The response to neoadjuvant chemoradiotherapy with 5-fluorouracil in locally advanced rectal cancer patients: a predictive proteomic signature

Abstract

Background: Colorectal cancer is the third most common and the fourth most lethal cancer in the world. In the majority of cases, patients are diagnosed at an advanced stage or even metastatic, thus explaining the high mortality. The standard treatment for patients with locally advanced non-metastatic rectal cancer is neoadjuvant radio-chemotherapy (NRCT) with 5-fluorouracil (5-FU) followed by surgery, but the resistance rate to this treatment remains high with approximately 30% of non-responders. The lack of evidence available in clinical practice to predict NRCT resistance to 5-FU and to guide clinical practice therefore encourages the search for biomarkers of this resistance.

Methods: From twenty-three formalin-fixed paraffin-embedded (FFPE) biopsies performed before NRCT with 5-FU of locally advanced non-metastatic rectal cancer patients, we extracted and analysed the tumor proteome of these patients. From clinical data, we were able to classify the twenty-three patients in our cohort into three treatment response groups: non-responders (NR), partial responders (PR) and total responders (TR), and to compare the proteomes of these different groups.

Results: We have highlighted 384 differentially abundant proteins between NR and PR, 248 between NR and TR and 417 between PR and TR. Among these proteins, we have identified many differentially abundant proteins identified as having a role in cancer (IFIT1, FASTKD2, PIP4K2B, ARID1B, SLC25A33: overexpressed in TR; CALD1, CPA3, B3GALT5, CD177, RIPK1: overexpressed in NR). We have also identified that DPYD, the main degradation enzyme of 5-FU, was overexpressed in NR, as well as several ribosomal and mitochondrial proteins also overexpressed in NR. Data are available via ProteomeXchange with identifier PXD008440.

Conclusions: From these retrospective study, we implemented a protein extraction protocol from FFPE biopsy to highlight protein differences between different response groups to RCTN with 5-FU in patients with locally advanced non-metastatic rectal cancer. These results will pave the way for a larger cohort for better sensitivity and specificity of the signature to guide decisions in the choice of treatment.

Keywords: 5-Fluorouracil; Mass spectrometry; Predictive biomarkers; Proteomics; Rectal cancer; Resistance to neoadjuvant radio-chemotherapy.

Fig. 1

Overview of the workflow to obtain the group-specific proteomic profile. a Clinical selection and treatment. Patients with locally advanced non-metastatic colorectal cancer (CRC) are treated by radiotherapy (RT) with a concomitant chemotherapy (CT), in a neoadjuvant situation (preoperatively) (NRCT). (*) The doses indicated are the standard doses applied. The dose of RT is 45 to 54 Gray (Gy) in total, divided into 25 fractions (5 days per week for 5 weeks). The concomitant CT is 1600 mg/m² per day of RT. It can be either intravenous (5-fluorouracil, 5-FU) or oral (Capecitabine or Xeloda^®) (in the majority of cases, unless contraindications). After treatment, patients are divided in three groups according to their response to the treatment: no response (NR; absence of decreased stage/size of the tumor), partial response (PR; decreased stage/size of the tumor), and complete response (TR; elimination of the tumor). b From the initial sample to the FFPE biopsy: the physiopathological step. The fixation in formalin is immediately realized after the biopsy. The degreasing and dehydration are necessary for the inclusion in paraffin. The cutting and H&E staining steps allowed the pathologist to select the areas of interest and make a punch to obtain the FFPE biopsy (more details in Fig. 2). Punches are realized with TMA Master II (3DHISTECH). All these steps are realized in the laboratory of pathology. c From the FFPE biopsy to its proteomic profile: the experimental step. First steps (dewaxing, homogenisation and concentration) are necessary to realize a dithiothréitol reduction and an iodoacetamide alkylation on the samples. These last are in-gel separated and trypsin digested, and then the peptides are extracted and purified before their separation on HPLC–MS/MS. Analysis are realized with different software such as MaxQuant software package version 1.5.2.8 [18] and Perseus software package [19] and allowed to establish a proteomic profile specific for each group of responders. 5-FU 5-fluorouracil, CRC colorectal cancer, CT chemotherapy, FFPE formalin-fixed paraffin-embedded, Gy Gray, H&E haematoxylin and eosin, HPLC–MS/MS high performance liquid chromatography separation coupled to mass spectrometry, NR non-responders, PR partial responders, RT radiotherapy, TMA tissue microarray, TR total responders

Fig. 2

Cut obtained from a FFPE biopsy performed in a patient with locally advanced non-metastatic CRC (magnification ×1.5). a Selecting the area of interest containing the tumor cells by the pathologist. b Punch of the region of interest with the TMA Master II (3DHISTECH). CRC colorectal cancer, FFPE formalin-fixed paraffin-embedded, TMA tissue microarray

Fig. 3

Distribution of protein intensities and multi-scatter plot for each biopsy. For each biopsy, on the left, histograms representing the distribution of protein intensities obtained (n = 1 and n = 2). The colors identify treatment response groups: total responders (TR) in green, partial responders (PR) in yellow and non-responders (NR) in red. For each biopsy, on the right, multi-scatter plot comparing the intensities obtained for n = 1 and n = 2. The Pearson correlation coefficient r is indicated in blue for each multi-scatter plot. NR non-responders, PR partial responders, TR total responders

Fig. 4

Heatmaps of differentially abundant proteins between groups of response and PCA plot showing the separation of these three groups. a Comparison between NR, PR and TR groups (NR non-responders, PR partial responders, TR total responders) resulting in 402 differentially abundant proteins. b Comparison between NR and PR groups resulting in 384 differentially abundant proteins (180 over-represented in NR and 204 in PR). c Comparison between NR and TR groups resulting in 248 differentially abundant proteins (139 over-represented in NR and 109 in TR). d Comparison between PR and TR groups resulting in 417 differentially abundant proteins (294 over-represented in PR and 123 in TR). Statistical tests used for these analyses are ANOVA test (1) and Student’s T test (2) both corrected with a p value of 0.05. e Principal component analysis (PCA) showing the separation between the three groups of response: TR in green diamond, PR in yellow circle and NR in red square. NR non-responders, PCA principal component analysis, PR partial responders, TR total responders

Fig. 5

Protein and gene expression profile in normal rectal tissue and heat map for CALD1, CPA3, B3GALT5, CNNM4, MTIF3, CD177 and RIPK1 gene sequence variants in 22 colorectal cancer cell lines. a Protein expression profile in rectal tissue. The values are arbitrary units (0: undetected protein, 4: low expression, 8: medium expression, 12: high expression). These data were obtained from Human Protein Atlas available from http://www.proteinatlas.org. b Gene expression profile in rectal tissue. RNA-seq tissue data is reported as mean TPM (transcripts per million). These data were obtained from Human Protein Atlas available from http://www.proteinatlas.org. c Heat map of gene sequence variants in 22 colorectal cancer cell lines (in green: no sequence variants reported in coding region; in red: sequence variants reported in coding region). These data were obtained from http://colonatlas.org/index.html [23]

Fig. 6

Representation by STRING software of differentially abundant proteins overexpressed in NR compared to TR (https://string-db.org/). The nature of the interactions is shown at the bottom left of the figure and lists the known interactions (from curated database, experimentally determined), the predicted interactions (gene neighbourhood, gene fusion, gene co-occurrence) and the others (text mining, co-expression, protein homology). Overexpressed proteins were found in NR versus TR with a ribosomal proteins (RPS and RPL), b NDUF mitochondrial respiratory chain complex I proteins and c DPYD, the enzyme responsible for 5-FU hepatic catabolism. 5-FU 5-fluorouracil, DPYD dihydropyrimidine dehydrogenase, NR non-responders, RPL 40S ribosomal protein, small subunit, RPS 60S ribosomal protein, large subunit, TR total responders