Colorectal cancer susceptibility loci as predictive markers of rectal cancer prognosis after surgery

Abstract

To understand the molecular mechanism of rectal cancer and develop markers for disease prognostication, we generated and explored a dataset from 243 rectal cancer patients by gene expression microarray analysis of cancer samples and matched controls, and SNP-arrays of germline DNA. We found that two of the loci most strongly linked with colorectal cancer (CRC) risk, 8q24 (upstream of MYC) and 18q21 (in the intron of SMAD7), as well as 20q13 (in the intron of LAMA5), are tightly associated with the prognosis of rectal cancer patients. For SNPs on 18q21 (rs12953717 and rs4464148) and 20q13 (rs4925386), alleles that correlate with higher risk for the development of CRC are associated with shorter disease free survival (DFS). However, for rs6983267 on 8q24, the low risk allele is associated with a higher risk for recurrence and metastasis after surgery, and importantly, is strongly correlated with the resistance of CRC cell lines to chemoradiotherapy (CRT). We also found that although MYC expression is dramatically increased in cancer, patients with higher levels of MYC have a better prognosis. The expression of SMAD7 is weakly correlated with DFS. Notably, the presence of the 8q24 and 18q21 SNP alleles is not correlated with expression levels of MYC and SMAD7. rs4464148, and probably rs6983267 and rs4925386, are linked with overall survival time of patients. In conclusion, we show that several CRC risk SNPs detect subpopulations of rectal cancer patients with poor prognosis, and that rs6983267 probably affects prognosis through interfering with the resistance of cancer cells to CRT.

Figure 1. Principal component analysis (PCA) of expression levels of rectal cancer samples and matched normal mucosa

PCA of gene expression of both control samples (green) and rectal cancer samples (red) (A), of expression fold change of rectal cancer samples compared with paired controls (B-I). Samples with (C) RAS mutation (red) or without mutation (green), with lymph node metastasis (red) or without (green) before (D) or after (E) CRT, (F) with depth of invasion before CRT including T2 (green), T3 (yellow) or T4 (crimson), (G) with depth of invasion after CRT including T0 (green), T1 (blue), T2 (yellow), T3 (red) or T4 (crimson), (H) with recurrence after surgery (red) or disease free (green), (I) death due to CRC (red) or survival (green).

Figure 2. Influence of CRC development associated SNPs on DFS of rectal cancer patients and on CRT resistance of CRC cell lines

Kaplan-Meier curves of DSF of 8q24 SNP rs6983267 (A), 18q21 SNPs rs12953717 (B) rs4464148 (C) and rs4939827 (D), 20q13 SNP rs4925386 (E) for rectal cancer patients. (F) Genotype of rs6983267 for CRC cell lines. Cell lines are ordered according to their CRT resistance as in Spitzner, et al. (2014) Figure C. Genotypes unfavorable to rectal cancer patients are highlighted with gray. See supplementary figure 2 for data of all 5 SNPs.

Figure 3. Expressions levels of MYC, SMAD7 and LAMA5 have no association with adjacent significant SNPs

Fold change of expression of MYC (A), SMAD7 (B) and LAMA4 (C) of rectal cancer samples comparing with paired controls. Expression of MYC in cancer samples with different alleles of rs6983267 (D), SMAD7 with different alleles of rs12953717 (E) rs4464148 (F) and rs4939827 (G) and LAMA5 with different alleles of rs4925386 (H). One way Anova P value was calculated.

Figure 4. Low MYC expression is significantly associated with disease free time of rectal cancer patients

Kaplan-Meier curves of disease free time of high and low expression of MYC (A), SMAD7 (B) and LAMA5 (C). Kaplan-Meier curves of survival of high and low expression of MYC (D), SMAD7 (E) and LAMA5 (F).