Abstract
BACKGROUND
The transforming growth factor-β (TGF-β) signaling pathway is involved in many aspects of tumori-genesis, including angiogenesis and metastasis. The authors evaluated this pathway in association with survival after a diagnosis of colon or rectal cancer.
METHODS
The study included 1553 patients with colon cancer and 754 patients with rectal cancer who had incident first primary disease and were followed for a minimum of 7 years after diagnosis. Genetic variations were evaluated in the genes TGF-β1 (2 single nucleotide polymorphisms [SNPs]), TGF-β receptor 1 (TGF-βR1) (3 SNPs), smooth muscle actin/mothers against decapentaplegic homolog 1 (Smad1) (5 SNPs), Smad2 (4 SNPs), Smad3 (37 SNPs), Smad4 (2 SNPs), Smad7 (11 SNPs), bone morphogenetic protein 1 (BMP1) (11 SNPs), BMP2 (5 SNPs), BMP4 (3 SNPs), bone morphogenetic protein receptor 1A (BMPR1A) (9 SNPs), BMPR1B (21 SNPs), BMPR2 (11 SNPs), growth differentiation factor 10 (GDF10) (7 SNPs), Runt-related transcription factor 1 (RUNX1) (40 SNPs), RUNX2 (19 SNPs), RUNX3 (9 SNPs), eukaryotic translation initiation factor 4E (eiF4E) (3 SNPs), eukaryotic translation initiation factor 4E-binding protein 3 (eiF4EBP2) (2 SNPs), eiF4EBP3 (2 SNPs), and mitogen-activated protein kinase 1 (MAPK1) (6 SNPs).
RESULTS
After adjusting for American Joint Committee on Cancer stage and tumor molecular phenotype, 12 genes and 18 SNPs were associated with survival in patients with colon cancer, and 7 genes and 15 tagSNPs were associated with survival after a diagnosis of rectal cancer. A summary score based on “at-risk” genotypes revealed a hazard rate ratio of 5.10 (95% confidence interval, 2.56-10.15) for the group with the greatest number of “at-risk” genotypes; for rectal cancer, the hazard rate ratio was 6.03 (95% confidence interval, 2.83-12.75).
CONCLUSIONS
The current findings suggest that the presence of several higher risk alleles in the TGF-β signaling pathway increase the likelihood of dying after a diagnosis of colon or rectal cancer.
Keywords: colon cancer, rectal cancer, transforming growth factor-β, bone morphogenetic protein, smooth muscle actin/mothers against decapentaplegic homolog, Runt-related transcription factors, survival, prognosis
The transforming growth factor-β (TGF-β) signaling pathway is an essential regulator of cellular proliferation, differentiation, apoptosis, and extracellular matrix remodeling and is involved in angiogenesis and inflammation.1 TGF-β mediates its cellular effects through interaction with smooth muscle actin/mothers against decapentaplegic (Smad) proteins. Bone morphogenetic proteins (BMPs), another part of the TGF-β signaling pathway, trigger an Smad-signaling cascade that has been linked to reduced cell proliferation and cellular growth kinetics of glioblastomas.2,3 The Runt-related transcription factors (RUNX), including RUNX1, RUNX2, and RUNX3, also are related to the TGF-β signaling pathway. Studies in RUNX3 knockout mice have demonstrated apoptotic defects in response to TGF-β, and RUNX2 transgenic mice reportedly were hypersensitive to TGF-β.4 It has been demonstrated that all 3 RUNX genes bind Smads,5-7 further altering the TGF-β signaling pathway.
Mitogen-activated protein kinase 1 (MAPK1), also known as extracellular signal-regulated kinase 2 (ERK2), is involved in eukaryotic signal transduction. It has been demonstrated that MAPK1 activates RUNX28 and is involved in both TGF-β and Smad signaling.9,10 Eukaryotic translation initiation factor 4E (eIF4E) plays a key role in cell growth and reportedly is overexpressed in colon tumors.11 Studies indicate that the expression of eIF4E in human colon cancer cells promotes the TGF-β stimulation of adhesion molecules.12
The importance of the TGF-β signaling pathway in colon and rectal cancer has been supported in numerous studies; our previous work illustrated the risk associated with genetic variation in these genes in colon and rectal cancer etiology.13-15 It is plausible biologically that this pathway may influence prognosis as well as initial disease development given its role in influencing angiogenesis, inflammation, and growth factors. For the current study, we evaluated the associations between genetic variability in the TGF-β signaling pathway and survival after a diagnosis of colon cancer or rectal cancer. We evaluated genetic variation in key elements of the TGF-β signaling pathway, including TGFβ1 and its receptor, Smad genes, BMP genes, RUNX genes, MAPK1, and eiF4E and its binding proteins.
MATERIALS AND METHODS
Two study populations were included in these analyses. The first study, a population-based case-control study of colon cancer, included cases (n = 1553 patients with complete genotype data) who were identified between October 1, 1991 and September 30, 199416 and who lived in the Twin-Cities Metropolitan Area, in an area served by the Kaiser Permanente Medical Care Program of Northern California (KPMCP), and in a 7-county area of Utah. The second study, which used identical data-collection methods, included cases with cancer of the rectosigmoid junction or rectum (n = 754 patients with complete genotype data) who were identified between May 1997 and May 2001 in Utah and KPMCP.17 Eligible cases were between ages 30 years and 79 years at the time of diagnosis; English speaking; mentally competent to complete the interview; had no previous history of colorectal carcinoma (CRC); and had no reported familial adenomatous polyposis, ulcerative colitis, or Crohn disease. All study procedures were approved by ethics committees at all study locations.
Tumor Registry Data
Tumor registry data were obtained to determine disease stage at diagnosis and months of survival after diagnosis. Disease stage was categorized using the sixth edition of the American Joint Committee on Cancer (AJCC) staging criteria. Disease staging was done centrally by 1 pathologist in Utah. Local tumor registries also provided information on patient follow-up, including vital status, cause of death, and contributing cause of death. Follow-up was obtained for all study participants and was terminated for the colon cancer study in 2000 and for the rectal cancer study in 2007. At those times, all study participants had >5 years of follow-up.
Tumor Marker Data
We previously evaluated tumors for CpG island methylator phenotype, microsatellite instability (MSI) (colon only), tumor protein 53 (TP53) mutations, and v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS2) mutations.18-21 Because the tumor molecular phenotype may influence survival,18,20,22 we adjusted hazard rate ratios (HRRs) for tumor molecular phenotype.
SNP Selection and Genotyping
SNPs were selected using the following parameters: r2 = 0.8 defined LD blocks using a Caucasian LD map, a minor allele frequency >0.1, ranged −1500 base pairs from the initiation codon to +1500 base pairs from the termination codon, and 1 SNP/LD bin. All markers were genotyped using a multiplexed bead array assay format based on Golden Gate chemistry (Illumina, San Diego, Calif). A genotyping call rate of 99.85% was attained. Blinded internal replicates represented 4.4% of the sample set. The duplicate concordance rate was 100%.
For TGFβ1, the representative markers reference SNP identifier 4803455 (rs4803455) and rs1800469 were chosen based on a prevalent minor allele frequency and previous findings described in the literature.23 Both were genotyped using a TaqMan assay from Applied Bio-systems (Foster City, Calif). Each 5-μL polymerase chain reaction (PCR) mixture contained 20 ng of genomic DNA, primers, probes, and TaqMan Universal PCR Master Mix (containing AmpErase uracil N glycosylase [UNG], AmpliTaq Gold enzyme, dinucleotide triphosphates, and reaction buffer). PCR was carried out under the following conditions: 50°C for 2 minutes to activate UNG, then 95°C for 10 minutes, followed by 40 cycles at 92°C for 15 seconds and 60°C for 1 minute using 384-well duel-block ABI 9700. Fluorescent endpoints of the TaqMan reactions were measured using a 7900HT sequence-detection instrument.
Statistical Methods
All statistical analyses were performed using the SAS statistical software package (version 9.2; SAS Institute, Inc., Cary, NC). Months of survival were calculated based on the month and year of diagnosis and the month and year of death or the date of last contact. Associations between SNPs and the risk of dying from CRC within 5 years after diagnosis were evaluated using Cox proportional hazards models to obtain multivariate HRRs and 95% confidence intervals (CIs). We adjusted the analyses for age at diagnosis, study center, race, sex, tumor molecular phenotype, and AJCC stage to estimate HRRs. Stepwise regression was used to determine which SNPs and their corresponding inheritance models contributed most significantly to survival in the candidate pathway for each study separately. The numbers of SNPs entered into the initial analyses: TGFβ1 (2 SNPs), TGFβR1 (3 SNPs), Smad1 (5 SNPs), Smad2 (4 SNPs), Smad3 (37 SNPs), Smad4 (2 SNPs), Smad7 (11 SNPs), BMP1 (11 SNPs), BMP2 (5 SNPs), BMP4 (3 SNPs), BMPR1A (9 SNPs), BMPR1B (21 SNPs), BMPR2 (11 SNPs), GDF10 (7 SNPs), RUNX1 (40 SNPs), RUNX2 (19 SNPs), RUNX3 (9 SNPs), eiF4E (3 SNPs), eiF4EBP2 (2 SNPs), eiF4EBP3 (2 SNPs), and MAPK1 (6 SNPs).
To summarize the risk associated with multiple variants across the pathway, we created a summary score that was based on all at-risk genotypes identified from stepwise regression models for colon and rectal cancer. The score for each SNP was based on the inheritance model and its associated risk. For the codominant or additive model, a score of zero, 1, or 2 was assigned directly related to the number of high-risk alleles; whereas a scores of zero or 2 were assigned for the dominant or recessive models, respectively. After assigning a score for each SNP identified from the stepwise regression models, the scores were summed across SNPs to generate an individual summary score. The continuous score variable was redefined as a categorical variable based on the frequency distribution within the study population. Our objective was to have approximately 10% of the sample in the referent category to provide stability to risk estimates.
RESULTS
Characteristics of the study population are listed in Table 1. Individuals who were younger lived longer after diagnosis than older individuals, as expected. Those with less advanced disease stage also had better survival. The majority of deaths were attributed to CRC, and most CRC deaths occurred within the first 5 years of follow-up, although a few deaths occurred in the first year after diagnosis. Patients who had KRAS2-mutated tumors had slightly lower survival rates, whereas patients who had MSI-positive tumors had greater survival.
Table 1.
Characteristics of Study Population
| Colon Cancer | Rectal Cancer | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| No. of Patients (%) | Survival, mo | No. Surviving (%) | No. of Patients (%) | Survival, mo | No. Surviving (%) | |||||||||||
| Characteristic | Dead | Alivea | Log-Rank P | Mean | Median | Range | At 1 Year | At 5 Years | Dead | Aliveb | Log-Rank P | Mean | Median | Range | At 1 Year | At 5 Years |
| Sex | ||||||||||||||||
| Men | 320 (37) | 549 (63) | <.01 | 61.9 | 61 | 3-120 | 848 (98) | 610 (70) | 171 (38) | 280 (62) | <.01 | 66.4 | 72 | 6-114 | 438 (97) | 317 (70) |
| Women | 200 (29) | 487 (71) | 64.3 | 64 | 5-120 | 673 (98) | 516 (75) | 88 (29) | 217 (71) | 71.1 | 76 | 7-115 | 302 (99) | 245 (80) | ||
| Age at time of diagnosis, y | ||||||||||||||||
| <60 | 117 (28) | 297 (72) | <.01 | 62.4 | 61 | 5-119 | 404 (98) | 312 (75) | 92 (29) | 220 (71) | <.01 | 70.4 | 75 | 10-115 | 309 (99) | 243 (78) |
| 60-69 | 145 (30) | 333 (70) | 64.9 | 64 | 5-119 | 470 (98) | 361 (76) | 74 (33) | 151 (67) | 67 | 73 | 7-113 | 222 (99) | 168 (75) | ||
| 70-79 | 218 (40) | 333 (60) | 61.3 | 61 | 3-120 | 536 (97) | 369 (67) | 80 (45) | 98 (55) | 66.2 | 69.5 | 6-113 | 171 (96) | 120 (67) | ||
| Cause of death | ||||||||||||||||
| Other causes | 211 (41) | 51.2 | 48 | 3-119 | 205 (97) | 79 (37) | 88 (34) | 57.5 | 59.5 | 8-115 | 86 (98) | 44 (50) | ||||
| CRC | 309 (59) | 29.1 | 26 | 5-88 | 280 (91) | 11 (4) | 171 (66) | 35.4 | 32 | 6-83 | 157 (92) | 21 (12) | ||||
| All causes | 485 (93) | 90 (17) | 243 (94) | 65 (25) | ||||||||||||
| Tumor phenotype | ||||||||||||||||
| CIMP-positive | 101 (37) | 172 (63) | <.01 | 63.2 | 62 | 7-118 | 266 (97) | 184 (67) | 28 (47) | 31 (53) | .14 | 65.3 | 71 | 10-109 | 58 (98) | 40 (68) |
| KRAS-mutated | 145 (42) | 203 (58) | 60.6 | 60 | 3-120 | 335 (96) | 226 (65) | 71 (41) | 102 (59) | .14 | 65.9 | 72 | 6-115 | 169 (98) | 116 (67) | |
| TP53-mutated | 192 (37) | 325 (63) | 59.2 | 57 | 7-120 | 501 (97) | 349 (68) | 96 (35) | 182 (65) | 69.2 | 74.5 | 8-113 | 271 (97) | 206 (74) | ||
| MSI-positivec | 38 (21) | 147 (79) | 67.5 | 65 | 5-118 | 192 (98) | 152 (82) | |||||||||
| AJCC stage | ||||||||||||||||
| I | 78 (17) | 391 (83) | <.01 | 71.7 | 71 | 9-119 | 466 (99) | 416 (89) | 79 (21) | 303 (79) | <.01 | 76.2 | 78 | 6-114 | 378 (99) | 335 (88) |
| II | 115 (28) | 291 (72) | 65.1 | 62.5 | 8-120 | 403 (99) | 317 (78) | 50 (40) | 74 (60) | 69.3 | 74.5 | 7-113 | 123 (99) | 89 (72) | ||
| III | 154 (41) | 220 (59) | 58.3 | 54.5 | 3-119 | 368 (98) | 240 (64) | 71 (40) | 105 (60) | 63.1 | 68 | 6-115 | 173 (98) | 119 (68) | ||
| IV | 115 (90) | 13 (10) | 29.6 | 20.5 | 5-113 | 106 (83) | 17 (13) | 54 (95) | 3 (5) | 28.9 | 23 | 7-87 | 49 (86) | 6 (11) | ||
| Unknownd | 58 (32) | 121 (68) | 68.9 | 68 | 9-118 | 178 (99) | 136 (76) | 5 (29) | 12 (71) | 67.3 | 67 | 23-112 | 17 (100) | 13 (76) | ||
CRC indicates colorectal carcinoma; CIMP, CpG island methylator phenotype-positive, KRAS, v-Kras2 Kirsten rat sarcoma viral oncogene homolog; TP53, tumor protein 53; MSI, microsatellite instability; AJCC, American Joint Committee on Cancer.
Colon cancer follow-up was through February 2000.
Rectal cancer follow-up was through March 2007.
MSI was not considered in patients with rectal cancer because of its rarity.
P values were based on AJCC stage I through IV.
Twelve genes in the TGF-β signaling pathway were associated with colon cancer survival (Table 2). TBFβR1, Smad7, BMP2, BMP4, GDF10, EIF4E, EIF4EBP3, and RUNX3 each had at least 1 tagSNP that was associated significantly with survival after accounting for stage at diagnosis, other SNPs, tumor molecular phenotype, and the standard adjustments of age, sex, race, and center. Smad3, BMPR1A, BMPR1B, and RUNX1 had multiple tagSNPs associated with colon cancer survival. Fifteen tagSNPs were associated with survival among those diagnosed with rectal cancer (Table 3). BMPR1A, BMPR1B, and RUNX3 each had 1 tagSNP associated with survival; whereas Smad3, Smad7, BMP1, and RUNX1 had multiple tagSNPs associated with survival after a diagnosis of rectal cancer.
Table 2.
Hazard Rate Ratios Associated With Transforming Growth Factor-β Signaling Pathway Genes and Colon Cancer
| Gene | Reference SNP (rs) No. | No. of Deaths/Person Years | HRRa | 95% CI |
|---|---|---|---|---|
| TGFBR1 | rs10733710 | |||
| GG | 202/3869 | 1.00 | ||
| GA/AA | 93/2475 | 0.64 | 0.48-0.87 | |
| SMAD3 | rs2414937 | |||
| GG | 176/4167 | 1.00 | ||
| GC/CC | 122/2215 | 1.51 | 1.14-2.01 | |
| rs11639295 | ||||
| CC/CT | 280/5705 | 1.00 | ||
| TT | 16/656 | 0.42 | 0.21-0.85 | |
| SMAD7 | rs3736242 | |||
| GG/GA | 283/6085 | 1.00 | ||
| AA | 15/297 | 1.88 | 1.06-3.32 | |
| BMP2 | rs3178250 | |||
| TT | 204/3759 | 1.00 | ||
| TC/CC | 94/2623 | 0.65 | 0.48-0.87 | |
| BMP4 | rs17563 | |||
| CC/CT | 227/4887 | 1.00 | ||
| TT | 71/1495 | 1.45 | 1.05-1.99 | |
| BMPR1A | rs2883420 | |||
| TT | 98/2462 | 1.00 | ||
| TC/CC | 200/3919 | 1.47 | 1.09-1.99 | |
| rs12765929 | ||||
| GG | 157/3272 | 1.00 | ||
| GT/TT | 140/3102 | 1.59 | 1.18-2.14 | |
| BMPR1B | rs12508087 | |||
| TT | 185/3746 | 1.00 | ||
| TA/AA | 113/2636 | 0.73 | 0.54-0.97 | |
| rs2120834 | ||||
| GG/GC | 254/5561 | 1.00 | ||
| CC | 44/821 | 1.52 | 1.03-2.26 | |
| rs2214395 | ||||
| AA/AG | 281/6226 | 1.00 | ||
| GG | 17/152 | 2.33 | 1.27-4.27 | |
| rs7698964 | ||||
| GG/GA | 292/6225 | 1.00 | ||
| AA | 4/124 | 0.20 | 0.06-0.68 | |
| GDF10 | rs12769499 | |||
| AA/AG | 290/6217 | 1.00 | ||
| GG | 7/145 | 2.41 | 1.03-5.62 | |
| EIF4E | rs12498533 | |||
| AA | 76/1873 | 1.00 | ||
| AC/CC | 214/4419 | 1.48 | 1.08-2.03 | |
| EIF4EBP3 | rs250425 | |||
| CC | 175/3872 | 1.00 | ||
| CT/TT | 117/2438 | 1.39 | 1.05-1.84 | |
| RUNX1 | rs2834645 | |||
| TT | 177/4136 | 1.00 | ||
| TC/CC | 121/2241 | 1.49 | 1.13-1.97 | |
| rs81341792 | ||||
| TT | 213/4414 | 1.00 | ||
| TC | 74/1808 | 0.56 | 0.40-0.79 | |
| CC | 11/160 | 1.84 | 0.93-3.65 | |
| RUNX3 | rs7517302 | |||
| TT | 79/2080 | 1.00 | ||
| TC/CC | 216/4248 | 1.50 | 1.09-2.06 |
SNP indicates single nucleotide polymorphism; HRR, hazard rate ratio; CI, confidence interval; TGFBR1, transforming growth factor-β receptor 1; G, guanine; A, adenine; SMAD3/SMAD7, smooth muscle actin/mothers against decapentaplegic homologs 3 and 7, respectively; C, cytosine; T, thymine; BMP3/BMP4, bone morphogenetic proteins 3 and 4, respectively; BMPR1A/BMPR1B, bone morphogenetic protein receptors 1A and 1B, respectively; GDF10, growth differentiation factor 10 (also known as bone morphogenetic protein 3B); EIF4E, eukaryotic translation initiation factor 4E; EIF4EBP3, eukaryotic translation initiation factor 4E-binding protein 3; RUNX1/RUNX3, Runt-related transcription factors 1 and 3, respectively.
HRRs and 95% CIs were adjusted for age, center, race, sex, tumor markers, American Joint Committee on Cancer stage, and other SNPs.
The dominant model was used to calculate the summary score.
Table 3.
Hazard Rate Ratios Associated With Transforming Growth Factor-β Signaling Pathway Genes and Rectal Cancer
| Gene | Reference SNP (rs) No. | No. of Deaths/Person Years | HRR | 95% CI |
|---|---|---|---|---|
| SMAD3 | rs12708492 | |||
| CC | 43/872 | 1.00 | ||
| CT/TT | 107/2379 | 0.38 | 0.24-0.59 | |
| rs16950687 | ||||
| AA/AG | 131/2974 | 1.00 | ||
| GG | 19/279 | 1.92 | 1.07-3.43 | |
| rs3825977 | ||||
| CC | 90/2164 | 1.00 | ||
| CT/TT | 60/1089 | 1.64 | 1.09-2.47 | |
| rs7181556 | ||||
| CC/CT | 137/3075 | 1.00 | ||
| TT | 13/178 | 2.30 | 1.17-4.52 | |
| SMAD7 | rs49398322 | |||
| AA | 74/1810 | 1.00 | ||
| AG | 68/1196 | 2.14 | 1.43-3.21 | |
| GG | 8/247 | 0.90 | 0.39-2.08 | |
| rs12953717 | ||||
| CC/CT | 131/2713 | 1.00 | ||
| TT | 19/540 | 0.38 | 0.21-0.71 | |
| BMP1 | rs12114940 | |||
| TT/TG | 118/2738 | 1.00 | ||
| GG | 32/515 | 2.36 | 1.47-3.77 | |
| rs4076873 | ||||
| AA | 89/2025 | 1.00 | ||
| AC/CC | 61/1228 | 1.93 | 1.32-2.83 | |
| BMPR1A | rs2168730 | |||
| AA | 89/1876 | 1.00 | ||
| AG/GG | 61/1377 | 0.65 | 0.44-0.97 | |
| BMPR1B | rs13134042 | |||
| GG | 78/1971 | 1.00 | ||
| GA/AA | 72/1282 | 2.07 | 1.42-3.02 | |
| RUNX1 | rs1475840 | |||
| AA | 43/766 | 1.00 | ||
| AG/GG | 107/2482 | 0.48 | 0.32-0.73 | |
| rs1883066 | ||||
| GG | 123/2432 | 1.00 | ||
| GC/CC | 27/821 | 0.51 | 0.31-0.84 | |
| rs2253319 | ||||
| CC | 61/1555 | 1.00 | ||
| CT/TT | 89/1698 | 1.68 | 1.13-2.51 | |
| rs7280028 | ||||
| TT/TC | 143/3175 | 1.00 | ||
| CC | 7/78 | 3.72 | 1.54-8.96 | |
| RUNX3 | rs2135756 | |||
| AA/AG | 127/2427 | 1.00 | ||
| GG | 23/826 | 0.32 | 0.19-0.55 |
SNP indicates single nucleotide polymorphism; HRR, hazard rate ratio; CI, confidence interval; SMAD3/SMAD7, smooth muscle actin/mothers against decapentaplegic homologs 3 and 7, respectively; C, cytosine; T, thymine; BMP1, bone morphogenetic protein 1; G, guanine; A, adenine; BMPR1A/BMPR1B, bone morphogenetic protein receptors 1A and 1B, respectively; RUNX1/RUNX3, Runt-related transcription factors 1 and 3, respectively.
HRRs and 95% CIs were adjusted for age, center, race, sex, tumor markers, American Joint Committee on Cancer stage, and other SNPs.
The dominant model was used to calculate the summary score.
To better estimate the hazard of dying across the TGF-β signaling pathway genes that were analyzed, we assessed the composite of the “at-risk” genotypes based on those genes that were important for colon and rectal cancer. For both colon and rectal cancer, we observed that, with increasing number of “at-risk” genotypes within the pathway, the likelihood of dying increased in a linear manner (Table 4). For colon cancer, the HRRs for patients in the highest category of risk was 5.10 (95% CI, 2.56-10.15); whereas, for patients with rectal cancer, it was 6.03 (95% CI, 2.86-12.75).
Table 4.
Hazard Rate Ratios Associated With Summary Scores Across Genes in the Transforming Growth Factor-β Signaling Pathway After Diagnosis With Colon or Rectal Cancer
| Colon Cancer | Rectal Cancer | ||||||
|---|---|---|---|---|---|---|---|
| No. of At-Risk Genotypesa | No. of Deaths/Person Years | HRRb | 95% CI | No. of At-Risk Genotypesa | No. of Deaths/Person Years | HRRb | 95% CI |
| 6-12 vs 6-12 | 20/944 | 1.00 | 2-8 vs 2-8 | 18/865 | 1.00 | ||
| 14-18 vs 6-12 | 174/3795 | 2.15 | 1.15-4.02 | 10-12 vs 2-8 | 7/1609 | 2.14 | 1.17-3.91 |
| 20-20 vs 6-12 | 51/944 | 4.43 | 2.23-8.80 | 14-14 vs 2-8 | 32/486 | 4.11 | 2.02-8.34 |
| 22-28 vs 6-12 | 53/699 | 5.10 | 2.56-10.15 | 16-22 vs 2-8 | 23/293 | 6.03 | 2.86-12.75 |
HRR indicates hazard rate ratio; CI, confidence interval.
Summary score genotypes were based on the single nucleotide polymorphisms listed in Table 2 for colon cancer and in Table 3 for rectal cancer.
HRRs were based on colorectal 5-year survival adjusted for age, center, race, sex, American Joint Committee on Cancer stage, and tumor molecular phenotype.
We assessed the stage-specific risk associated with our TGF-β summary score (Table 5). Few individuals were diagnosed with AJCC stage I cancer, which made risk estimates imprecise. HRRs increased in a linear manner with increasing numbers of high-risk genotypes for all disease stages except stage I, in which no increased risk was observed for colon cancer and an increased risk was observed for patients who had rectal cancer with >14 at-risk genotypes. For colon cancer, the HRRs were slightly greater for stage II cancers than for either stage III or IV cancers. For rectal cancer, the risk was not significant for patients who had stage III disease, whereas patients who had stage IV disease had an increased risk when they had the most at-risk genotypes.
Table 5.
Hazard Rate Ratios for American Joint Committee on Cancer Stage Associated With Summary Scores Across Genes in the Transforming Growth Factor-β Signaling Pathway After Diagnosis With Colon Cancer or Rectal Cancer
| Colon Cancer | Rectal Cancer | |||||||
|---|---|---|---|---|---|---|---|---|
| AJCC Stage | No. of At-Risk Genotypesa | No. of Deaths/Person Years | HRRb | 95% CI | No. of At-Risk Genotypesa | No. of Deaths/Person Years | HRRb | 95% CI |
| I | 6-16 | 5/4470 | 1.00 | 2-10 | 4/2244 | 1.00 | ||
| 18-28 | 7/3678 | 1.27 | 0.21-7.58 | 12-12 | 4/1044 | 5.57 | 0.89-34.95 | |
| 14-14 | 8/628 | 14.42 | 2.81-73.92 | |||||
| 16-22 | 7/373 | 9.98 | 1.81-55.22 | |||||
| II | 6-14 | 10/2692 | 1.00 | 2-10 | 8/2244 | 1.00 | ||
| 16-20 | 23/4579 | 1.58 | 0.62-4.04 | 12-12 | 11/1044 | 3.35 | 1.14-9.80 | |
| 22-28 | 15/877 | 6.43 | 2.36-17.53 | 14-22 | 8/1001 | 4.62 | 1.45-14.69 | |
| III | 6-14 | 18/2692 | 1.00 | 2-8 | 8/1149 | 1.00 | ||
| 16-18 | 47/3419 | 1.54 | 0.85-2.80 | 10-12 | 25/2139 | 1.24 | 0.52-2.92 | |
| 20-20 | 20/1159 | 2.79 | 1.41-5.49 | 14-14 | 9/628 | 1.98 | 0.56-6.99 | |
| 22-28 | 18/877 | 2.90 | 1.44-5.84 | 16-22 | 5/373 | 3.04 | 0.88-10.57 | |
| IV | 6-14 | 21/2692 | 1.00 | 2-8 | 8/1149 | 1.00 | ||
| 16-16 | 27/1778 | 2.01 | 0.95-4.28 | 10-12 | 26/2139 | 1.35 | 0.37-4.97 | |
| 18-18 | 33/1641 | 3.92 | 1.75-8.80 | 14-22 | 16/1001 | 11.05 | 2.62-46.67 | |
| 20-28 | 24/2037 | 3.11 | 1.28-7.60 | |||||
AJCC indicates American Joint Committee on Cancer; HRR, hazard rate ratio; CI, confidence interval.
DISCUSSION
The current data suggest that genetic variation in the TGF-β signaling pathway may play an important role in survival after a diagnosis of colon or rectal cancer. Multiple genes and SNPs that were evaluated within this pathway were associated with survival in both colon cancer and rectal cancer. Furthermore, our data suggest that having multiple high-risk genotypes in the pathway increased risk beyond that observed for each independent genotype.
The importance of the TGF-β signaling pathway in the etiology of colon cancer is well documented.24-29 The TGF-β signaling pathway is critical for tissue homeostasis, and TGF-β is central in the inhibition of cell proliferation, apoptosis, and inflammation. Mutations in TGFβR2 occur in approximately 30% of colon cancers.29 Smad4 and BMPR1A mutations are reported commonly in patients with familial juvenile polyposis, a condition that increases the risk of CRC.29 We have reported on associations with genetic variability in this pathway and the risk of colon and rectal cancers. We have demonstrated that having multiple high-risk genotypes increased the risk beyond that of individual genotypes when all 3 separate pathway components were included: TGFβ1 and Smad13,14 and the RUNX, MAPK1, eiF4E,15 and BMP genes.30 The TGF-β pathway also has been identified as 1 of the most important pathways influencing colon cancer risk based on gene-set analysis data from genome-wide association studies.31
Although the biologic rationale exist for involvement of the TGF-β signaling pathway in survival, limited information is available on genetic variation in the pathway and survival. Forsti and colleagues investigated at 9 polymorphisms in the TGF-β signaling pathway and CRC among 308 patients with CRC.32 Those authors observed that the TGF-βRA IVS7G+24A minor allele was associated with better survival and less aggressive tumors than the more common allele. We did not observe any clear differences in the HRR based on AJCC disease stage. Others have reported on differences in survival based on tumor protein expression of these genes. Improved disease-free survival after a diagnosis of CRC was observed when TGF-β expression was increased.33 Studies of Smad2 and Smad4 in patients with CRC have demonstrated that reduced protein expression of Smad2 and Smad4 results in a poor prognosis.34,35 In fact, the deletion of Smad4 has been considered a negative predictor of chemotherapy benefit for patients with CRC.36 Smad7 also was associated with survival in a study of 264 CRC tumors. Smad7 deletions improved survival, whereas duplication of Smad7, as determined by gene copy, decreased survival. This affect was attributed to the blocking TGF-β signaling by Smad7, because TGF-β signaling is necessary for cell growth arrest and apoptosis.
A key question that we address in this study is whether having multiple variants in a pathway has a greater influence on risk than an individual variant. It is possible that having a single at-risk genotype impairs or inactivates the pathway and that additional harm is not encountered with multiple at-risk genotypes. However, it also is possible that risk increases with increasing number of deleterious variations, resulting in an overall unstable or weak pathway. Our results suggest that the integrity of the pathway is diminished with additional risk genotypes in the pathway. This weakened pathway may result in an increased risk of developing CRC and, in accord with the data presented here, may influence survival after diagnosis.
This study also has limitations. First, we assessed many genes and SNPs within the TGF-β signaling pathway, although other genes and SNPs that may contribute to risk were not included. We used a variety of statistical methods to consolidate the information from over 200 tagSNPs. We used stepwise regression to identify the tagSNPs that were most important within the pathway; however, we recognize that entry or exit into the model depended on other variables in the model. We developed a summary score with the hope of being able to summarize risk across the candidate pathway. Our method of scoring consisted of a simple summation of the number of at-risk genotypes and was not weighted by the magnitude or precision of the estimate. Estimation of risk was based on categorization of the summary score. It should be recognized that HRRs vary depending on the referent group and cutoff points used. We tried to establish categories based on the study data with the objective of having stable categories. However, although the exact point estimate varied with cutoff points, the same strong linear association for increasing HRRs with increasing numbers of at-risk genotypes was observed for all models. The data-set did not contain complete treatment data; however, treatment was determined uniformly according to disease stage; therefore, stage represented a surrogate for adjustment of treatment.
The current study has numerous strengths, including extensive data available to adjust for multiple tumor phenotypes that may influence survival. Our dataset includes extensive follow-up beyond the first 5 years after diagnosis, when most CRC deaths occur. We also have complete AJCC stage data for most study participants. Stage is probably the single most important contributor to survival and was the key determinant of treatment when our study was conducted. We have used a candidate pathway approach relying on tagSNPs for complete gene coverage. Thus, our approach is to assess variation in the gene and the pathway that influences risk. We believe that our candidate pathway approach provides the most complete gene coverage of the genes available.
In summary, we observed that genetic variation in the TGF-β signaling pathway increased the likelihood of dying after a diagnosis of colon cancer or rectal cancer. The risk associated with the pathway increased with the number of high-risk genotypes. These findings merit replication in other studies.
Acknowledgments
We acknowledge the contributions of Sandra Edwards, Roger Edwards, Leslie Palmer, Donna Schaffer, Drs. Kristin Anderson and John Potter, and Judy Morse for data management and collection.
The contents of this article are solely the responsibility of the authors and do not necessarily represent the official view of the National Cancer Institute.
CONFLICT OF INTEREST DISCLOSURES
This study was funded by National Cancer Institute grants CA48998 and CA61757. This research also was supported by the Utah Cancer Registry, which is funded by contract N01-PC-67000 from the National Cancer Institute with additional support from the State of Utah Department of Health, the Northern California Cancer Registry, and the Sacramento Tumor Registry.
References
- 1.Gordon KJ, Blobe GC. Role of transforming growth factor-beta superfamily signaling pathways in human disease. Biochim Biophys Acta. 2008;1782:197–228. doi: 10.1016/j.bbadis.2008.01.006. [DOI] [PubMed] [Google Scholar]
- 2.Piccirillo SG, Vescovi AL. Bone morphogenetic proteins regulate tumorigenicity in human glioblastoma stem cells. Ernst Schering Found Symp Proc. 2006;5:59–81. doi: 10.1007/2789_2007_044. [DOI] [PubMed] [Google Scholar]
- 3.Piccirillo SG, Reynolds BA, Zanetti N, et al. Bone morphogenetic proteins inhibit the tumorigenic potential of human brain tumour-initiating cells. Nature. 2006;444:761–765. doi: 10.1038/nature05349. [DOI] [PubMed] [Google Scholar]
- 4.Blyth K, Cameron ER, Neil JC. The RUNX genes: gain or loss of function in cancer. Nat Rev Cancer. 2005;5:376–387. doi: 10.1038/nrc1607. [DOI] [PubMed] [Google Scholar]
- 5.Lee KS, Hong SH, Bae SC. Both the Smad and p38 MAPK pathways play a crucial role in Runx2 expression following induction by transforming growth factor-beta and bone morphogenetic protein. Oncogene. 2002;21:7156–7163. doi: 10.1038/sj.onc.1205937. [DOI] [PubMed] [Google Scholar]
- 6.Cameron ER, Blyth K, Hanlon L, et al. The Runx genes as dominant oncogenes. Blood Cells Mol Dis. 2003;30:194–200. doi: 10.1016/s1079-9796(03)00031-7. [DOI] [PubMed] [Google Scholar]
- 7.Ito Y, Osato M, Ito K. RUNX and cancer. Ann Acad Med Singapore. 2003;32(5 suppl):S6–S7. [PubMed] [Google Scholar]
- 8.Ge C, Xiao G, Jiang D, et al. Identification and functional characterization of ERK/MAPK phosphorylation sites in the Runx2 transcription factor. J Biol Chem. 2009;284:32533–32543. doi: 10.1074/jbc.M109.040980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Guo H, Zhang HY, Wang SL, Ye L, Yang GH, Bu H. Smad4 and ERK2 stimulated by transforming growth factor beta1 in rhabdomyosarcoma. Chin Med J (Engl) 2007;120:515–521. [PubMed] [Google Scholar]
- 10.Javelaud D, Mauviel A. Smad signal transduction. Interplays between the Smad and Map kinase signaling pathways. Proteins Cell Regul. 2006;5:317–334. [Google Scholar]
- 11.Berkel HJ, Turbat-Herrera EA, Shi R, de Benedetti A. Expression of the translation initiation factor eIF4E in the polyp-cancer sequence in the colon. Cancer Epidemiol Biomarkers Prev. 2001;10:663–666. [PubMed] [Google Scholar]
- 12.Rajasekhar VK, Holland EC. Postgenomic global analysis of translational control induced by oncogenic signaling. Oncogene. 2004;23:3248–3264. doi: 10.1038/sj.onc.1207546. [DOI] [PubMed] [Google Scholar]
- 13.Slattery ML, Herrick JS, Lundgreen A, Wolff RK. Genetic variation in the TGF-β signaling pathway and colon and rectal cancer risk. Cancer Epidemiol Biomarkers Prevent. 2011;20:57–69. doi: 10.1158/1055-9965.EPI-10-0843. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Slattery ML, Herrick J, Curtin K, et al. Increased risk of colon cancer associated with a genetic polymorphism of SMAD7. Cancer Res. 2010;70:1479–1485. doi: 10.1158/0008-5472.CAN-08-1792. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Slattery ML, Lundgreen A, Herrick SJ, Caan BJ, Potter JD, Wolff RK. Associations between genetic variation in RUNX1, RUNX2, RUNX3, MAPK1, and eIF4E and risk of colon and rectal cancer: additional support for a TGF-β signaling pathway. Carcinogenesis. 2010 doi: 10.1093/carcin/bgq245. published online ahead of print November 28, 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Slattery ML, Potter JD, Duncan DM, Berry TD. Dietary fats and colon cancer: assessment of risk associated with specific fatty acids. Int J Cancer. 1997;73:670–677. doi: 10.1002/(sici)1097-0215(19971127)73:5<670::aid-ijc10>3.0.co;2-a. [DOI] [PubMed] [Google Scholar]
- 17.Slattery ML, Caan BJ, Benson J, Murtaugh M. Energy balance and rectal cancer: an evaluation of energy intake, energy expenditure, and body mass index. Nutr Cancer. 2003;46:166–171. doi: 10.1207/S15327914NC4602_09. [DOI] [PubMed] [Google Scholar]
- 18.Samowitz WS, Curtin K, Ma KN, et al. Prognostic significance of p53 mutations in colon cancer at the population level. Int J Cancer. 2002;99:597–602. doi: 10.1002/ijc.10405. [DOI] [PubMed] [Google Scholar]
- 19.Slattery ML, Curtin K, Anderson K, et al. Associations between cigarette smoking, lifestyle factors, and microsatellite instability in colon tumors. J Natl Cancer Inst. 2000;92:1831–1836. doi: 10.1093/jnci/92.22.1831. [DOI] [PubMed] [Google Scholar]
- 20.Samowitz WS, Curtin K, Schaffer D, Robertson M, Leppert M, Slattery ML. Relationship of Ki-ras mutations in colon cancers to tumor location, stage, and survival: a population-based study. Cancer Epidemiol Biomarkers Prev. 2000;9:1193–1197. [PubMed] [Google Scholar]
- 21.Slattery ML, Curtin K, Sweeney C, et al. Diet and lifestyle factor associations with CpG island methylator phenotype and BRAF mutations in colon cancer. Int J Cancer. 2007;120:656–663. doi: 10.1002/ijc.22342. [DOI] [PubMed] [Google Scholar]
- 22.Samowitz WS, Curtin K, Ma KN, et al. Microsatellite instability in sporadic colon cancer is associated with an improved prognosis at the population level. Cancer Epidemiol Biomarkers Prev. 2001;10:917–923. [PubMed] [Google Scholar]
- 23.Zha Y, Leung KH, Lo KK, et al. TGFB1 as a susceptibility gene for high myopia: a replication study with new findings. Arch Ophthalmol. 2009;127:541–548. doi: 10.1001/archophthalmol.2008.623. [DOI] [PubMed] [Google Scholar]
- 24.Xu Y, Pasche B. TGF-beta signaling alterations and susceptibility to colorectal cancer. Hum Mol Genet. 2007;16(spec no. 1):R14–R20. doi: 10.1093/hmg/ddl486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Saltzman BS, Yamamoto JF, Decker R, et al. Association of genetic variation in the transforming growth factor beta-1 gene with serum levels and risk of colorectal neoplasia. Cancer Res. 2008;68:1236–1244. doi: 10.1158/0008-5472.CAN-07-2144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Subramaniam MM, Chan JY, Yeoh KG, Quek T, Ito K, Salto-Tellez M. Molecular pathology of RUNX3 in human carcinogenesis. Biochim Biophys Acta. 2009;1796:315–331. doi: 10.1016/j.bbcan.2009.07.004. [DOI] [PubMed] [Google Scholar]
- 27.Gulubova M, Manolova I, Ananiev J, Julianov A, Yovchev Y, Peeva K. Role of TGF-beta1, its receptor TGFbetaRII, and Smad proteins in the progression of colorectal cancer. Int J Colorectal Dis. 2010;25:591–599. doi: 10.1007/s00384-010-0906-9. [DOI] [PubMed] [Google Scholar]
- 28.Hong S, Lee HJ, Kim SJ, Hahm KB. Connection between inflammation and carcinogenesis in gastrointestinal tract: focus on TGF-beta signaling. World J Gastroenterol. 2010;16:2080–2093. doi: 10.3748/wjg.v16.i17.2080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Bellam N, Pasche B. TGF-beta signaling alterations and colon cancer. Cancer Treat Res. 2010;155:85–103. doi: 10.1007/978-1-4419-6033-7_5. [DOI] [PubMed] [Google Scholar]
- 30.Slattery ML, Lundgreen A, Herrick JS, Kadlubar S, Caan BJ, Potter JD, Wolff RK. Genetic variation in bone morphogenetic protein (BMP) and colon and rectal cancer. Int J Cancer. doi: 10.1002/ijc.26047. In press. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Chen LS, Hutter CM, Potter JD, et al. Insights into colon cancer etiology via a regularized approach to gene set analysis of GWAS data. Am J Hum Genet. 2010;86:860–871. doi: 10.1016/j.ajhg.2010.04.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Forsti A, Li X, Wagner K, et al. Polymorphisms in the transforming growth factor beta 1 pathway in relation to colorectal cancer progression. Genes Chromosomes Cancer. 2010;49:270–281. doi: 10.1002/gcc.20738. [DOI] [PubMed] [Google Scholar]
- 33.Tsamandas AC, Kardamakis D, Ravazoula P, et al. The potential role of TGFbeta1, TGFbeta2 and TGFbeta3 protein expression in colorectal carcinomas. Correlation with classic histopathologic factors and patient survival. Strahlenther Onkol. 2004;180:201–208. doi: 10.1007/s00066-004-1149-x. [DOI] [PubMed] [Google Scholar]
- 34.Isaksson-Mettavainio M, Palmqvist R, Forssell J, Stenling R, Oberg A. SMAD4/DPC4 expression and prognosis in human colorectal cancer. Anticancer Res. 2006;26(1B):507–510. [PubMed] [Google Scholar]
- 35.Xie W, Rimm DL, Lin Y, Shih WJ, Reiss M. Loss of Smad signaling in human colorectal cancer is associated with advanced disease and poor prognosis. Cancer J. 2003;9:302–312. doi: 10.1097/00130404-200307000-00013. [DOI] [PubMed] [Google Scholar]
- 36.Boulay JL, Mild G, Lowy A, et al. SMAD7 is a prognostic marker in patients with colorectal cancer. Int J Cancer. 2003;104:446–449. doi: 10.1002/ijc.10908. [DOI] [PubMed] [Google Scholar]