PTPRJ Haplotypes and Colorectal Cancer Risk

Colorectal cancer (CRC) is the third most common cancer in the United States, not including skin cancer. Whereas mutations in genes such as APC, MLH1, MSH2, and MYH confer a high risk of colon cancer and track through families, they only account for ∼5% of all CRC cases (1). For individuals in the general population, the risk of developing CRC is due to a combination of environmental factors, such as diet, and low penetrance cancer susceptibility and resistance genes that act in the context of other genetic variants and environmental factors (2).

Mouse models of cancer susceptibility are proving useful for the identification of low penetrance cancer susceptibility genes (3, 4). Using mice susceptible to and mice resistant to CRC, Ptprj was identified as a strong candidate gene for Scc1, a colon cancer susceptibility locus (4). Ptprj is a good candidate for a cancer susceptibility gene. It has several functional roles consistent with tumor suppression including inhibition of cell growth, migration, and angiogenesis (5–8). It also has been shown to dephosphorylate a number of tyrosine kinase receptors important in carcinogenesis including PDGFRβ, RET, and VEGFR2 (7, 9, 10). PTPRJ also seems to play a role in human CRC. In loss of heterozygosity studies in humans, PTPRJ was deleted in 19 of 39 informative colorectal carcinomas (4). Allele-specific loss of PTPRJ was assessed, and 13 of 16 colon tumors heterozygous for an A1176C (Q276P; rs1566734) polymorphism showed loss of the A allele, suggesting that this variant may be a cancer resistance allele in humans. Other studies have reported PTPRJ loss as an early event in colorectal tumorigenesis. Forty-seven percent of adenomas and 10% of aberrant crypt foci show PTPRJ loss, providing evidence for a role in tumor initiation or promotion (11, 12). PTPRJ is a candidate tumor suppressor gene for other cancer types. Loss of 11p11, the locus harboring PTPRJ, is observed in 50% of lung tumors, 78% of breast tumors, and 38% of thyroid anaplastic carcinomas (4, 13). Based on PTRPJ loss of heterozygosity in breast tumors, PTPRJ haplotype tagging single nucleotide polymorphisms (SNP) were assessed for breast cancer risk in individuals from the United Kingdom. One haplotype showed a protective effect for breast cancer (14). In this study, there was no association for cancer risk with a R326Q variant (rs1503185) that was in 100% linkage disequilibrium with the A1176C allele. The A1176C variant was not directly tested.

Because PTPRJ shows frequent early loss in colon adenomas and aberrant crypt foci and the PTPRJ A1176C allele showed preferential allelic loss in colon tumors, we hypothesized that the A1176C might function as a low penetrance CRC susceptibility allele. To determine if the PTPRJ A1176C contributes to increased CRC risk, we tested this variant in 1,897 incidence cases of CRC and 1,954 controls from the Molecular Epidemiology of Colorectal Cancer study. To further clarify the risk between PTPRJ and CRC, we also did haplotype analysis of PTPRJ using six tagSNPs.

The Molecular Epidemiology of Colorectal Cancer study is a population-based, matched case-control study of 1,897 incident CRC cases and 1,954 controls described previously (15, 16). Eligible cases included any person newly diagnosed with CRC between January 1, 1999, and March 31, 2004, from northern Israel. Potential controls were identified from the same region by generating a list of individuals from the Clalit Health Services database, matched with respect to year of birth, sex, clinical code, and Jewish versus non-Jewish ethnicity. Written informed consent was obtained from all study participants. This study was approved by the Institutional Review Boards at the University of Michigan, the Ohio State University, and Carmel Medical Center in Haifa.

The PTPRJ A1176C variant and six SNPs described in Lesueur et al. (rs7122335, rs2904315, rs4752894, rs1503185, rs905476, and rs1566729; ref. 14) were genotyped using the ABI PRISM 7900 sequence detection system. χ² tests were used to test Hardy-Weinberg equilibrium in the controls. The genotypic specific colon cancer risks were estimated as odds ratios (OR) with associated 95% confidence intervals (CI) by unconditional logistic regression as implemented in SAS v.9.0. Haplotypes were estimated using the estimation-maximization likelihood method and score statistics calculated for significance as implemented in R using the haplo.stats library (17).⁶

We genotyped 1,897 cases and 1,954 controls for the PTPRJ A1176C variant. There were no appreciable differences in gender (male cases, 50.8%; male controls, 50.0%), mean age of diagnosis [cases, 70.1 years (±11.8 years); controls, 70.6 years (±11.8 years)] or in Ashkenazi Jewish heritage (cases, 68.9%; controls, 63.3%). There was no deviation of genotype frequencies in controls from those expected under Hardy-Weinberg equilibrium. After adjustment for ethnicity, age, and first-degree family history of CRC, there was no meaningful increased risk of CRC associated with the presence of the PTPRJ A1176C variant (heterozygote OR, 1.03; 95% CI, 0.90-1.18; homozygotes OR, 1.13; 95% CI, 0.88-1.44; Table 1).

To better understand the relationship between PTPRJ and CRC, we genotyped tagSNPs reported as contributing to a protective haplotype for breast cancer risk (14) in a subset of the sample. Five of the six tagSNPs are located in introns and have no known functional role. One SNP, rs1503185, results in an amino acid substitution of a Gln for an Arg at amino acid 326. The functional significance of this substitution is not known. Of the 1,042 cases and 920 controls genotyped for 6 tagSNPs, there were no significant differences in gender (male cases, 52.2%; male controls, 50.8%), mean age of diagnosis [cases, 72.9 (±10.1 y), controls 73.8 (±10.03 y)], or Ashkenazi Jewish heritage (cases, 92.5%; controls, 73.8%). None of the SNPs were individually associated with risk of CRC (Table 2), although one SNP, rs1503185, showed a trend toward being protective in the rare homozygote versus common homozygote comparison (OR, 0.63; 95% CI, 0.43-0.93).

The 6 SNPs generated 6 common (>1%) haplotypes representing ∼97% of all possible haplotypes in this population. One haplotype was overrepresented in cases when compared with controls, corresponding to a 34% increase in risk of CRC (Table 3), but there was no significant difference overall in haplotype frequencies between cases and controls (global test P statistical = 0.19). These results did not change when restricting the sample to subjects of Ashkenazi Jewish ethnicity or including the PTPRJ A1176C variant in the haplotype estimation (data not shown).

This is the first study, to our knowledge, that has tested the PTPRJ A1176C variant or tagSNPs in PTPRJ for CRC risk. Ptprj was identified as a low penetrance CRC cancer resistance allele in mice but had not previously been tested for risk in human CRC populations. In this large case/control population, we saw no evidence for increased CRC risk of the A1176C allele or combination of haplotype tagSNPs. These results also do not rule out the possibility that an untagged variant or haplotype confers risk, that variants in this gene are involved in epistatic interactions, which confer risk, or that population-specific difference or exposures in this study cohort, which consists of mostly Whites, individuals of Ashkenazi Jewish ancestry, may modify the effects of the PTPRJ A1176C variant. It is also possible that in a larger study population, the effects of increased risk associated with haplotype geno.glm.9 and the protective effect of rs1503185 could be established. Our study has exactly 80% power to detect a relative risk of similar magnitude for CRC to that previously reported for breast cancer, assuming a disease frequency of 6% and allele frequency of 17% for a log-additive model (18). As rs1503185 results in an amino acid substitution (QR326), this variant may have functional significance and therefore warrants further study.

The effects of PTPRJ on CRC risk may be modest or may be dependent on interacting genetic factors because other studies suggest these possibilities. In the crosses that led to the identification of Ptprj as a low penetrance CRC resistance gene, Ptprj interacts with two additional loci to further increase CRC risk (19). Given the mild protective effect seen for PTPRJ in over 4,000 breast cancer cases and controls (OR, 0.81; 95% CI, 0.72-0.92; ref 14) and the results presented here, a more comprehensive understanding of the genetic variability in PTPRJ in light of potential modifying factors, including genetic variation at potential interaction loci, is needed to better understand the cancer risk associated with PTPRJ.

In conclusion, this study showed no significant evidence for the A1176C allele of PTPRJ or previously described haplotypes of tagSNPS in PTPRJ on CRC risk.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Allan Balmain for thoughtful discussions.