Clinical impact of heterogeneously distributed tumor-infiltrating lymphocytes on the prognosis of colorectal cancer

Patients

This study was approved by the ethics committee of People’s Hospital of Baoan District (Date: March 8, 2021; NO: BYL20210337). The informed consent for research was waived since the following three conditions can be met: (1) it is difficult to obtain informed consent, (2) the research is of high social importance, (3) the research does not infringe the rights and interests of study subjects.

Seventy-eight CRC patients were retrospectively collected from the pathology archive of People’s Hospital of Baoan District between 2017 and 2019. All enrolled patients underwent complete surgical resection so that their surgical margins were microscopically negative for residual tumors. None of the patients had been treated with preoperative chemotherapy or radiotherapy. Patients who were death or recurrence within 1 month after surgery, were excluded. Their clinical and histopathological data were reviewed by two experienced gastrointestinal pathologists according to the American Joint Committee on Cancer the tumor-node-metastasis (TNM) staging system. The observation time in this study was the interval between diagnosis and last contact (death, local recurrence or lost to follow-up).

Immunohistochemistry

Formalin-fixed paraffin-embedding tissue blocks were serially sectioned at 5-µm thickness. Prior to immunohistochemistry, tissue slides were assessed by hematoxylin and eosin staining to confirm the presence of tumor and adjacent cells. Then, they were deparaffinized in xylene, ethanol and water, and were treated with citrate buffer (0.01 M, pH 6.0) and heated in a microwave oven at 90 °C for 30 min. After blocking by 5% (v/v) human serum, tissue slides were incubated with the rabbit polyclonal anti-CD3 antibody (ab5690; dilution, 1:100; Abcam, Cambridge, UK) or anti-CD8 antibody (ab4055; dilution, 1:200; Abcam) at 37 °C for 120 min, followed by incubation with the HRP-conjugated secondary antibody(goat anti-rabbit IgG H&L; ab214880; Abcam) for 15 min at 37 °C .

Cell quantification

3D Histech Midi Scanner System was used to digitalize stained slides that were viewed in the Panoramic Viewer system. Five visual fields with the most abundant infiltration of T cells were selected from each slice to calculate the mean density of TILs. Numbers of positively immunostained cells were double-blindly evaluated by two experienced pathologists.

Statistical analysis

The patients’ parameters that were taken into account for statistical analysis contained clinical (serum level of carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9), chemotherapy regimen, and recurrence as well as age, sex), histopathologic (tumor size, histological type, pathologic TNM, and differentiation), and immunologic variables (CD3+ and CD8+ cells). Pearson’s χ2 test or Fisher’s exact (two-sided) test was used for examining the difference between groups. Survival curves were estimated by the Kaplan–Meier method, and the comparison between groups were performed using the log-rank test. Multivariate logistic regressions were used to identify clinicopathological features associated with the recurrence of patients with early stage diseases. Proportional hazards models (Cox) were performed to identify the variables associated with patient’s survival. All statistical analysis described above was conducted using SPSS 22.0, and p-value <0.05 was thought to be statistically significant.

Study population

We retrospectively analyzed 78 CRC patients without preoperative treatment. Their clinicopathological characteristics were presented in Table 1. Histologically, most patients were diagnosed as adenocarcinoma (59/78, 75.6%), and presented with deep tumor invasion (T3-T4, 67/78, 85.9%). The incidence of lymph node metastasis was 42.3% (33/78). Primary tumors were commonly located on the left side (56/78, 71.8%). 38 patients (48.7%) experienced local recurrence after surgery. The median duration of PFS was 31.1 months (19.46–35.49 months, 95% confidence interval [CI]). Among the clinicopathological parameters, the depth of tumor invasion (T stage, P = 0.033), lymph node invasion (N stage, P < 0.0001) and TNM stage (P < 0.0001) were significantly associated with PFS (Table 1).

TILs distribution pattern in CRC

In the present study, to characterize the distribution pattern of TILs, the whole slide was digitally imaged and segmented into three regions for counting the stained TILs. We found both CD3+ and CD8+ TILs displayed a similar distribution pattern. They were mainly located along the invasive margin (CD3+TILs^IM, 1,063.22 ± 99.13; CD8+TILs^IM, 386.75 ± 102.01), and, to a lesser extent, at the tumor center (CD3+TILs^TC, 122.54 ± 48.18; CD8+TILs^TC, 70.82 ± 27.03) and at the tumor stroma (CD3+TILs^TS, 336.61 ±  100.80; CD8+TILs^TS, 165.41 ± 37.22). Statistical significances were observed among the density of TILs at these difference areas (Figs. 1G and 1H), indicating that TILs were unevenly distributed in CRC. Although intratumoal TILs showed marginal increases in CRC patients with higher peritumoral TILs, these did not reach statistical significance (CD3+TILs^TC vs CD3+TILs^IM, P = 0.535; CD3+TILs^TC vs CD3+TILs^TS, P = 0.240; CD8+TILs^TC vs CD8+TILs^IM, P = 0.093; CD8+TILs^TC vs CD8+TILs^TS, P = 0.058). The ratio of CD3+ to CD8+ TILs was also evaluated. Tumor center displayed a significantly lower CD3+/CD8+ ratio compared with that in the invasive margin (P < 0.001, Fig. 1I), suggesting that the composition of infiltrated T cells was significantly different between tumor center and invasive margin.

The correlation between TILs densities and clinicopathological features

The cohort was divided into low and high-density groups according to the median threshold, and spearman’s rank correlation was used to explore the correlation between TILs densities and clinicopathologic features. Only intratumoral and peritumoural TILs were considered in the following analysis since there was no significant correlation between TILs in tumor stroma and clinicopathological features (Table S1). As shown in Table 2, the associations of clinicopathological features with CD3+ and CD8+ TILs at the same region were not identical. For CD3+TILs, low density of CD3+ TILs^IM was associated with larger tumor size (odds ratio [OR], 0.309; 95% CI [0.121–0.786]), deeper tumor invasion (OR, 0.181; 95% CI [0.036–0.897]), positive lymph node metastasis (OR, 0.303; 95% CI [0.118–0.778]) and advanced stage (OR, 0.245; 95% CI [0.095–0.634]). Intratumoral CD3+, CD3+ TILs^TC, was more frequently observed in primary tumor located at the right side (OR, 0.324; 95% CI [0.109–0.964]). For CD8+ TILs, intratumoral CD8+, CD8+ TILs^TC, was significantly higher in tumors with the well differentiation (OR, 0.228; 95% CI [0.066–0.782]) and without lymph node metastasis (OR, 0.347; 95% CI [0.138–0.871]). Peritumoural CD8+, CD8+TILs^IM, was also inversely correlated with tumor stage (OR, 0.326; 95% CI [0.126–0.843]). No significant association was observed between densities of TILs with age, sex, as well as the level of serum CEA and CA19-9.

The correlation with total patient outcome

Table 3 showed the results of univariate survival analysis for the clinicopathological features. We found that the degree of tumor invasion (HR, 0.325; 95% CI [0.158–0.669]; p = 0.002), and lymph node metastasis (HR, 0.260; 95% CI [0.137–0.495], p = 0.0001) were the factors associated with CRC patient’s survival. Kaplan–Meier analysis showed that the higher densities of CD3+ and CD8+ TILs at the tumor center and invasive margin were implicated in favorable outcomes (Fig. 2).

Multivariate logistic regression was performed to identify independent prognostic factors, in which all features found to be prognostic in univariate analysis were included. Multivariate analysis revealed that N stage (HR, 0.474; 95% CI [0.260–0.862]; P = 0.017), CD3+ TILs^IM, (HR, 1.893; 95% CI [1.071–3.460]; P = 0.028) and CD8+ TILs^TC (HR, 2.050; 95% CI [1.211–3.704]; P = 0.013) showed independent prognostic significance. These observations also confirmed that high densities of CD3+ TILs^IM and CD8+ TILs^TC, correlated with good outcome.

The correlation with early-stage patient outcome

Of the 43 patients with stage I/II CRC, 7 patients (16.28%) developed disease recurrence. No significant difference was observed between recurrence and non-recurrence patients in their clinicopathological features, including age, sex, serum CEA and CA19-9 level, tumor size and the depth of tumor invasion. We next studied whether the characteristic of TILs could offer a prognostic indicator for the recurrence of early-stage CRC patients. As shown in Table 4, recurrent CRC patients had a higher CD3+/CD8+ TILs ratio in invasive margin, whereas no statistical difference was seen for densities of CD3+ or CD8+ TILs in different regions. These results also suggested that compared with the absolute number or density of TILs, the composition of TILs had a more important impact on the prognosis of early-stage CRC patients.