Prognostic significance of additional histologic features for subclassification of pathological T3 colon cancer

Background Additional histologic features of T3 colon cancer, such as tumour depth invasion beyond muscularis propria and elastic lamina invasion (ELI), have taken interest for a more accurate staging. Methods Patients with pT3 and pT4a (control group) colon adenocarcinoma were retrospectively collected from our institutional database. The study group was divided according to depth of tumour invasion < 5 mm and ≥ 5 mm, and into ELI − and ELI + . Chi-square test was used to compare the clinicopathological characteristics. OS and DFS were estimated using Kaplan–Meier method and compared with the log-rank test. Univariable and multivariable Cox proportional hazard models were employed to assess the effect on OS and DFS. Results Out of 290 pT3 tumours, 168 (58%) had a depth of tumour invasion < 5 mm and 122 (42%) ≥ 5 mm. The 5-year OS and DFS were 85.2, 68.7 and 60.9%, and 81.4, 73.9 and 60.1% in pT3 < 5 mm, pT3 ≥ 5 mm, and pT4a respectively (p = 0.001, p = 0.072). Considering ELI − (n = 157, 54%) and ELI + (n = 133, 46%), the 5-year OS and DFS were 78.9, 76.7, and 60.9%, and 75.5, 81.5, and 60.1% in ELI  − , ELI + and pT4a respectively (p = 0.955, p = 0.462). At multivariable analysis, the depth of invasion was found to be an independent predictive factor for OS (HR 2.04, 95%CI 1.28–3.24, p = 0.003) and DFS (HR 1.98, 95%CI 1.24–3.18, p = 0.004), while ELI did not result a prognostic factor for OS nor DFS. Conclusion In pT3 colon cancer, depth of tumour invasion ≥ 5 mm is an independent risk factor for OS and DFS, whereas ELI did not result a prognostic factor affecting OS nor DFS. Supplementary Information The online version contains supplementary material available at 10.1007/s10147-022-02192-y.


Introduction
Colorectal cancer (CRC) is the third most common tumour diagnosed and the third cause of cancer-related deaths in the United States, accounting for 150,000 new estimated cases in 2020 [1]. To date, surgical resection of the tumour with en-bloc removal of the regional lymph nodes is the standard of care for non-metastatic colon cancer.
On the contrary, stage III patients are usually treated with adjuvant chemotherapy to reduce the risk of recurrence [9]. Lorenzo Macchi and Quoc Riccardo Bao equally contributed to this manuscript.
Additional prognostic factors for survival after resection were investigated [12][13][14][15]. Whereas in rectal cancer pT3 sub-classification was widely validated [16][17][18], few studies evaluated additional pathological features for a prognostic subdivision of pT3 colon cancer. Sub-classification of T3 colon cancer based on the depth of infiltration (DOI) beyond the muscularis propria, or peritoneal elastic lamina invasion (ELI) were proposed [19][20][21][22][23][24][25][26]. Yoo et al. demonstrated that a subdivision based on the measurement of the maximal DOI is correlated to nodal and distant metastasis and OS [19]. In contrast, Mrak et al. reported that subdivision based on DOI does not provide any additional information about longterm oncologic outcome [20]. Furthermore, supported by the established role of elastic lamina as landmark of visceral pleura invasion in lung cancer staging [27], several studies evaluated peritoneal ELI as a prognostic marker in CRC. Authors reported that pT3 tumours breaching the peritoneal elastic lamina (ELI +) were associated to a worse survival than pT3 tumours not breaching the peritoneal elastic lamina (ELI − ) [23][24][25][26]. On the contrary, Grin et al. found no significant differences in survival between ELI + and ELI-pT3 tumours in stage II CRC [28].
The aim of our study was to evaluate and compare the clinical significance of pT3 subdivision of intraperitoneal CRC based on the DOI beyond muscularis propria and ELI. To our knowledge, this is the first study that compares these two parameters for a prognostic sub-classification of pT3 colon cancer.

Patients selection
All patients surgically treated for primary CRC from January 1 st 2008 to December 31st 2018 were collected from the prospectively maintained colorectal database of the General Surgery 3, University Hospital of Padova. Inclusion criteria were radical resection (R0) for pT3 (study group) and pT4a (control group) colorectal adenocarcinoma. Patients with extraperitoneal rectal adenocarcinoma, who underwent neoadjuvant treatment, with metastatic disease, pT1-pT2-pT4b disease, or histology other than adenocarcinoma were excluded. Patients lost on follow-up were also excluded.

Clinicopathological and treatment
For each patient data regarding age, sex, body mass index (BMI), ASA score, serum carcinoembryonic antigen (CEA), location (right/left colon), surgical approach, tumour maximal diameter, grading, lymphovascular and perineural invasion, metastatic lymph nodes, and adjuvant treatment were collected. All patients underwent standard oncological surgical resection. Follow-up was performed according to national guidelines [29]. Local recurrence was defined as recurrent disease at the site of the original CRC, whereas distant recurrence was defined as any disease identified outside the primary site.

Histopathologic analysis
All pathological specimens were staged according to the American Joint Committee on Cancer TNM classification 8th edition [30]. According to our local gross sampling protocol, at least 4 samples from the primary tumour are analysed. All pT3 surgical specimens (n = 1160) were jointly reviewed by two gastrointestinal dedicated pathologists, who were blinded to clinical and others pathological data, to assess the sample over four most adequate to determine the DOI. Samples (n = 17) with orientation artefacts due to inadequate paraffin embedding were excluded from the selection. DOI was measured in mm from the end of longitudinal muscle layer into the nearby adipose tissue (Fig. 1). The morphometric evaluation of DOI was taken from a dedicated gastrointestinal pathologist with the support of the digital microimaging device Leica DMD108™. The presence of peritoneal ELI was jointly evaluated by two gastrointestinal pathologists, who were blinded to clinical and others pathological data, as landmark for tumour invasion between adipose tissue and serosal layer. Additional elastic stain was performed only in doubtful cases (n = 47) (Fig. 2).

Statistical analysis
Considering the DOI, the study group was divided by a cutoff determined by a Receiver Operating Curve (ROC) for disease-free survival (DFS) (table of sensitivity, specificity and Youden's J index are shown in Supplementary file, Table 1). Moreover, the study group was divided into two groups, according to the peritoneal ELI, into ELI − and ELI + .
Descriptive statistics was reported as absolute numbers and percentages, and variables distribution in the different subgroups was evaluated using the Chi-squared test. OS and DFS were evaluated using the Kaplan-Meier method and pairwise comparisons were conducted with the log-rank test. P-value associated with the pairwise comparisons underwent Benjamini-Hochberg correction to account for multiplicity of testing. The incidence of recurrence in the subgroups was evaluated using Cumulative Incidence Functions (CIF) in a competing risk framework. Differences in probabilities were evaluated using the Gray's test. Univariable and multivariable Cox proportional hazard models were employed to assess the effect of the variables of interest on mortality and disease relapse. Results were reported as Hazard Ratio (HR), 95% confidence interval and p value. The level for statistical significance was determined at p < 0.05. All statistical analyses were performed using R software (version 4.0.3) [31].

Clinicopathological characteristics
Clinicopathological characteristics are summarized in Table 1. In the study group (290 patients) there were 164(56.6%) men and 126(43.5%) women, and the median age of was 73(37-95) years. The location of the tumour was right colon in 137(47.2%) and 153(52.8%) left colon. Open surgical approach was performed in 184(63.4%) patients, while a laparoscopic approach in 106(36.6%) patients. Lymph nodes metastasis were retrieved in 121(41.8%) patients. Adjuvant chemotherapy was performed in 143(49.3%) patients. At a median follow-up of 50.5  months, local and distant recurrence occurred in 9(3.1%) and 51(17.6%) patients respectively. In the control group, 62 pT4a patients was considered for outcomes comparison.

Univariable and multivariable analysis
At univariable analysis, factors significantly associated with a decreased OS were age, ASA score, lymphatic invasion, lymph nodes metastasis, and DOI (pT3 ≥ 5 mm) ( Factors significantly associated with a decreased DFS were sex, lymphatic invasion, perineural invasion, lymph

Discussion
Assessment of additional pathological features might allow an accurate risk stratification to identify high-risk stage II CRC patients who benefit from adjuvant therapy. A pT3 subclassification based on DOI was proposed by several authors, while others strongly suggested a sub-classification based on peritoneal ELI. We used a subdivision of pT3 using a cut-off, determined by using a ROC for DFS, of 5 mm DOI beyond the muscular layer, which resulted to affect survival. A DOI > 15 mm was proposed by Merkel et al. as a major risk factors for stage II colon carcinoma [32], while various cut-offs between 2 and 10 mm were demonstrated for the subdivision of T3 CRC [16,33]. Yoo et al., similarly to Pollheimer et al., divided pT3 patients using 4 different cut-offs (< 1 mm; 1-5 mm; 5-15 mm; > 15 mm), reporting 5 mm cut-off as the strongest prognostic factor in both their analysis [19,34]. Recently, Nomura et al. reported 148 pT3 CRC subdivided into T3a (< 1 mm); T3b (1-5 mm); T3c (> 5 mm) and confirmed a 5 mm as optimal cut-off predictor for recurrence [21].
At univariable and multivariable analysis pT3 ≥ 5 mm category resulted as independent prognostic factor both for OS and DFS. Similarly, Bori et al. divided 593 CRC by a 5 mm cut-off, reporting that pT3 < 5 mm is associated with an improved long-term outcomes in terms of nodal involvement and distant metastasis [35]. Furthermore, similar results were reported by several authors [19,21,36], in particular Akagi et al. reported that pT3 ≥ 5 mm was the strongest independent risk factor for recurrence on 202 cases of stage II colon cancer, and proposed adjuvant chemotherapy as indicated for these patients [37]. Based on this rationale, our sub-analysis on pT3N0 and pT3N + aimed to determine if the DOI ≥ 5 mm may be a recommendation for adjuvant treatment. In this setting, the DOI failed to result as predictive factors for survival in pT3N0 patients, although it was confirmed as a strong independent predictor of OS and DFS in pT3N + patients (HR 3.26, 95%CI 1.56-6.78, p = 0.002, and HR 2.31 95%CI 1.15-4.65, p = 0.019 respectively). In this setting, DOI confirmed to be an index of advanced disease in pT3N + patients, as far as Kaplan-Meier survival analysis showed that in N + group pT3 ≥ 5 mm patients had a survival even worse than pT4 control group (5-year OS 50.0 Vs 60.9%, 5-year DFS 59.3 Vs 60.1% respectively). For this reason, we also strongly suggest to consider DOI as a high-risk factor, that may help in case of doubt for adjuvant treatment, even in pT3N0 patients.
Alternatively to the measurement of DOI, Shinto et al. used the peritoneal ELI as a cut-off, finding an increased  demonstrate peritoneal ELI by tumour that should be considered for routine use in all CRC [26,38]. Lu et al., proposed ELI assessed with elastic stain as prognostic factors for stage II colon cancer, and might be an indication to postoperative adjuvant chemotherapy [39]. Recently, a meta-analysis including six studies, recommended the sub-categorization of pT3 CRC by ELI for better prognostic assessment and treatment strategy of patients with CRC [40]. Our analysis failed to confirm any significant impact of ELI on oncological outcome, and ELI as predictive factor at univariable analysis. To note, lymph nodes metastasis and adjuvant treatment have a main role in long-term outcome, and the prognostic role of ELI may have been mitigated by the high rate of Stage III who underwent adjuvant treatment (70% in ELI + in pT3N + patients). As reported by Yokota et al., even if ELI resulted as independent risk factor for RFS and OS, RFS resulted almost identical when comparing pT3N + / ELI + to pT3N + /ELI-in patients with no adjuvant treatment [25]. Along with DOI and ELI, the circumferential resection margin (CRM) was investigated in colon cancer also, whereas its prognostic role in rectal cancer is well known. NCDB registry showed that a positive CRM were common in pT4 (26-32% of patients), whereas in pT2 and pT3 colon cancer is less common (6 and 11% of patients respectively) [41]. A positive CRM should be an indication for adjuvant treatment in stage II colon cancer, however in only 9% of these patients were reported [42]. Nevertheless, the histopathological parameters that we considered are more common in T3 colon cancer patients, representing in both cases more than 40% of the patients. Furthermore, beside pathological analysis in the last years circulating tumour DNA (ctDNA) is assuming an increasing role in the prognosis of colon cancer. Our group previously reported that an increased level of ctDNA was associated to a poor prognosis [43], whereas an Australian multicenter RCT reported ctDNA as independent predictors of recurrence-free survival [44]. Unfortunately, even considering these promising data, the ctDNA is not commonly tested in the normal followup of colon cancer, whereas the evaluation of DOI and ELI are an unexpensive extension of normal pathological examinations.
There are some limitations in this study. First, this is a single-institution retrospective study. Even if we considered the limitation of a single institution study, in our study group almost 300 colon cancer were re-evaluated by dedicated colorectal pathologists. Second, the patients included in the study covered a wide timespan of 10 years, during which several changes occurred in staging modalities, preoperative treatment, anaesthesiological, pathological, and surgical techniques, follow-up and adjuvant treatment. All these changes may have a potential impact on OS and DFS. Lastly, we did not identify other confounding

Conclusion
In our study, the DOI beyond the muscular layer of colonic wall, using a cut-off of 5 mm, is an independent risk factor both for OS and DFS, and may be considered as highrisk feature in pT3 colon cancer. ELI was not resulted to be a prognostic factor affecting OS and DFS. To our knowledge, this is the first study that compares these 2  parameters for a prognostic sub-classification of pT3 colon cancer.