International Validation of the Immunoscore Biopsy in Patients With Rectal Cancer Managed by a Watch-and-Wait Strategy

Abstract

Purpose: No biomarker capable of improving selection and monitoring of patients with rectal cancer managed by watch-and-wait (W&W) strategy is currently available. Prognostic performance of the Immunoscore biopsy (IS_B) was recently suggested in a preliminary study.

Methods: This international validation study included 249 patients with clinical complete response (cCR) managed by W&W strategy. Intratumoral CD3+ and CD8+ T cells were quantified on pretreatment rectal biopsies by digital pathology and converted to IS_B. The primary end point was time to recurrence (TTR; the time from the end of neoadjuvant treatment to the date of local regrowth or distant metastasis). Associations between IS_B and outcomes were analyzed by stratified Cox regression adjusted for confounders. Immune status of tumor-draining lymph nodes (n = 161) of 17 additional patients treated by neoadjuvant chemoradiotherapy and surgery was investigated by 3'RNA-Seq and immunofluorescence.

Results: Recurrence-free rates at 5 years were 91.3% (82.4%-100.0%), 62.5% (53.2%-73.3%), and 53.1% (42.4%-66.5%) with IS_B High, IS_B Intermediate, and IS_B Low, respectively (hazard ratio [HR; Low v High], 6.51; 95% CI, 1.99 to 21.28; log-rank P = .0004). IS_B was also significantly associated with disease-free survival (log-rank P = .0002), and predicted both local regrowth and distant metastasis. In multivariate analysis, IS_B was independent of patient age, sex, tumor location, cT stage (T, primary tumor; c, clinical), cN stage (N, regional lymph node; c, clinical), and was the strongest predictor for TTR (HR [IS_B High v Low], 6.93; 95% CI, 2.08 to 23.15; P = .0017). The addition of IS_B to a clinical-based model significantly improved the prediction of recurrence. Finally, B-cell proliferation and memory in draining lymph nodes was evidenced in the draining lymph nodes of patients with cCR.

Conclusion: The IS_B is validated as a biomarker to predict both local regrowth and distant metastasis, with a gradual scaling of the risk of pejorative outcome.

FIG 1.

IS_B in patients with rectal cancer managed by the W&W strategy: study design, IS_B methodology, and prognostic value. (A) Flow chart of the IS_B multicenter study design (further details are provided in the Data Supplement). (B) Left: representative image of biopsy analysis; the tumor region is selected (pink), and normal tissue or dysplasia (blue) are excluded from the analysis. Middle: representative detection by the software of positive CD3+ and CD8+ T cells infiltrating the rectal tumor of the patient with IS_B High. Right: chart illustrating the IS_B calculation method. Densities of CD3+ and CD8+ T cells (cells/mm²) in the tumor region are converted into percentile values using predefined cutoffs. The mean percentile of the two markers is calculated to generate IS_B mean score value, where IS_B Low, IS_B Int, and IS_B High subgroups are reflected by 0%-25%, >25% to 70%, and >70% to 100% percentile, respectively. (C) Time to recurrence and (D) DFS according to IS_B Low (red), IS_B Int (blue), and IS_B High (green) in W&W patients with cCR. Log-rank statistical test is stratified by the type of neoadjuvant radiotherapy (standard without intensification, external intensification, or [contact] brachytherapy). ^aEl Sissy et al. *P < .05; **P < .01; ***P < .001; ****P < .0001. cCR, clinical complete response; DFS, disease-free survival; Int, Intermediate; IS_B, Immunoscore biopsy; P_tft, P test for trend; TTR, time to recurrence; W&W, watch-and-wait.

FIG 2.

Multivariable time to recurrence survival analysis in the test and validation cohorts (n = 317). The model is stratified by neoadjuvant radiotherapy type (standard without intensification, external intensification, or [contact] brachytherapy). c, clinical; HR, hazard ratio; Int, Intermediate; IS_B, Immunoscore biopsy; N, regional lymph node; T, primary tumor; TTR, time to recurrence.

FIG 3.

IS_B and clinical outcome. Comparison with clinicopathologic parameters. Cumulative incidence of (A) local regrowth and (B) distant metastasis in W&W patients from the test and the validation cohorts (n = 322 patients), according to IS_B categories. (C) The 3- and 5-year organ-preservation rates in W&W patients with cCR according to IS_B categories. (D) Box plot of the predictive accuracy for TTR on the basis of the incremental area under the curve with 1,000× bootstrap resampling for each parameter. The log likelihood ratio test between clinical model and clinical model plus IS_B is shown. The relative importance of each risk parameter to recurrence risk using the χ² proportion test is shown in the pie chart. ^§cN, tumor location, and sex represent <2% of the relative contribution. *P < .05; **P < .01; ***P < .001; ****P < .0001. c, clinical; cCR, clinical complete response; iAUC, integrated area under the ROC curve; Int, Intermediate; IS_B, immunoscore biopsy; N, regional lymph node; P_tft, P test for trend; T, primary tumor; TTR, time to recurrence; W&W, watch-and-wait.

FIG 4.

Immune status of the draining lymph nodes of rectal cancer patients with or without cCR to neoadjuvant chemoradiotherapy. Illustrations from the left to the right of (1) double immunostaining of Ki-67 (brown) and Lyve-1 (red) in lymph nodes, (2) recognition of regions of interests: follicle (blue), germinal center (yellow), and medulla (red) by the software (Classifier module, Halo), (3) focus on a follicle with Ki-67+ cells, and (4) detection of nuclei (blue) and Ki67+ cells (brown). (B) Ki67+ cell densities in lymph node regions (cortex, paracortex, and medulla) of cCR (R) or non-cCR (NR) to neoadjuvant chemoradiotherapy. (C) Heatmaps of genes associated with B cells according to proliferative (Ki67) status of lymph nodes and response status (R) of the patients to nCRT. Gene expression level is represented as a color gradient from low (blue) to high (red) intensity. Only overexpressed genes among pathologic complete response proliferative (Ki67 hi) lymph nodes in responders are shown (18/30 expressed genes). (D) Upper left: immunofluorescence staining for proliferative B-cell panel (Ki67 in yellow, CD20 in pink, and double positive in orange) in a draining lymph node. Lower left: immunofluorescence staining for memory B-cell panel (CD27 in yellow, CD38 in pink, and CD20 in blue) in a draining lymph node. Upper right: proliferative B-cell densities in lymph nodes of responders compared with nonresponders. Lower right: memory B-cell densities in lymph nodes of responders Ki67 high compared with control (nonresponders Ki67 low). *P < .05; **P < .01; ***P < .001; ****P < .0001. cCR, clinical complete response; cpm, read counts per million of reads; nCRT, neoadjuvant chemoradiotherapy; NR, nonresponders; R, responders.