Vitamin D Receptor Expression Is Associated with PIK3CA and KRAS Mutations in Colorectal Cancer

The vitamin D receptor (VDR) is a member of the steroid hormone receptor superfamily and regulates gene expression in a ligand-dependent manner (1). Plasma vitamin D level and estimated whole-body vitamin D level have been associated with decreased incidence and mortality in various cancers (1), including colorectal cancer (2, 3). Vitamin D and VDR seem to play important roles in preventing tumor development and progression through induction of cellular differentiation and inhibition of proliferation (1, 4-9). VDR is overexpressed or repressed in various human cancers (1, 10-15). In colorectal cancer, VDR is expressed in early-stage neoplasias but repressed in high-grade and metastatic cancers (12).

An activation of epidermal growth factor receptor (EGFR) triggers a chain of downstream signaling events that include the phosphatidylinositol 3-kinase (PI3K)–AKT and RAS–mitogen-activated protein kinase (MAPK) pathways. Accumulating evidence suggests that VDR binds to PI3K and inhibits cell proliferation by inducing differentiation (1, 16) and that the PI3K-AKT pathway is activated by 1,25(OH)₂D, the active form of vitamin D (17). In addition, VDR expression has been associated with RAS-MAPK pathway activation in leukemia cells (18). These observations suggest that VDR-expressing cells may need activation of the PI3K-AKT or RAS-MAPK pathway to acquire malignant characteristics. Activating mutations in PIK3CA and KRAS play important roles in colorectal carcinogenesis. Thus, we hypothesized that VDR expression is associated with PIK3CA and KRAS mutations in colorectal cancer. If this hypothesis is true, therapy or chemoprevention targeting VDR and downstream pathways may be influenced by PIK3CA or KRAS mutations. In addition, VDR-mediated action of 1,25(OH)₂D can limit colon cancer cell growth particularly when induced by activation of EGFR (19). Thus, response to EGFR-targeted therapy may be modified by VDR status in tumor cells.

We therefore used 619 colorectal cancers identified in two prospective cohort studies and examined the relation of VDR expression with patient survival, PIK3CA and KRAS mutations, and other related molecular features, including BRAF mutation, microsatellite instability (MSI), and the CpG island methylator phenotype (CIMP), which were potential confounders.

We used the databases of two independent prospective cohort studies: the Nurses' Health Study (n = 121,700 women followed since 1976; ref. 20) and the Health Professionals Follow-up Study (n = 51,500 men followed since 1986; ref. 20). Data on height and weight were obtained by biennial questionnaire. A subset of the cohort participants developed colorectal cancers during prospective follow-up. Data on tumor location and tumor-node-metastasis stage were obtained through medical record review by study physicians. We collected paraffin-embedded tissue blocks from hospitals where cohort participants with colorectal cancers had undergone resections of primary tumors. Up to 2002, there were 1,834 incident colorectal cancer cases with adequate clinical information. We successfully retrieved formalin-fixed, paraffin-embedded tumor tissue in 998 cases (54%). We have previously shown that there are no differences in demographic, nutritional, or exposure features between patients with and without available tumor tissue (20). Among 998 cases with available tumor tissue, we were able to construct tissue microarrays using 625 cases. Among them, we obtained valid VDR expression data in 619 cases, which were eligible for the current study (Table 1). In any of our previous studies, we have not examined VDR expression in colorectal cancers. This current analysis represents a new analysis of VDR on the existing colorectal cancer database that has been previously characterized for MSI, KRAS, PIK3CA, and BRAF (21, 22), which is analogous to novel analysis using the well-characterized cell lines or animal models. Patients were observed until death or June 30, 2006, whichever came first. Ascertainment of deaths included reporting by the family or postal authorities. In addition, the names of persistent nonresponders were searched in the National Death Index. More than 98% of deaths in the cohorts were identified by these methods. The cause of death was assigned by physicians blinded to information on lifestyle exposures and molecular features in colorectal cancer. Written informed consent was obtained from all study subjects. Tissue collection and analyses were approved by the Harvard School of Public Health and Brigham and Women's Hospital Institutional Review Boards.

H&E-stained tissue sections were examined by a pathologist (S.O.) unaware of other data. The tumor grade was categorized as low (≥50% gland formation) versus high (<50% gland formation). The presence and extent of extracellular mucin and signet ring cells were categorized as 0% (no mucin or signet ring cells) or ≥1% of the tumor volume.

DNA was extracted from paraffin-embedded tumor tissue sections, and PCR and Pyrosequencing targeted for KRAS (codons 12 and 13; ref. 23), BRAF (codon 600; ref. 24), and PIK3CA (exons 9 and 20) were done as previously described (25). MSI analysis was done using 10 microsatellite markers (D2S123, D5S346, D17S250, BAT25, BAT26, BAT40, D18S55, D18S56, D18S67, and D18S487; ref. 26). MSI-high was defined as the presence of instability in ≥30% of the markers. MSI-low was defined as instability in 10% to 29% of the markers, and “microsatellite stable” (MSS) tumors were defined as tumors with no unstable marker.

Sodium bisulfite treatment on genomic DNA and subsequent real-time PCR (MethyLight) were validated and done as previously described (27). We quantified DNA methylation in eight CIMP-specific promoters [CACNA1G, CDKN2A (p16), CRABP1, IGF2, MLH1, NEUROG1, RUNX3, and SOCS1; refs. 28-30]. CIMP-high was defined as the presence of six or more of eight methylated promoters, CIMP-low as the presence of one of eight to five of eight methylated promoters, and CIMP-0 as the absence (zero of eight) of methylated promoters, according to the previously established criteria (30). To accurately quantify relatively high methylation levels in LINE-1 repetitive elements, we used Pyrosequencing (31).

Tissue microarrays were constructed as previously described (20). Methods of immunohistochemical procedures and interpretations were previously described for p53, p21 (32, 33), β-catenin (34), and cyclooxygenase-2 (COX-2; refs. 20, 26). Appropriate positive and negative controls were included in each run of immunohistochemistry for all markers. For VDR immunohistochemistry, antigen retrieval was done, and deparaffinized tissue sections in Antigen Retrieval Citra Solution (Biogenex Laboratories) were treated with microwave for 15 min. Tissue sections were incubated with 3% H₂O₂ (10 min) to block endogenous peroxidase (DakoCytomation), with 10% normal goat serum (Vector Laboratories) in PBS (10 min). Primary antibody against VDR [polyclonal rabbit anti-VDR (C-20), 1:200 dilution; Santa Cruz Biotechnology] was applied, and the slides were maintained for 1 h at room temperature. Next, we applied Envision System horseradish peroxidase–labeled polymer anti-rabbit (Dako) for 30 min followed by visualizing signal with diaminobenzidine (5 min) and methyl green counterstain. Appropriate positive and negative controls were included in each run of VDR immunohistochemistry. VDR overexpression (Fig. 1) was evaluated by one of the investigators (K.No.) unaware of other data. Cytoplasmic VDR expression was recorded as no, weak, or moderate/strong expression, and the proportion of cells with expression was recorded. Among the 619 tumors, 82 showed strong cytoplasmic expression, 139 showed moderate cytoplasmic expression, 12 showed ≥50% of tumor cells with weak cytoplasmic expression, 53 showed 30% to 49% of tumor cells with weak cytoplasmic expression, 112 showed 1% to 29% of tumor cells with weak cytoplasmic expression, and 221 tumors showed no staining in tumor cells. VDR positivity (i.e., overexpression) was defined as moderate/strong cytoplasmic expression in any fraction of tumor cells or at least 50% of tumor cells with weak cytoplasmic expression. Either the absence of staining or the presence of weak cytoplasmic expression in <50% of tumor cells was interpreted as negative. This cut point was based on the frequency of PIK3CA mutation in colorectal cancer groups categorized by VDR status. The frequency of PIK3CA mutation was 21% (40 of 191) in tumors with moderate or strong cytoplasmic expression, 36% (4 of 11) in tumors with weak cytoplasmic expression in ≥50% of tumor cells, 11% (5 of 47) in tumors with weak cytoplasmic expression in 30% to 49% of tumor cells, 12% (12 of 103) in tumors with weak cytoplasmic expression in 1% to 29% of tumor cells, and 11% (23 of 202) in tumors with no staining.

A random selection of 139 cases was examined for VDR by a second observer (Y.B.) unaware of other data, and concordance between the two observers was 0.81 (κ = 0.62, P < 0.0001), indicating substantial agreement. Each of the other immunohistochemical markers was interpreted by a pathologist unaware of other data (p53, p21, and COX-2 by S.O.; β-catenin by K.No.). For each of the other immunohistochemical markers, a second observer (S.O. for β-catenin; K.S. for p21; K.No. for p53; R. Dehari, Kanagawa Cancer Center, for COX-2) examined a random sample of 108 to 402 tumors, unaware of other data. The κ coefficient between the two observers was 0.65 for β-catenin (P < 0.0001; n = 402), 0.62 for p21 (P < 0.0001; n = 179), 0.75 for p53 (P < 0.0001; n = 118), and 0.62 for COX-2 (P < 0.0001; n = 108), indicating substantial agreement.

For all statistical analyses, we used SAS program (version 9.1; SAS Institute). All P values were two sided, and statistical significance was set at P ≤ 0.05; however, whenever appropriate, P values were conservatively interpreted considering multiple hypotheses testing. For categorical data, the χ² test (or Fisher's exact test when any expected cell count was less than five) was done. To assess independent relations of VDR with several variables, a multivariate logistic regression analysis was done. Odds ratio (OR) was adjusted for age at diagnosis (<65 versus ≥65 y), sex, tumor location (proximal versus distal), body mass index (BMI; ≥30 versus <30 kg/m²), tumor stage (I-II versus III-IV), grade (low versus high), family history of colorectal cancer in any first-degree relative (present versus absent), mucinous component (present versus absent), CIMP status (high versus low/0), MSI status (high versus low/MSS), LINE-1 methylation (continuous), β-catenin, COX-2, p53, p21, BRAF, KRAS, and PIK3CA. For cases with missing data on KRAS (0.3% missing) and PIK3CA (11% missing), we assigned separate (“missing”) indicator variables and included those cases in the multivariate model. For cases with missing information in other variables [BMI (5.8% missing), tumor location (1.1%), tumor stage (11%), tumor grade (3.2%), mucinous component (13%), CIMP (2.6%), MSI (0.5%), LINE-1 (3.7%), BRAF (2.9%), β-catenin (11%), p53 (0.8%), p21 (2.3%), and COX-2 (0.6%)], we included those cases in a majority category to minimize the number of indicator variables. We confirmed that excluding cases with missing information in any of the covariates did not substantially alter results (data not shown).

For survival analysis, Kaplan-Meier method was used and log-rank test was used to test significance of a deviation from the null hypothesis. For analyses of colorectal cancer–specific mortality, death as a result of colorectal cancer was the primary end point and deaths as a result of other causes were censored. To assess independent effect of VDR status on mortality, we constructed a multivariate, stage-matched (stratified) Cox proportional hazards model to compute a hazard ratio (HR) according to VDR status, adjusted for age at diagnosis, sex, BMI, year of diagnosis, tumor grade, tumor location, mucinous component, family history, CIMP, MSI, KRAS, BRAF, PIK3CA, p53, β-catenin, p21, COX-2, and LINE-1 methylation. Tumor stage (I, IIA, IIB, IIIA, IIIB, IIIC, IV, unknown) was used as a matching variable using the “strata” option in the SAS “proc phreg” command to minimize residual confounding and overfitting. We also used stage-matched Cox model including only stage as a stratifying variable without any other covariate in the model. The proportionality of hazards assumption was satisfied by evaluating time-dependent variables, which were the cross product of the VDR variable and survival time (P = 0.62 for colorectal cancer–specific mortality; P = 0.85 for overall mortality). Cases with missing data in any of the covariates were dealt as in the multivariate logistic regression analysis described above, except for stage, KRAS, and PIK3CA. For cases with missing data on KRAS (or PIK3CA), we included those cases in the wild-type category. An interaction was assessed by including the cross product of the VDR variable and another variable of interest (excluding data-missing cases) in a multivariate Cox model, and the Wald test was done.

We examined VDR overexpression by immunohistochemistry in 619 colorectal cancers identified in two independent prospective cohort studies and detected VDR overexpression in 233 (38%) tumors. Table 1 shows the frequencies of VDR expression according to various clinical and pathologic features. VDR overexpression was significantly less common among high-grade tumors [OR, 0.41; 95% confidence interval (95% CI), 0.20-0.81; P = 0.008].

Table 1 shows the frequencies of VDR overexpression according to various molecular features in colorectal cancer. VDR overexpression was significantly more common among KRAS-mutated tumors (OR, 1.55; 95% CI, 1.11-2.16; P = 0.010) and PIK3CA-mutated tumors (OR, 2.17; 95% CI, 1.36-3.47; P = 0.001). To examine combined effect of KRAS and PIK3CA mutations on VDR overexpression, we classified tumors into four subtypes according to KRAS and PIK3CA status (Table 1). VDR overexpression was significantly more common in KRAS-mutated/PIK3CA-mutated tumors (58%, 29 of 50; OR, 2.97; 95% CI, 1.61-5.47; P = 0.0003) than in KRAS wild-type/PIK3CA wild-type tumors (32%, 99 of 312). We also examined the frequency of PIK3CA or KRAS mutation according to VDR staining intensity in tumor cells (Table 2). The frequencies of PIK3CA and KRAS mutations increased as intensity of VDR staining increased, and reached plateau at moderate intensity of staining.

We determined CIMP status using MethyLight assays on a panel of eight CIMP-specific promoters (CACNA1G, CDKN2A, CRABP1, IGF2, MLH1, NEUROG1, RUNX3, and SOCS1; refs. 29, 30). VDR overexpression was not significantly associated with CIMP (Table 1). On the other hand, VDR overexpression was more common in MSI-low tumors (51%, 30 of 59; OR, 1.80; 95% CI, 1.05-3.11; P = 0.032) than in MSS tumors (36% = 168/461); nonetheless, this could be a chance association given multiple hypothesis testing on multiple tumor markers (excluding PIK3CA and KRAS).

We did multivariate logistic regression analysis to confirm independent relations between VDR and mutations of KRAS and PIK3CA after adjusting for clinical, pathologic, and other molecular variables (Table 3). VDR overexpression was significantly associated with PIK3CA mutation (adjusted OR, 2.36; 95% CI, 1.43-3.91; P = 0.0008) and KRAS mutation (adjusted OR, 1.53; 95% CI, 1.04-2.23; P = 0.029).

Using our cohort database, we previously showed that molecular features in colon cancer, such as BRAF mutation, PIK3CA mutation, and LINE-1 hypomethylation, were significantly associated with inferior prognosis (21, 22, 35). We assessed the influence of VDR overexpression on survival of patients with stage I to IV colorectal cancers. During follow-up of 599 patients who were eligible for survival analysis, there were 260 deaths, including 158 colorectal cancer–specific deaths. In Kaplan-Meier analysis, VDR status was not significantly associated with colorectal cancer–specific survival (log-rank P = 0.57) or overall survival (log-rank P = 0.31; Fig. 2).

We did Cox regression analysis to assess mortality according to VDR status (Table 4). VDR status was not significantly related with patient survival in univariate analysis, stage-matched analysis, or multivariate analysis in colorectal cancer. We also examined whether any of the clinical, pathologic, and molecular variables significantly modified the effect of VDR overexpression on patient survival. There was no evidence for a significant interaction between VDR overexpression and any of the variables examined, including KRAS and PIK3CA (all P_interaction > 0.05).

We conducted this study to examine the relationship between VDR expression and mutations in PIK3CA and KRAS in colorectal cancer. Accumulating evidence indicates a substantial role of vitamin D in prevention of various forms of human cancer (1, 2, 17, 36-38). A potential link between VDR and the PI3K-AKT or RAS-MAPK pathway has been suggested (17, 18). Thus, examining VDR in colorectal cancer may shed lights on biological mechanisms of vitamin D action and its failure. We have found that VDR overexpression in colorectal cancer is significantly associated with both PIK3CA and KRAS mutations independent of other clinical and molecular features. Our data support the hypothesis that the VDR pathway interacts with the PI3K-AKT and RAS-MAPK pathways in colonic neoplastic cells.

Our resource of a large number of colorectal cancers derived from the two prospective cohort studies has enabled us to precisely estimate the frequency of colorectal cancers with a specific molecular feature (such as VDR expression, KRAS mutation, and PIK3CA mutation). The large number of cases has also provided a sufficient power in our multivariate logistic regression analysis and survival analysis.

Studying risk-modifying factors or molecular changes is important in cancer research (39-44). Previous studies have consistently reported the preventive effect of vitamin D on colorectal cancers (1, 2, 36-38) and that higher plasma levels of vitamin D confer a greater reduction in the risk of colorectal cancer (2). Studies have reported that 1,25(OH)₂D potentiates the effects of many cytotoxic and antiproliferative drugs (1, 17) and that higher plasma levels of 25(OH)D are associated with a significant reduction in colon cancer mortality (2). Thus, accumulating evidence indicates important roles of vitamin D in preventing the development and progression of colorectal cancer. Interestingly, VDR and 1-α-hydroxylase (encoded by CYP27B1), which converts 25(OH)D into 1,25(OH)₂D, are frequently overexpressed in colon cancer cells (1, 4-6, 8, 9). The antiproliferative action of 1,25(OH)₂D seems to depend on VDR expression level and differentiation status of tumor cells (45, 46). Therefore, it is possible that the effect of serum vitamin D level, which is protective against cancer incidence and mortality, may differ according to VDR expression status in colorectal cancer. Additional studies are necessary to address this intriguing hypothesis.

A potential link between VDR and the PI3K-AKT or RAS-MAPK pathway has been suggested (16-18, 47-51). In these pathways, PIK3CA or KRAS mutation plays an important role in the progression of colorectal cancer. Mutant PIK3CA stimulates the downstream AKT pathway and promotes cell growth in various cancers, including colorectal cancer. The PI3K-AKT pathway has been known to mediate signals from growth factors, which is influenced by the state of energy balance. In addition, PI3K-AKT signaling is influenced by KRAS (52). In fact, PIK3CA mutation is positively associated with KRAS mutation in colorectal cancer (25, 53). Our data support a potential link between PIK3CA and VDR and suggest that VDR expression may affect the regulation of PI3K-AKT pathway in colorectal cancer. In myeloid leukemia cells, VDR activated by 1,25(OH)₂D can inhibit tumor cell proliferation by inducing differentiation, which depends on the formation of activated VDR and PI3K complexes (1, 16). A combination of 1,25(OH)₂D with AKT pathway inhibitors is strongly antiproliferative and should be considered for differentiation therapy of myeloid leukemia (17). These previous observations (16, 17) and our data suggest that VDR-expressing cells may need the activation of PI3K-AKT pathway to acquire malignant characteristics and that therapy or chemoprevention targeting VDR and downstream pathways may be influenced by PIK3CA mutation in colon cancer cells.

The vitamin D pathway may also interact with RAS signaling. A recent case-control study has reported that the VDR polyadenylic acid, rs10735810 (so-called “FokI SNP”), and rs11568820 (so-called “CDX2 SNP”) polymorphisms are associated with KRAS mutation in colon cancer (54). VDR protein expression is down-regulated in KRAS-mutated colon cancer cells (47). In contrast, VDR expression has been associated with the activation of the RAS-MAPK pathway in leukemia cells (18), which agrees with our current data. In addition, RAS-transformed human keratinocytes are shown to be resistant to the growth-inhibitory effects of 1,25(OH)₂D (48-50). Further analysis is needed to clarify how vitamin D and its downstream pathway influence colorectal cancer development in relation to KRAS mutation.

We did not observe a significant relation between VDR expression and MSI-high or the CIMP in colorectal cancer. A molecular classification of colorectal cancer based on MSI and CIMP is increasingly important (55) because MSI and CIMP status represent genomic and epigenomic alterations, respectively, in tumor cells and largely determine clinical, pathologic, and molecular characteristics (55). Nonetheless, our data do not support a link between VDR and CIMP or MSI in colorectal cancer. A recent study has reported that the VDR rs10735810 (FokI SNP) polymorphism is significantly associated with CIMP-high and inversely with MSI-high (54); however, either of these could be a chance association given multiple hypothesis testing.

In the current study, we have shown that VDR expression is inversely associated with high-grade tumors in univariate analysis but not in multivariate analysis. Thus, our data are not incompatible with the inverse association between VDR expression and high tumor grade in the previous study (12) but do not support a mechanistic link between VDR loss and high tumor grade. With regard to disease stage, the previous study (12) has shown that VDR expression in colon cancer cells is commonly lost in a metastatic focus but not in a primary lesion. Therefore, the absence of the relation between loss of VDR expression in primary tumor and advanced disease stage in the current study is not incompatible with the previous data (12).

As one limitation in our cohort studies, data on cancer treatment were limited. Nonetheless, it is unlikely that chemotherapy use differed according to tumoral VDR status because such data were not available to patients or treating physicians. In addition, beyond cause of mortality, data on cancer recurrences were not available in these cohorts. Nonetheless, given that the median survival for metastatic colorectal cancer was approximately 10 to 12 months during much of the time period of this study, colorectal cancer–specific survival should be a reasonable surrogate for colorectal cancer–specific outcomes.

As another limitation of this study, there are currently no standardized methods to assess VDR overexpression in colorectal cancer. Thus, our current study is exploratory by nature, and our data and method to determine a cut point for VDR overexpression need to be validated and confirmed by independent data sets.

In summary, this large cohort of colorectal cancers has shown that VDR expression is significantly associated with PIK3CA and KRAS mutations independent of clinical, pathologic, and molecular features. On the other hand, VDR expression is not significantly related with patient survival. Our data support the hypothesis that PIK3CA and/or KRAS mutations may influence the biological effect of VDR and its downstream pathway. Thus, targeting VDR for chemoprevention or cancer therapy likely needs to consider the effect of KRAS or PIK3CA mutation. Likewise, therapy targeting EGFR or the downstream RAS or PI3K pathway may be influenced by VDR status. Considering that VDR regulates the transcription of various genes involved in cellular differentiation and inhibition of proliferation, our findings may have considerable clinical implications. Further studies are necessary to elucidate exact roles of vitamin D and VDR in prevention of colorectal neoplasias as well as to examine a potential mechanistic link between VDR and the RAS-MAPK and PI3K-AKT pathways.

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 the Nurses' Health Study and Health Professionals Follow-up Study cohort participants who have generously agreed to provide us with biological specimens and information through responses to questionnaires and Frank Speizer, Walter Willett, Susan Hankinson, Meir Stampfer, and many other staff members who implemented and have maintained the cohort studies.