involvement [32, 48, 49, 75, 81, 89, 94], administration of chemotherapy [4, 14, 20, 24, 32, 39, 48, 49, 93], administration of anti-VEGF therapy [2, 13, 21, 23], CEA levels [5, 12,13,14, 28, 32], age [4, 6, 18, 22, 25, 28, 29], Eastern Cooperative Oncology Group performance status (ECOG-PS) [4, 14, 38,39,40], World Health Organization physiology score (WHO-PS) [2, 12, 49, 66], American Society of Anesthesiology (ASA) score [22, 93], primary tumor N-stage [2, 6, 18, 23, 25, 26], T-stage [25, 49], peritoneal dissemination [50, 83], adjacent organ invasion [20, 32], ascites [83], white blood cell count or neutrophil count [2, 4, 12], and levels of hemoglobin [2, 4], alkaline phosphatase [4, 66], aspartate aminotransferase [5], bilirubin [4], lactate dehydrogenase [38], and serum albumin [4]. The present review has quantified the duration of survival of each treatment strategy, quantified the survival benefit of PTR in different subgroups of patients, and has quantified the risk of morbidity of each individual treatment strategy, to hopefully assist the discussion within the multidisciplinary team on an individualized patient basis, as well as with the patient, to allow for an informed decision to be made.