Tumor deposits are associated with a higher risk of peritoneal disease in non‐metastatic colorectal cancer patients

Tumor deposit (TD) is a poor prognostic factor in colorectal cancer (CRC) patients. This study aimed to determine whether TD carry the same risk of peritoneal recurrence as known high‐risk (HR) features in CRC patients.

(AJCC) 7th edition, published in 2009, they were redefined and considered a separate entity. 3 In the absence of LNM (pN0), TDs are assigned an N1c designation-upgrading the patient to stage III at minimum. CRC patients of this stage are advised to undergo adjuvant therapy. 4 The AJCC 8th edition (2017) further clarifies that TDs should not have any identifiable lymph node, vascular, or neural tissue on histopathological examination. 4 Regardless of how TDs are classified, the literature identifies them as an independent, poor prognostic factor for increased likelihood of metastasis, and worse survival. 5,6 In the presence of LNM, TDs are ignored under the current classification system. This means that a patient categorized as N1a or N1b will not be upgraded if found to have TDs. Herein lies the limitations of the current system because TD are not equivalent or inferior to LNM in a prognostic sense. In fact, Bouquot et al. reported that N1c CRC patients had a worse 3-year overall survival compared to N1a or N1b. 7 However, according to the current AJCC 8th edition staging system, N1a, N1b, and N1c are all equal prognostic groups. 4 To preserve valuable prognostic information that TDs hold, perhaps they need to be reclassified once again.
Given that TD are pericolonic aggregates of carcinoma that are known to predict a high likelihood of metastasis, our hypothesis was that tumors with TD are more likely to recur in the peritoneum. This phenomenon is not well studied. While other papers have looked at recurrence as a whole, literature on the patterns of recurrence is scarce. 8 Furthermore, factors typically considered as high-risk (HR) features; obstruction, perforation, and T4 tumors, are known predictors of peritoneal failure. [9][10][11] Nonetheless, we hypothesized that TD may be a more significant predictor of peritoneal recurrence than HR features and that their presence may necessitate novel surveillance and treatment strategies.

| Study source
This is a retrospective cohort study that utilized our institutional electronic medical record, EPIC (Epic Systems Corporation). Using International Classification of Diseases (ICD)-10 codes C- 18

| Patient characteristics
Using SlicerDicer functionality in EPIC, CRC patients were identified. For data extraction, natural language processing (NLP) was used to tag patient charts (all clinical notes and pathology reports) containing any of the four keywords: TD, T4, perforation, and obstruction. Approximately 1000 charts were tagged for at least one key word. All these 1000 charts were manually checked for inclusion criteria (final n = 151). To determine if NLP was complete in picking up charts, the excluded~500 charts were randomly sampled and manually chart checked to ensure no misidentification of charts had occurred. The TD group was defined as the presence of TD noted on pathology reports whereas the HR group was defined as the presence of one or more of the following features: T4, perforation, or obstruction. T4 was primarily noted on pathology reports whereas perforation or obstruction was defined as their mention on either pathology, colonoscopy, radiologly, or clinical notes. If patients had more than one HR feature they were still categorized in the HR group however any patient having TD and a HR feature, was categorized in the TD group (Supporting Information: Figure S1).

| Study variables
Chart review was conducted to gather demographic variables.
Pathology reports were used to extract information on tumor grade, histology, size, stage; lymph node, TD, or margin status; and perineural or lymphovascular invasion. TD counts were categorized as 1−4 or ≥5 based on AJCC 8th edition guidelines.

| Tumor characteristics
The most common tumor location was ascending colon in the TD group (24.0%) and sigmoid colon in the HR group (31.7%) ( Overall cancer-specific mortality was found to be 24.  Multiple other studies have also reported TDs to be associated with LNM. [13][14][15] In their meta-analysis, Nagtegaal et al. reported that TDs increased as the N-stage increased and that the relative risk of TD was 4.2 (95% CI: 3.2−5.6) in the presence of LNMs. 6 In another study that evaluated 35 CRC patients with TDs, the authors reported TD patients to have worse outcomes when 0−3 LNM (N0 or N1) were present but when 4 or more were found (N2), prognosis was worse regardless of TD status. 14 These findings were verified in an updated meta-analysis of 90 455 CRC patients over 19 studies that reported N1c patients had worse disease-free survival than N1 patients but better than that of N2 patients. 5 Despite building evidence, current staging guidelines assign N1a/b and N1c an equal prognostic role in staging criteria. Interestingly, the latest edition of the AJCC guidelines also state that TDs should not have any identifiable lymph nodal tissue. This raises the question of whether TD should be placed within the N-stage to begin with. Our study provides evidence that and inhibits E-cadherin) that increases angio-invasiveness and cellular motility, hence facilitating tumor invasion. 16 Goldstein 3.21) and that the addition of TD to LNM did not significantly increase the risk of peritoneal metastasis (OR 6.97). 6 Although TDs have been identified as early as 1935, their origin remains to be discovered. 7 Literature reports them to have components of vascular, neural, lymphatic, or a combination of all three, tissues, on histopathologic analysis. 2,17 Whether TD are in-transit metastasis or a complete metastatic replacement of lymph nodes, it is evident that these tumors have access to multiple anatomic pathways leading to early and extensive disease spread. 6 Not surprisingly then, we noticed along with many other authors, that patients with TD had a higher risk of mortality. 2,5,7,13,18 Guidelines from National Comprehensive Cancer Network and For surveillance, peritoneal disease is difficult to detect on current imaging modalities such as computed tomography scans. 24 Our data suggests that utility of a diagnostic laparoscopy at 1-year for surveillance in these patients should be explored further.
There are several limitations that must be considered while interpreting our results. This is a single-institution study, with a limited sample size despite identifying more than 1500 CRC patients initially. Due to the lack of relevant variables, large databases (such as NCDB or SEER) could not be used to answer the research question. The next step will be to perform a multicenter prospective study to further corroborate the findings of our study. Furthermore, the retrospective nature can only suggest associations rather than causation.
Many patients (~50%) in the TD group also had HR features. Due to sample size limitation such patients could not be eliminated and hence a TD only group was not chosen. Although, multivariate analysis controlling for HR features did report that TDs were independently associated with cancer recurrence, when lymph node positivity was adjusted for, statistical significance was lost. It is possible that this is primarily due to small sample sizes given a p-value (p = 0.06) approaching significance. Furthermore, a number of other factors such as lymphovascular invasion or perineural invasion could not be controlled for either, on the multivariate analysis as a result of sample size limitations. Since the TD group had significantly higher proportion of perineural or lymphovascular invasion, and since our analysis did not control for any of these factors, it is possible that the worse outcomes seen were not due to TDs themselves.
Patients were selected from 2010 to 2015 to allow sufficient follow-up time; however, treatment regimens for this disease have continued to evolve. 25 Patients who received neoadjuvant chemotherapy were also part of the cohort (although a very small number) and how this impacted tumor biology is unknown. Ideally, an analysis of patients who received adjuvant therapy is also warranted with a discussion on how the type of therapy and its timing factors in, however that requires a separate study with multi-institutional collaboration resulting in higher sample size.
Finally, the HR group had a median follow-up that was twice that of the TD group and it is unclear how that affected outcomes.

| CONCLUSIONS
This study suggests that non-metastatic CRC patients with TDs identified on pathology might have an increased risk of peritoneal recurrence, which typically occurs at 1 year from surgical resection. It also provides evidence that TD patients have outcomes comparable to, if not worse than, HR CRC patients. Based on these data, we suggest investigation into enhanced surveillance of TD patientsperhaps diagnostic laparoscopy at or before 1-year and adoption of more aggressive systemic therapy treatment to combat the significantly higher failure rate.