Ninety percent of the adverse outcomes occur in 10% of patients: can we identify the populations at high risk of developing peritoneal metastases after curative surgery for colorectal cancer?

Abstract

Background: Peritoneal metastases (PM) occur in 3.4–6.3% after curative surgery for non-metastatic colorectal cancer. Systematic “2nd look” surgery helps overcoming the diagnostic problem but can be only proposed to selected patients. The aim of this study was to update the knowledge on risk factors of developing PM after curative surgery for colorectal cancer.

Methods: A systematic review of the literature published between 2011 and 2016 was made, searching for all clinical studies reporting the incidence of recurrent PM after curative surgery for colorectal cancer and factors associated with the primary tumour that were likely to influence this recurrence rate.

Results: Seven new clinical studies were considered informative for risk factors and added to the 16 reviewed in 2013. Even if the level of evidence was low, data suggested rates of recurrent PM at 1 year between 54% and 71% after completely resected synchronous PM, between 62% and 71% after resection of isolated synchronous ovarian metastases, of 27% after surgery for a perforated primary tumour, of 16% after surgery for a pT4 tumour, and between 11% and 36% after surgery for a mucinous histological subtype. No new risk factor was identified.

Conclusions: Evidence regarding the incidence of recurrent PM after curative surgery for colorectal cancer is poor. Situations at higher risk of recurrent PM are synchronous PM, synchronous isolated ovarian metastases, perforated primary tumour with serosa invasion and mucinous histological subtype.

Keywords:

• Risk factors
• peritoneal metastases
• colorectal cancer
• HIPEC

Introduction

Peritoneal metastases (PM) occur in 3.4–6.3% of patients after curative surgery for non-metastatic colorectal cancer [1–7]. Patients with PM have a spontaneous poor prognosis and median survival is 22 months at best with the latest drug regimens [7–9]. In comparison, a more aggressive approach combining complete cytoreductive surgery (CCRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) enables a 5-year overall survival rate of 42% and a median survival of 41 months [10]. One of the most important prognostic factors after CCRS plus HIPEC is the extent of the peritoneal disease measured with the peritoneal cancer index (PCI) [9,10]. Unfortunately, a major diagnostic problem remains because PM are asymptomatic at an early stage and all non-invasive imaging have disappointing detection rate in nodules smaller than 5 mm [11]. To overcome this problem, the concept of “2nd look” surgery plus HIPEC was developed, enabling early detection of PM [11]. Nevertheless, this invasive approach can only be proposed to selected patients. After a systematic review of the literature published in 2013, an evidence-based categorisation was proposed with a group of patient deemed at high risk of PM (patients with synchronous PM or patients with ovarian metastases both completely resected with the primary tumour and patients with perforated primary tumour) [12]. Since that publication, major studies have been reported on the subject, allowing a more precise evaluation of patient at risk of PM.

The aim of this systematic review was to update the knowledge on risk factors of developing PM after curative surgery for colorectal cancer.

Methods

Search strategy

All retrospective and prospective studies explicitly describing the incidence of recurrent PM after curative resection of colorectal cancer since the publication of systematic review of the literature published in 2013 were eligible for inclusion in this systematic review [12]. All case reports were excluded from the analysis.

Study inclusion and data extraction

A systematic review of the literature was conducted by one independent reviewer using PubMed, MEDLINE, Embase, Ovid, the Cochrane database, and Google Scholar Search to identify all clinical studies published in English between 2011 and 2017 reporting the incidence of recurrent PM. Combinations of the following search terms were used to identify the studies: “colorectal cancer”, “colon cancer”, “colon carcinoma”, “peritoneal metastases”, “peritoneal recurrence”, “peritoneal carcinomatosis”, “ovarian metastasis”, “obstructive cancer”, “mucinous adenocarcinoma”, “signet ring cell adenocarcinoma”, “serosal invasion”, “pattern of recurrent colorectal cancer”, “peritoneal cytology”, and “lavage”. A 3-step eliminative selection was applied (a title-based selection, followed by an abstract-based selection, and followed by a full text selection). Studies were excluded if they included non-colorectal cancer, if the incidence of recurrent PM were not specified, if patients having undergone a curative resection could not be isolated and if the patients had received an intraperitoneal chemotherapy. All studies reporting disease-free and overall survival without explicit recurrent PM rate were not selected for the analysis.

Data analysis and quality of included studies

Once studies identified, the incidence of recurrent PM was categorised according to associated level of risk. The studies were graded from 1 to 5 in terms of the methodological quality following the Oxford Centre for Evidence-based Medicine levels of Evidence, according to the number of patients, the randomised or non-randomised design and whether they were prospective or retrospective (Table 1) [13].

Table 1. Oxford centre for evidence-based medicine – levels of evidence (March 2009).

Level
1a	SR (with homogeneity^a) of RCTs
1b	Individual RCT
1c	All or none^b
2a	SR (with homogeneity^a) of cohort studies
2b	Individual cohort study (including low quality RCT)
2c	“Outcomes” Research
3a	SR (with homogeneity^a) of case-control studies
3b	Individual Case-Control Study
4	Case-series (and poor quality cohort and case-control studies)
5	Expert opinion without explicit critical appraisal

SR: systematic review; RCT: randomised controlled trial.

^a Homogeneity means being free of worrisome variations.

^b Met when all patients died before the treatment became available, but some now survive or when some patients died before the treatment became available, but none now die.

Results

The subsequent search strategy identified 259 new articles to add to the 6522 articles initially analysed. The majority of the studies were excluded and seven studies were eventually considered informative concerning the risk for developing PM and added to the 16 studies reported in the previous review [12,14–20]. All 23 studies were non-randomised. Six were prospective and 17 were retrospective. The total number of patients was 598 and six factors associated with recurrent PM were analysable: synchronous PM completely resected during the primary tumour surgery, synchronous ovarian metastases, a perforated primary tumour, the histological mucinous subtype, positive peritoneal cytology, and serosa invasion (pT4).

Synchronous peritoneal metastases (completely resected during the primary tumour surgery)

Unexpected synchronous PM are discovered peroperatively in 4.8–7.3% of patients with colorectal cancer scheduled for surgery [12]. Five non-randomised studies were eventually available for PM analysis. In the three retrospective studies, the reported recurrent PM rate was ranging between 29% and 75% but there was a strong heterogeneity among diagnostic methods and follow-up schemes [12,14,16]. The two available prospective studies, evaluating the development of PM with a systematic “2nd look” 12 months after the initial surgery, reported rates of recurrent PM of 54% and 71% [11,20]. The median PCI were 8 and 10 at the time of the systematic “2nd look” surgery.

In conclusion, recurrent PM rate after completely resected synchronous PM during the primary tumour surgery is estimated between 54% and 71% at 1 year, with a high level of evidence if we exclude lower quality retrospective studies (2b).

Synchronous ovarian metastasis

Unexpected synchronous ovarian metastases are synchronously found in 0.8–7.4% of all patients with colorectal cancers but are associated with synchronous PM in 29–72% of the cases [12]. Three retrospective studies including a total of 35 patients after resection of isolated ovarian metastases (without associated PM) reported recurrent PM rates of 8–56%. In the two prospective series reporting a total of 35 patients evaluated with “2nd look” surgery after 12 months, the rates of recurrent PM were 62% and 71% [11,20].

In conclusion, the rate of recurrent PM in patients with isolated synchronous ovarian metastases is estimated between 62% and 71% at 1 year with a high level of evidence (2b).

Perforated primary tumour

The incidence of colorectal cancer presenting with a perforated primary tumour is ranging from 1.6% to 5.4% [21–23]. This incidence is probably overestimated because perforations above the occlusive tumour are frequently mixed with truly perforated primary tumours. Three retrospective studies including a total of 134 patients with a perforated colorectal cancer reported rates of recurrent PM of 19–54% [12]. In the only prospective study available, PM were diagnosed in 27% (3/11) of patients after a systematic “2nd look” surgery 12 months after the initial surgery [11]. No data are available on iatrogenous perforation.

In conclusion, the incidence of recurrent PM after surgery for a perforated primary tumour is 27% at one year with a low level of evidence (3b), the impossibility of excluding perforations above the obstacle from the analysis inducing a potential risk of underestimation.

Primary tumour with serosa and/or adjacent organ invasion

Serosa and/or adjacent organ invasion is found in 13.7% of all colorectal cancer [24]. Four non-randomised retrospective studies including total of 671 patients reported rates of recurrent PM of 8–19% [12]. In non-metastatic patients, full-thickness invasion through the serosa (pT4a) seemed more strongly associated with PM compared to overtly invaded adjacent organs (pT4b). Even if we have no clear explanation for that, the required “en-bloc” resection with neighbouring organs in pT4b tumour could have improved the resection margins [12]. One recent prospective study focussing on a population of 173 patients reported recurrent PM rates of 4.5% at 1 year and 9% at 3 years in pT3 tumours vs. 15.6% at 1 year and 36.7% at 3 years in pT4 tumours [15].

In conclusion, the incidence of recurrent PM after surgery for a pT4 tumour is 15.6% at 1 year and 36.7% at 3 years with a low level of evidence (3a).

Mucinous histological subtype

Mucinous adenocarcinoma (MA) and signet ring cell carcinoma (SRC) account, respectively, for 3–15% and for 0.1–2.4% of all colorectal cancers [12]. Synchronous PM at the time of primary surgery are frequent with these histological subtypes, in 11–75% of MA and in 71% of SRC [12]. When focussing on non-metastatic patients after curative resection, one retrospective study including 127 patients reported a recurrent PM rate of 36% in MA, of 11% in adenocarcinomas with a mucinous component and of 27% in adenocarcinomas with a signet ring cell component. No data were available on PM rate after surgery for SRC. In a recent study, 22% of patients operated for a pT3/pT4 MA developed PM after a median delay of 12.7 months [18].

In conclusion, recurrent PM in patients with MA or with a mucinous/SRC component is estimated between 11% and 36% with a low level of evidence (3b).

Positive peritoneal cytology

The incidence of positive peritoneal cytology, without distinguishing tumour stages and detecting methods, ranges between 2.2% and 52% [12]. Six non-randomised prospective (2) and retrospective (4) studies including total of 73 patients with a positive peritoneal cytology reported rates of recurrent PM ranging from 0% to 36% [12]. After 2013, a large prospective study including 963 patients with a colorectal cancer who had preoperative cytology was reported [17]. The positivity rate of the peritoneal lavage in this large study was very low, 5.7% in colon cancer and 0.6% in rectal cancer [17]. Comparable PM rates were found between patients with or without positive peritoneal cytology [17].

In conclusion, the incidence of recurrent PM after surgery for colorectal cancer with positive peritoneal cytology is estimated between 0% and 36% with a very poor level of evidence (3b–4).

Prediction score

This topic was explored recently in a nation-wide population-based study including 11 124 patients with a non-metastatic colorectal cancer to determine the incidence, prevalence, and risk factors for recurrent PM [19]. In this large registry, recurrent PM were detected in 4.2% of patients after a median delay of 16 months. Independent predictive factors for developing recurrent PM were a colon primary tumour, particularly right-sided (HR 1.77), a pT3 tumour (HR 3.82) or pT4 tumour (HR 9.98), a N1 (HR 7.41) or N2 (HR 4.66) lymph node status, emergency procedures (HR 2.11), and non-radical resection of the primary tumour (R1 HR 1.96 or R2 HR 2.75). These findings were included in a statistical analysis to build a score predicting on an individual basis 1 year and 3 year risk of recurrent PM [19]. This prediction score was subsequently validated and is now available online at http://www.imm.ki.se/biostatistics/calculators/pcrisk [19].

Other risk factors

We failed to find new information on the ruled-out “risk factors” of the systematic review of 2013 (lymph node invasion, laparoscopic resection, occlusive tumour, bleeding tumour, and tumour location) or on other potential “risk factors” (peroperative bleeding, postoperative complications, tumour biology, radicality, biomarkers, etc.) [12].

Discussion

Risk categorisation in 2017

The update of the 2012 systematic review found no new risk factor of recurrent PM. Nevertheless, we were able to determine more precisely the level of risk bound with each known factor. The main diagnostic bias of this new review remains the poor reliability of imaging in detecting PM. Nevertheless, new prospective studies, including a systematic “2nd look” surgery series published after 2013 gave a much stronger level of evidence compared to the earliest review [20]. Using these available evidence-based data and considering the quality of the studies, we were able to categorise patients’ risk as follows.

Very high-risk patients

Patients with isolated synchronous ovarian metastases and with synchronous PM and with an intraperitoneally perforated primary tumour run a risk of developing recurrent PM of 27–71% at 1 year [12,20]. In the first two categories, we can safely say that we are not dealing anymore with a situation of risk but in an established state of PM. Nevertheless, for the sake of further studies, we left them in the risk categorisation but we upgraded this level of risk from high to very high.

High-risk patients

Patients with tumour invading the serosa and beyond (pT4) or with a mucinous histological subtype run a risk of developing recurrent PM of 16–22% at 1 year [12,15,18]. The main drawback before proposing an aggressive proactive treatment to these patients is the ability to identify them preoperatively or peroperatively to avoid any overtreatment. In the retrospective series of Sammartino et al. [18], five patients with a suspected T3–4 tumour had HIPEC but had in reality a pT2N0 tumour on the definitive pathologic analysis.

Standard risk patients

No evidence of an increased incidence of recurrent PM was found in patient with bleeding/occlusive primary tumours, with invaded lymph node or after laparoscopic resection [12]. The running risk of PM is evaluated in this category between 3.4% and 6.3% [1–7,12].

Inconclusive data and prediction score

The influence of positive peritoneal cytology on recurrent PM rate is still debated in patients with a colorectal cancer. The efficiency of the procedure itself is very poor. In the largest prospective study, prevalence of positive peritoneal cytology was only 5.7% in colon cancer and 0.6% in rectal cancer, inducing a power bias to find any relationship with recurrent PM [17]. The rarity of this event combined with a strong heterogeneity in sampling techniques makes it too unreliable to be taken in account when deciding a morbid treatment like HIPEC. Help can be found in other predicting tools, such as the Segelman Prediction Score [19] Factors like age, tumour site (left/transverse/right colon, rectum), pT stage, pN stage, number of examined lymph nodes, radicality (R0/R1/R2), elective or emergent surgery, preoperative radiotherapy, and adjuvant chemotherapy, of 8044 patients with a stage I/II/III colorectal cancer were computed to a build a monogram predicting recurrent PM rate at 1, 2, and 3 year [19]. The given risk within this score is running between 0.4% and 43.6% at 1 year.

“2nd look” or prophylactic HIPEC in very high-risk patients?

Based on these results, the French Network for Rare Peritoneal Malignancies (RENAPE) initiated a multicentric prospective randomised phase III trial called “ProphyloCHIP” (NCT 01226394) comparing a systematic “2nd look” surgery plus HIPEC vs. non-invasive follow-up after 6 months of systemic chemotherapy in patients with isolated synchronous ovarian metastases or with synchronous PM or with a intraperitoneally perforated primary tumour. The 3-year disease-free survival is the primary outcome measure and preliminary data can be expected in the middle of 2017. Another similar study was initiated in the United States but suffered an early withdrawal, without published data (NCT01095523) [25]. Considering the level of risk in this population, a more proactive management could be recommended. With the potential survival benefit of “2nd look” surgery plus HIPEC reported in two prospective studies, innovative strategies aimed at decreasing the risk of recurrent PM through prophylactic HIPEC in very high-risk patients may be the logical evolution [11,20]. A randomised phase III trial (NCT02179489) is actually ongoing in China, comparing the realisation of upfront adjuvant HIPEC with Mitomycin C and surveillance alone in patients with resectable minimal PM or with ovarian metastases or with a pT4 tumour (confirmed peroperatively with a pathological analysis) or with an intra-abdominal primary tumour rupture. The primary outcome measure is the 3-year disease-free survival. Both strategies have their pro and con. In favour of upfront prophylactic HIPEC during the resection of the primary tumour when confronted with a very high-risk situation is the possibility of a one-stage full procedure. Unfortunately, performing upfront HIPEC is not recommended if the risk of PM is identified peroperatively, giving no possibility for the patient to receive adequate information and give his/her consent. Another complicated situation is encountered when HIPEC is not directly available in the operating facility. For these reasons, both approaches (“2nd look” plus HIPEC and prophylactic HIPEC) cannot be considered mutually exclusive and must be incorporated in a unique strategy to prevent recurrent PM in very high-risk patients, giving the surgeon the possibility to choose which strategy could best fit to the situation.

“2nd look” or prophylactic HIPEC in high-risk patients?

Based on the recurrent PM rate of 15.6% at 1 year in pT4 tumours, extending the indications of 2nd look plus HIPEC or prophylactic HIPEC could be discussed in this subset of patients [15]. Two retrospective non-randomised studies are available [18,26]. Tentes et al. [26] performed HIPEC in 40 patients with a T3/T4 tumour without synchronous PM and reported recurrent PM rate of 0% after 17 months of median follow-up. The other study reported 25 patients with a mucinous T3/T4 tumour without synchronous PM who underwent HIPEC [18]. These patients were compared with 50 consecutive patients treated with standard surgical resection during the same period in the same hospital. The early postoperative results were not statistically different between the two groups but the recurrent PM rate was only 4% after 48 months of median follow-up in patients with HIPEC vs. 28% in patients without HIPEC. Even if both populations were not comparable, the peritoneal recurrence rate was significantly lower in the group that received HIPEC (4% vs. 22%), with similar overall recurrence rates (28% vs. 32%). The authors concluded in a potential increase in local control. This topic is now directly questioned in three prospective randomised phase III trial [27]. The COLOPEC trial (NCT02231086) coordinated by the Dutch Colorectal Cancer Group started in 2015. The aim of this study was to evaluate the efficacy of early adjuvant HIPEC using oxaliplatin, delivered simultaneously or within 10 d after curative resection of the primary tumour either T4N0–2M0 or intra-abdominally perforated. If adjuvant HIPEC could not be performed within 10 d the procedure, an extra postoperative delay of 5–8 weeks after surgery was allowed before HIPEC. The primary endpoint was peritoneal recurrence-free survival at 18 months. Registered but not open yet, the “PROMENADE” trial (NCT02974556) will evaluate prophylactic HIPEC with oxaliplatin after a curative resection of high-risk T3 (≥ 5 mm tumour invasion beyond the muscularis propria) and T4 any N, M0 colon cancer. Until we have the results of these studies, we cannot to recommend extending the indications of “2nd look” surgery plus HIPEC or prophylactic HIPEC. One exception could be made. The authors of the Segelman Prediction Score suggested prophylactic HIPEC could be discussed in patients with a 30% or higher PM rate using their score [19]. This score was recently validated with an external and recent cohort as a strong model to be used to identify high-risk patients for planned subsequent treatment [28]. We must nevertheless remain very cautious when proposing aggressive treatment like HIPEC outside a clinical trial. In the two retrospective study of adjuvant HIPEC, one patient out of 65 died postoperatively. These situations must carefully be discussed and balanced in expert multidisciplinary team meeting.

Prophylactic HIPEC and “2nd look” in non-colorectal cancer

Other histological subtypes could be eligible for proactive treatment, knowing the associated risk factors [29,30]. Recurrent PM is occurring in 29–46% of patients after curative surgery for gastric cancer [30]. Risk factors are tumour invading the serosa (pT3/pT4), tumour with invaded lymph nodes (N1/N2), positive peritoneal cytology, and signet ring cell (SRC) histological subtype [30]. No “2nd look” study was ever published but considering the results of intraperitoneal chemotherapy in gastric cancer, seven randomised controlled trials are actually including patients in locally advanced non-metastatic high risk patient (NCT01882933, NCT02381847, NCT02356276, NCT01683864, NCT02960061, NCT02240524, and NCT02969122). No results are yet available. The linkage between appendiceal mucinous neoplasm (MN) and pseudomyxoma peritonei (PMP) is known. In a retrospective study on 25 patients with a MN without PM (either macroscopic or microscopic, i.e. without PMP), perforated MN was associated with a 65% PMP rate after a median follow-up of 50 months compared with 17% without perforation. This could lead to either a prophylactic HIPEC or a “2nd look” surgery to these selected patients. Other risk factor have been identified in other histological subtype, like the loss of heterozygosity in chromosome 18 in neuroendocrine tumour or tumour rupture in abdominal soft tissue sarcoma, but we have no sufficiently efficient locoregional treatment to discuss any aggressive proactive management [31].

Other means of PM prophylaxis in colorectal cancer

No other treatment than HIPEC has ever been used in prophylaxis in colorectal cancer. Even if chemotherapy given intravenously fails to prevent 54–71% of PM in very high-risk patient, other ways of delivery have been described lately. As intra-arterial hepatic chemotherapy has been used in colorectal cancer liver metastases, intraperitoneal chemotherapy could be an option, as it has already been used in gastric cancer with PM [32,33]. The only drawback is the potential no-treatment zone in the peritoneal cavity because of early postoperative adhesion, if used postoperatively. A solution could be to identify preoperatively the candidates for prophylaxis and treat the peritoneum before surgery. Pressurized intraperitoneal aerosol chemotherapy could be an interesting alternative carrier for such preoperative locoregional chemotherapy [34].

Conclusions

Evidence regarding the incidence of recurrent PM after curative surgery for colorectal cancer is still poor but nevertheless increased since 2013. Situations at higher risk of recurrent PM are synchronous PM, synchronous isolated ovarian metastases, perforated primary tumour, tumour with serosa invasion, and mucinous histological subtype, running a risk from 11% to 71% at 1 year. Further studies are now required to evaluate effective means to prevent PM occurrence or to diagnose PM at an early stage.

Disclosure statement

All authors disclose any financial and personal relationships with other people or organisations that could inappropriately influence or bias this work.