Metastatic progression is associated with dynamic changes in the local microenvironment

Abstract

Most cancer-associated deaths result from metastasis. However, it remains unknown whether the size, microenvironment or other features of a metastatic lesion dictate its behaviour or determine the efficacy of chemotherapy in the adjuvant (micrometastatic) setting. Here we delineate the natural history of metastasis in an autochthonous model of pancreatic ductal adenocarcinoma (PDAC), using lineage tracing to examine the evolution of disseminated cancer cells and their associated microenvironment. With increasing size, lesions shift from mesenchymal to epithelial histology, become hypovascular and accumulate a desmoplastic stroma, ultimately recapitulating the primary tumours from which they arose. Moreover, treatment with gemcitabine and nab-paclitaxel significantly reduces the overall number of metastases by inducing cell death in lesions of all sizes, challenging the paradigm that PDAC stroma imposes a critical barrier to drug delivery. These results illuminate the cellular dynamics of metastatic progression and suggest that adjuvant chemotherapy affords a survival benefit by directly targeting micrometastases.

Figure 1. Metastatic landscape of the KPCY model.

(a) Representative images of YFP⁺ liver metastases and primary tumour. Scale bars, 50 μm. (b) Size distribution of metastatic lesions in the liver grouped according to size (n=23 mice). (c) Quantification of KI67⁺ (P=0.9439) tumour cells in liver metastases (n=6 mice, 159 lesions). (d) Quantification of CC3⁺ (P=0.1621) tumour cells in liver metastases (n=4 mice, 46 lesions). P-values were calculated by one-sample t-test against ‘single cell' means. Data are presented as mean±s.d.

Figure 2. Metastatic growth is associated with a more epithelial phenotype.

(a–d) Representative images of metastases and primary tumour stained for DAPI (blue), YFP (red) and ECAD (a), CLDN7 (b), FSP1 (c) and ZEB1 (d) (green). Scale bars, 50 μm. (e) Quantification of EMT in metastases. The percent of positive cells for each lesion was determined and the percentages were averaged across lesions of the same size (ECAD, P<0.05; CLDN7, NS; FSP1, P<0.05; ZEB1, NS). n≥8 mice, ≥100 lesions for each stain. Data are presented as mean±s.d. P-values were calculated by one-way ANOVA and one sample t-tests against ‘single cell' means; *P<0.05.

Figure 3. Desmoplasia accumulates as lesions grow.

(a) Representative images of metastases and primary tumour stained for DAPI (blue), YFP (red) and α-SMA (green). Scale bars, 50 μm. (b) Contact between metastases and α-SMA⁺ fibroblasts. Each lesion was binned by size and scored for direct contact with an α-SMA⁺ cell (n=9 mice, 167 lesions). (c) α-SMA⁺ fibroblast density at metastatic lesions. Percent α-SMA⁺ area was quantified within one cell diameter of metastatic lesions (n=9 mice, 167 lesions). (d–g) Extracellular matrix (ECM) density at metastatic lesions. Metastatic livers were stained for ECM components COL1 (d), HABP (e), SPARC (f) and FN (g). Percent positive area was quantified within one cell diameter of each lesion. n≥5 mice, ≥50 lesions for each stain. Data are presented as mean±s.d. P-values were calculated by one-way ANOVA and one sample t-tests against ‘single cell' means; *P<0.05; **P<0.01; ^#P<0.001; ^##P<0.0001.

Figure 4. Metastatic growth is associated with functional hypovascularity.

(a) Representative images of metastases and primary tumour stained for VECAD (green), YFP (red) and DAPI (blue). Scale bars, 50 μm. (b) Metastatic vessel density. VECAD⁺ blood vessels were quantified at metastases within × 40 fields. Data are presented as mean±s.d.; n =7 mice, 100 lesions. (c) Proximity to blood vessels. For each lesion, the distance between five random tumour cells and the nearest VECAD⁺ blood vessel was determined. Data are presented as mean±s.d. n=7 mice, 100 lesions. P-values were calculated by one-way ANOVA and one sample t-tests against ‘single cell' means. (d) Functional hypovascularity at large metastases. Representative images of fluorescent dextran accumulation (red) at metastatic lesions (YFP, green; DAPI, blue). (e) Representative images of human primary PDAC and matched liver metastases stained for CD31 (brown). Scale bars, 50 μm. (f) Mean vessel density quantification of human primary PDAC and matched liver metastases (n=25 cases). Data are presented as mean±s.d.; *P<0.05; **P<0.01; ^#P<0.001; ^##P<0.0001.

Figure 5. Chemotherapy reduces metastatic burden.

(a) Liver metastatic burden after long-term chemotherapy. Each dot represents one mouse. Untreated group consists of historical controls from Fig. 1b (untreated, n=23; treated, n=8). Bars represent means±s.d.; *P<0.05, Student's t-test with Welch's correction. (b) Size distribution of metastatic lesions after long-term chemotherapy. Data are presented as mean±s.d., *P<0.05, Student's t-test with Welch's correction. (c) Apoptosis rates in metastases after single dose chemotherapy. CC3 positivity in metastases was assessed 10-12 h after treatment. Each dot represents a lesion; line represents the mean. *P<0.05 by Mann-Whitney test.