Metastasis to the rectum: A systematic review of the literature

Background: Metastatic spread to the rectum is a rare ﬁ nding, and management of rectal metastases (RM) is not standardized. The aim of the present study was to review the evidence on diagnosis, management and outcomes of RM. Methods: A computerized literature search through MEDLINE/PubMed, Embase and the Cochrane databases was performed, applying a combination of terms related to RM. Articles and abstracts were screened and ﬁ nal selection was done after cross-referencing and by use of prede ﬁ ned eligibility criteria. Results: Final analysis was based on 99 publications totaling 162 patients with RM from 16 different primary tumors. Most common origins of RM were breast (42 patients), stomach (38 patients), and prostate (16 patients). RM occurred metachronously in the majority of patients (77%). The main treat-ment was surgical resection (n ¼ 32), followed by chemotherapy (n ¼ 16). Median overall survival for breast RM, stomach RM, and prostate RM were 24 months (95% CI 9-39 months), 7 months (95% CI 0-14 months), and 24 months (95% CI 7-41 months), respectively. Conclusion: RM is a rare and highly heterogeneous condition. Surgical treatment appears to be a valuable treatment option in selected patients, while overall prognosis depends mainly on the primary tumor. © 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Metastases to the gastrointestinal tract are rare.Overall, incidence of metastases to the upper and lower gastrointestinal tract is 0.03% and 0.05% of all metastases, respectively [1].In the limited literature, gastrointestinal metastases are treated mostly as one group.However, management, treatment and prognosis vary considerably according to the metastatic site and the underlying primary tumor.
Metastases to the rectum are particularly rare and its true incidence is unknown.Tumoral invasion of the rectum by a metastasis carries specific risks, such as obstruction or perforation [2,3].Moreover, due to the anatomical position of the rectum and its proximity to other structures (prostate, bladder, uterus), rectal metastases (RM) can induce various symptoms, be non-resectable, or create functional problems.Lastly, surgical resection is more complex than other intestinal resections and stoma creation might be required.The specificity and clinical implications of RM underline the relevance and interest of a comprehensive review of the available evidence on this uncommon pathology.
The aim of the present study was to systematically review the literature on RM from various etiologies, in terms of diagnosis, treatment and outcomes.

Search
A computerized literature search was performed through MEDLINE/PubMed, Embase and the Cochrane databases, applying the following terms: "neoplasm metastasis" (MeSH term), "rectum" (MeSH term), "rectal metastasis" (free text), "rectal metastases" (free text), and "secondary neoplasm" (free text).These terms were combined using "AND" and "OR".Additional articles and abstracts were identified by cross-referencing.Only articles published in English language from implementation of above-cited databases until December 2020 were considered for the present review.

Eligibility criteria
Hematogenous, lymphogenic or seeding RM (distant metastases) were considered whereas direct tumoral invasion into the rectum was an exclusion criterion.Only full-text articles were included.Letters to the editor, posters, and clinical images were excluded.

Data extraction
The following data were extracted: sex, age, RM clinical presentation, RM diagnostic procedures, site of primary tumor, histology and treatment of primary tumor, synchronous or metachronous RM (if metachronous, time between primary tumor diagnosis and RM diagnosis), presence of other metastases, RM treatment, overall survival from RM discovery, and specific markers.

Statistics
Continuous variables were presented as median with interquartile range and categorical variables as proportion and percentage.The Kaplan-Meier method was used to calculate the overall survival (OS).OS was reported as median with 95% confidence interval (CI).SPSS Statistics for Mac OS X, version 25 (IBM Corp., Armonk, NY, USA) was used for all statistical analyses.
As no direct patient data were used, approval of the ethics commission was waived.This systematic review was conducted and reported following the PRISMA guidelines.

Overall results
A total of 722 publications was identified and after duplicate removal, 99 articles were finally included totaling 162 patients with RM (70 women, 50 men and 42 cases with no gender precision) .Summary of the search is shown in the PRISMA flowchart (Fig. 1).Median age was 65 years (IQR .Breast cancer was the most common origin of RM (42 patients), followed by gastric cancer (38 patients), prostate cancer (16 patients), and melanoma (14 patients).Results are presented below according to the origin of the primary tumor.
The main treatment performed was resection surgery (n ¼ 32, abdominoperineal resections, anterior low resections, Hartmann's operations, or total pelvic exenterations), sometimes combined with stoma (n ¼ 8), with (n ¼ 15) or without (n ¼ 25) neoadjuvant or adjuvant therapy.Chemotherapy (ChT) alone was the second most used treatment (n ¼ 16).Stoma alone was performed in 9 patients, as well as in 5 patients with neoadjuvant or adjuvant therapy.Table 1 summarizes the number of patients, the resection rates and the survival ranges based on the origin of the primary tumor.Survival curves of patients with RM from urogenital, gynecologic, and gastrointestinal origins are shown in Fig. 2.

Urogenital cancers (n ¼ 38)
Median age of the 16 patients with prostate cancer was 73 years (IQR 69-85) [2e10].Six patients had other metastases in addition to RM. Median OS was 24 months (95% CI 7-41).Three of 11 patients did not receive any therapy for the RM, whereas the rest underwent   [16,24,59e63].Two patients did not have any other metastases at the time of RM occurrence and 2 had peritoneal carcinosis.Regarding endometrium cancer, both patients underwent surgery completed with RT for the primary cancer and for the RM (Hartmann's procedure and low anterior resection).One patient died 11 month later and the other one was still alive 5 years after RM diagnosis.Findings regarding gynecologic cancers are summarized in Table 3 and Supplementary Table 1.

Melanoma (n ¼ 15)
RM treatment was specified in only 4 out of 15 patients [17,64e70].Three had abdominoperineal resection and 1 had argon plasma coagulation and immunotherapy.As a particularity, it is the only group where anemia was described as clinical presentation of RM.Supplementary Table 2 summarizes the findings of included studies regarding melanomas.

Pancreas and liver cancers (n ¼ 5)
Median age of the 5 patients was 68 (IQR  [71e75].In 3 patients, RM were discovered synchronously as the primary cancer.Only 2 patients had data on OS (alive at 12 months and dead at 19 months).Summary of the data found for pancreas and liver cancer   RM are shown in Supplementary Table 3.

Lung cancer (n ¼ 10)
Median age was 62 years (IQR 41-81) and 80% of the patients were male [92e97].Survival after RM occurrence ranged from 5 weeks to 20 months.Supplementary Table 4 summarizes the findings of included studies regarding lung cancer.

Others (nasopharynx and femoral bone cancers) (n ¼ 3)
OS of patients with nasopharynx RM appeared to be very short [98,99].One of the patients underwent an unspecified surgical treatment and died 15 days later whereas the other one did not receive any treatment and survived 1 month.Hayasaka et al. described the only case of RM from fibrous histiocytoma of the femoral bone found in the literature [100].The patient did not have other metastases and underwent tumorectomy.Survival report was not specified.Supplementary Table 5 summarizes the findings of included studies regarding nasopharynx and femoral bone cancers.

Discussion
RM is a rare condition with challenging management.Treatment and prognosis mainly depend on the underlying primary and global disease extent.Surgical resection entails favorable outcomes in selected patients.
Regarding gastrointestinal metastases, symptoms are generally aspecific (bleeding, abdominal pain) or even absent.However, gastrointestinal bleeding is most often present as the majority of gastrointestinal metastases are located in the stomach and duodenum [16].The best diagnostic method remains endoscopy with biopsy [16].Even though gastrointestinal metastases are often associated with an advanced stage of the disease, aggressive treatment with chemotherapy and/or surgery allow better results in terms of symptom relief and survival prolongation than conservative palliative treatment [16].Overall prognosis is poor but depends on the patient condition, the origin of the primary tumor and the presence of other metastases.
Decision to perform surgery for RM should consider several important factors: patient desire, presence of symptoms decreasing the quality of life, life expectancy, treatment alternative, and origin of the primary tumor.Resection surgery should be performed to improve the quality of life and prevent complications in patients with a good overall health status with at least several months of life expectancy.
As RM is a rare phenomenon, its precise incidence is unknown.Patient reported outcomes, quality of life after surgery (after resection or stoma), complications of RM treatment are important elements that are not described in the literature.
Surgery is currently recognized as the main treatment in gastrointestinal metastases from melanoma, showing consistently better survival rates than systemic therapies, even for patients with disseminated diseases [64,68e70].This was described in studies with outcomes calculated for any types of gastrointestinal metastases (small and large intestine, rectum and anus).Nonetheless, the most frequently observed metastasis site from melanoma in these studies was by far the small intestine (67-91%) [64,68].Caution should be taken to apply such conclusions to RM from melanoma, as they only represented a small number of cases (2-4%).Moreover, taking into account advances and development in immunotherapy, this approach has to be considered as a potential treatment strategy for melanoma RM [101,102].Regarding RM from lung cancer, Hu et al. reported similar results showing that absence of surgery appeared as a negative prognostic factor for survival outcome.[15] Again, RM represented a minor proportion (1.4%) in this study and outcomes were calculated by including all secondary cancer sites (duodenum, jejunum, ileum, appendix, caecum, colon, and rectum).Outcomes of metastases treated with small bowel resections, esophagectomies or abdominoperineal resections cannot be compared due to the differences of pathophysiology and surgery types.
Regarding immunohistochemical markers, it was observed that 85% (23/27) of RM from breast tumors were positive for the estrogen receptor (ER) while HER-2 was negative in 87% (13/15) of the cases.Recently, several articles demonstrated that ER-negative tumors had a higher proportion of visceral metastases, and that ERpositive tumors tended to first metastasize to bone [103e107].The results of this review showed an opposite pattern regarding ER status, as a significant majority of the tumors were positive for this receptor and only 15% of the women (5/32) had bone metastases.It has also been suggested that ER-positive tumors tend to relapse later, whereas secondary events are more frequently observed in the first 5-7 years among ER-negative cancers [106,108].In the present review, a median time of 6.5 years between primary and secondary cancer for the ER-positive tumors was found, which is shorter than usual time observed in the literature for this subgroup.Those findings may suggest the existence of a cluster group among ER-positive tumors which tends to metastasize sooner and more frequently to the rectum.These results should nevertheless be taken with caution because of the risk of publication bias.
Certain types of RM seem to have a better prognosis than others in terms of OS after occurrence of RM.In this review, among the articles where survival was mentioned, 2/4 patients with ovarian cancer RM survived >2 years and 14/27 patients with breast cancer RM survived >1 year after RM occurrence.On the contrary, only 2/ 10 patients with RM from gastric origin survived over a year.Taking aside the performed treatment, this suggests that RM from gastrointestinal origin might have a worse prognosis than RM from gynecological origin (Fig. 2).
Several limitations of the present review need to be mentioned.As articles on RM from different primary cancers (16 types) were collected, generalization for all RM is not possible.Another factor was the heterogeneity and variable quality of the selected articles.Some articles were fully complete in terms of patient follow-up, clinical and technical data, while others were missing some critical information.As all included articles were case reports or case series, quality assessment was not performed.Finally, heterogeneity of the data precluded any meta-analyses.
In conclusion, literature on RM remains scarce and is of poor quality.Surgical resection for RM appears to be indicated in selected patients and goes along with favorable outcomes.However, indication mainly depends on the underlying primary and global disease extent and requires careful evaluation by a multidisciplinary tumor board.RM from breast or ovarian origins might have a better OS than RM from gastrointestinal origins, regardless of undertaken treatments.

Table 1
Summary table based on the origins of the rectal metastases.
a Only one patient had data on survival time.

Table 2
Studies reporting rectal metastases (RM) from urogenital cancers.

Table 4
Studies reporting rectal metastases (RM) from gastrointestinal cancers.