Cellular senescence predicts treatment outcome in metastasised colorectal cancer

Abstract

Background: Cellular senescence is a terminal cell-cycle arrest that occurs in response to activated oncogenes and DNA-damaging chemotherapy. Whether cancer cell senescence at diagnosis might be predictive for treatment outcome is unknown.

Methods: A senescence index (SI) was developed and used to retrospectively correlate the treatment outcome of 30 UICC stage IV colorectal cancer (CRC) patients with their SI at diagnosis.

Results: 5-Fluorouracil/leucovorin-treated CRC patients achieved a significantly longer progression-free survival when presenting with SI-positive tumours before therapy (median 12.0 vs 6.0 months; P=0.044).

Conclusion: Cancer cell senescence predicts treatment outcome in metastasised CRC. Prospective analyses of larger patient cohorts are needed.

Figure 1

Stratification of stage IV CRC samples by a senescence index (SI) predicts PFS to first-line 5-FU/LV therapy. (A) Proportions of senescent cases in cryo-preserved colorectal tissue sections of the indicated groups (normal crypt mucosa (n=5, 0 senescent); adenoma (n=12, 8 senescent), invasive carcinoma (n=6, 1 senescent)). Note that 9 out of 12 adenoma and 2 out of 6 carcinoma cases would score ‘senescent’ if judged in low-proliferating areas only (see D). (B) Average frequency of senescent cells per group (as in A), measured as the percentage of SA-β-gal-positive (blue) cells, and compared to the rate of Ki67-positive (red nuclear staining) cells (top: error bars represent the standard deviation; bottom: matched areas of representative photomicrographs – a non-senescent normal crypt mucosa, a senescent adenoma and a formally non-senescent carcinoma (but showing numerous senescent cells)). Notably, in normal colorectal mucosa, only cells of the most differentiated luminal mucosa stain SA-β-gal-positive. (C) FFPE sections of pelleted Ras-infected senescent (sen.) human fibroblasts (SA-β-gal-positive and Ki67-negative frozen material as a reference) show much stronger immunoreactivity for p-ERK, HP1γ and cytoplasmic PAI-1 when compared with non-senescent (non-sen.) mock-infected fibroblasts. (D) SI values, based on the expression of these three markers, were obtained in 30 cases of stage IV CRC specimens at diagnosis, and used to stratify PFS following 5-FU/LV first-line chemotherapy (senescent (n=12; blue line) vs non-senescent (n=17; red line); one case scored ‘not conclusive’ (see Figure 2B)). Genomic sequencing in a subset of these 29 specimens identified K-Ras codon 12 or 13 mutations in 5 out of 6 senescent, but only 2 out of 11 non-senescent cases (inset). Note that a primary stratification by the K-Ras mutation status unveiled no significant differences in PFS (P=0.128).

Figure 2

Generation and validation of the senescence index (SI). (A) Technical flow diagram of the generation and application of the SI in a Ras-transduced human fibroblast cell line and in colorectal tissue specimens. (B) Formula and cut-off values of the SI based on the expression p-ERK, HP1γ and PAI-1. Coefficients reflect the reciprocal average percentage of cells that stained positive for the respective marker in the learning adenomas to normalise the relative weight of the three markers. The correction value –19 was chosen to set the discrimination threshold at 0, with the range between −1 and +1 considered ‘non-conclusive’. (C) The majority of validation set adenomas is senescent when analysed by the SI. A set of seven matched samples (frozen and FFPE material from the same adenoma) underscores the high concordance between the SA-β-gal/Ki67- (see definition in Materials and Methods section) and the SI-based assignment.