Abstract
Introduction
Lymph node metastases and tumor characteristics predict breast cancer prognosis but correlate imperfectly with likelihood of metastatic relapse. Discovery of genetic polymorphisms affecting metastasis may improve identification of patients requiring aggressive adjuvant therapy to prevent recurrence. We investigated associations between several variants in the BRMS1 and SIPA1 metastasis-modifying genes and lymph node metastases, tumor subtype and grade, recurrence, disease-free survival and overall survival.
Methods
This cross-sectional and prospective prognostic analysis included 859 patients who received surgery for incident breast cancer at Roswell Park Cancer Institute, participated in the DataBank and BioRepository shared resource, and had DNA, clinical, and pathology data available for analysis. Genotyping for BRMS1 (rs11537993, rs3116068, and rs1052566) and SIPA1 (rs75894763, rs746429, rs3741378, and rs2306364) polymorphisms was performed using Sequenom® iPLEX Gold and Taqman® real-time PCR assays. Logistic and Cox proportional hazards regressions were used to estimate odds ratios (OR) and hazard ratios (HR), respectively.
Results
BRMS1 rs1052566 heterozygous individuals were more likely to have node positive tumors (OR=1.58, 95% CI 1.13-2.23), although there was no dose-response relationship, and those with at least one variant allele were less likely to have the luminal B subtype (AG+AA: OR=0.59, 95% CI 0.36-0.98). BRMS1 rs3116068 was associated with increased likelihood of having the luminal B and the HER2-enriched tumor subtype (Ptrend=0.03). Two SIPA1 SNPs, rs746429 and rs2306364, were associated with decreased risk of triple negative tumors (Ptrend=0.04 and 0.07, respectively). Presence of 8 or more risk alleles was associated with an increased likelihood of having a node positive tumor (OR=2.14, 95% CI 1.18-3.36, Ptrend = 0.002). There were no significant associations with survival.
Conclusions
Polymorphisms in metastasis-associated genes may be related to tumor characteristics and lymph node metastasis, but not survival. Future evaluation of metastasis modifying gene variants is necessary to better understand the biology of metastasis.
Keywords: Breast cancer, metastasis, single nucleotide polymorphism, recurrence, survival
Introduction
While early stage breast cancer has excellent prognosis, it is incurable once distant metastasis has occurred [1, 2]. Metastasis to regional lymph nodes is correlated with a higher risk of developing distant metastases [3, 4], as are tumor size and grade, estrogen and progesterone receptor status (ER and PR, respectively), and HER2 amplification. In general, larger tumors are correlated with increased likelihood of lymph node metastases at diagnosis and distant metastases [5, 6], while ER, PR, and HER2 status are markers of tumor aggressiveness and also determine suitability for targeted treatments [7].
Even with these known prognostic factors, however, the patients who will ultimately experience a recurrence are not clearly identified. Genetic variability may explain some of this heterogeneity in metastatic ability, particularly in genes affecting the metastatic cascade. Many metastasis-related genes have been identified, two of which are BRMS1 (breast cancer metastasis suppressor 1) and SIPA1 (signal-induced proliferation-associated 1).
BRMS1 can function as a metastasis suppressor gene [8–10] that affects apoptosis, colonization, cell adhesion and invasive potential by mitigating the effects of anti-apoptotic gene products regulated by the NF-kB pathway [11, 12]. No studies examining single nucleotide polymorphisms (SNPs) in BRMS1 have been published, although several expression studies have analyzed the relationship between BRMS1 and breast tumor characteristics and prognosis [13–17]. SIPA1 can affect metastatic efficiency by modifying cell adhesion [18] and expression of extracellular matrix genes [19] and has been shown to promote metastasis in vivo [20]. Several SIPA1 SNP association studies have been published, with conflicting reports with respect to prognosis [21–23]. These data indicate that BRMS1 and SIPA1 abnormalities could affect tumor aggressiveness, metastasis and the risk of recurrence in breast cancer patients.
Based on this previously published data, we selected several SNPs in BRMS1 and SIPA1 to investigate as potential candidate markers of tumor aggressiveness and recurrence. To investigate these relationships, we analyzed three SNPs in BRMS1 [rs11537993 (Leu67Leu); rs3116068 (3′ UTR); and rs1052566 (Ala273Val)] and four SNPs in SIPA1 [rs75894763 (Val621Val); rs746429 (Ala920Ala); rs3741378 (Ser182Phe); and rs2306364 (Ala342Ala)] with respect to lymph node metastasis, tumor grade and subtype, time to recurrence, disease-free survival and overall survival in women diagnosed with primary, incident breast cancer.
Methods
Study Population and Outcomes
We identified 859 women diagnosed between October 2003 and May 2010 with stage I-III incident, primary, histologically confirmed breast cancer, who received surgery and treatment at Roswell Park Cancer Institute (RPCI), provided informed consent to RPCI's DataBank and BioRepository (DBBR), and had a DNA sample available. The DBBR, as previously described [24], is a comprehensive data and sample bank containing high quality pre-treatment biospecimens and associated clinical and epidemiologic data. All patients diagnosed with cancer at RPCI are invited to participate. After consent and prior to treatment, including surgery, blood samples are collected, processed, and aliquoted for storage in liquid nitrogen. Epidemiologic data obtained via self-administered questionnaire were available for 688 of the participants in this analysis.
Outcomes were lymph node metastases, tumor subtype, tumor grade, time to recurrence, disease-free survival, and overall survival. Time to recurrence was defined as the time from diagnosis to date of first recurrence (local and regional recurrence and development of distant metastases) or last follow-up. Disease-free survival was defined as the time from diagnosis to the date of first recurrence, death from any cause, or last follow-up. Overall survival was defined as the time from diagnosis to the date of death from any cause or last follow-up. Clinical data were obtained from RPCI clinical databases and supplemented with data abstracted from medical records and the RPCI Tumor Registry. Vital status and recurrence data were obtained from the RPCI Tumor Registry and the National Comprehensive Cancer Network (NCCN) Breast Cancer outcomes database. The RPCI Tumor Registry conducts yearly follow-up on patients who were last seen at RPCI 13 months prior and known to be alive. Vital status and recurrences are ascertained via RPCI medical record abstraction, Social Security Death Index and Legacy.com searches, and/or letters sent to the patient, the patient's physician, or a family member. NCCN-coordinated linkage with the National Death Index for patients defined as “lost to follow-up” was completed December 8, 2011.
Fifteen participants missing HER2 status and 1 missing ER and PR status could not be classified by subtype. Additionally, 10 participants were missing lymph node status and 15 were missing tumor grade. Vital status was available for all participants. Follow-up ended July 2012 and follow-up time ranged from 4 to 101 months. This study was approved by the RPCI Institutional Review Board.
Genotyping
The NCBI dbSNP resource was used to identify SNPs in the BRMS1 and SIPA1 genes [25]. We initially selected 13 SNPs in BRMS1 [rs17850564 (Asp175Asp); rs11537993 (Leu67Leu); rs75053504 (A/G); rs3116068 (A/G); and rs1052566 (C/T)) and SIPA1 (rs3741378 (Ser182Phe); rs76570058 (Pro1038Thr); rs75861149 (Gly368Gly); rs2306364 (Ala342Ala); rs75894763 (Val621Val); rs746429 (Ala920Ala); rs77600626 (Gly249Glu); and rs76089059 (Ala997Ala)] for genotyping, based on presence in protein coding, 3′ untranslated, or promoter regions, and heterozygosity of ≥ 0.10.
Genotyping of all 13 SNPs was conducted by the RPCI Genomics Facility using Sequenom MassARRAY® iPLEX Gold matrix-assisted laser desorption-ionization time-of-flight mass spectrometry assays. Genotyping of several SNPs (rs1052566, rs746429, rs3471378, rs2306364, rs77600626, and rs76089059) was unsuccessful using this platform, and an additional 4 SNPs (rs17850564, rs75053504, rs75861149, and rs76570058) were monomorphic and not analyzed further. Probes for Taqman® (Applied Biosciences) real-time PCR genotyping assays were available for 4 of the SNPs that failed Sequenom® genotyping (rs1052566, rs746429, rs3471378, and rs2306364). Therefore, we were ultimately able to obtain genotyping data for analysis of 7 SNPs using either the Sequenom® (rs11537993, rs3116068, rs75894763) or Taqman® (rs1052566, rs746429, rs3471378, and rs2306364) platforms. Two SNPs in SIPA1, rs746429 and rs2306364, were in strong linkage disequilibrium (r2 = 0.809). Duplicate samples were genotyped to assess intra- and inter-plate reliability. Genotyping call rates ranged from 96.2-99.8%.
Cross-Sectional Analysis
Statistical analyses were performed using SAS version 9.3. Demographic variables and tumor characteristics were compared by lymph node status, tumor subtype, and tumor grade using Chi-squared and Fisher's exact tests as appropriate. Complete-case regression techniques were used to analyze the relationships between SNPs and lymph node metastasis, tumor grade, and tumor subtype. Potential covariates included age at diagnosis, tumor size, tumor grade, ER, PR, and HER2 status, lymph node metastasis, race, education, body mass index (BMI), age at menarche, menopausal status, age at menopause, parity, age at first birth, family history of breast cancer, history of benign breast disease, and hormone replacement therapy (HRT) use. Primary analyses incorporated data from the entire study population of 859 participants. Because we conducted complete-case analyses, participants missing epidemiologic questionnaire data dropped out of models including epidemiologic covariates. To minimize bias potentially introduced by these missing data, we initially included only tumor characteristics in adjusted models, as these data were available for the majority of our sample. Epidemiologic variables were included in separate models to assess the effect of their inclusion on odds ratio estimates. A variable was included in adjusted models if it was associated with the outcome and/or SNP(s), using Chi-squared or Fisher's exact tests of significance. Participants with and without questionnaire data had similar distributions of tumor characteristics and treatment modalities. Participants missing questionnaires were slightly younger and somewhat more likely to have node positive tumors, but differences were not significant. For all outcomes, sensitivity analyses in which we excluded participants who self-identified as non-white (5.4%) were performed.
Unconditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CI) for the associations between each of the 7 SNPs and lymph node status and tumor grade. We first constructed age-adjusted models and then subsequently added ER and PR status, tumor size, and tumor grade. A third model additionally included HER2 status, race, education, HRT, and menopausal status. Finally, we restricted our analysis to include only stage 2 and 3 participants, as these patients are eligible to have lymph node metastases (by definition, stage 1 is node-negative).
Moderate and low grade tumors were combined, creating a dichotomous grade variable with categories of low/moderate grade (well-differentiated and moderately differentiated tumors) and high grade (poorly differentiated and undifferentiated tumors). Using a similar strategy as outlined above, we first adjusted for age and tumor size. In separate models we added nodal status, ER, PR, and HER2 status, race, HRT, and menopausal status.
Generalized logit multinomial logistic regression was used to examine associations between each SNP and tumor subtype, using the luminal A subtype as the comparison group. Adjusted models included age and tumor size; age, tumor size and nodal status; and age, tumor size, nodal status, HRT, race, and menopausal status.
For all analyses, Ptrend was calculated by coding genotypes as 0, 1, or 2 for homozygous wildtype, heterozygous, or homozygous variant genotypes, respectively, and treating the SNP as a continuous variable in the regression models.
Survival Analysis
Cox proportional hazards regression was used to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the relationships between the SNPs and overall survival, time to recurrence, and disease-free survival. Log-rank tests were used to identify predictors for inclusion in multivariate proportional hazards regression models. Variables tested as predictors were age at diagnosis, tumor size, tumor grade, ER, PR, and HER2 status, race, education, BMI, age at menarche, menopausal status, parity, family history of breast cancer, history of benign breast disease, hormone replacement therapy use, radiation treatment, chemotherapy, hormonal treatment (tamoxifen, etc), and Charlson comorbidity score.
Significant variables were age, ER and PR status, tumor grade, tumor size, comorbidity score, radiation treatment, chemotherapy, hormone treatment, education, age at menarche, and menopausal status. Between 7 and 12 events, depending on the outcome, occurred among participants missing chemotherapy, hormone treatment, or comorbidity score. Adjustment for these variables would therefore result in loss of a large number of events. Similarly, adjustment for epidemiologic variables resulted in a large proportion of participants dropping out of our analyses due to missing data. To minimize bias due to drop-out and maximize power, we initially limited model covariates to age, ER, PR, tumor grade, tumor size, and radiation treatment. We then tested the addition of other covariates (comorbidity score, chemotherapy, hormone therapy, education, age at menarche, and menopausal status) to assess their impact on the SNP-survival outcome associations. Finally, we conducted sensitivity analyses in models adjusted for age, ER, PR, radiation, tumor grade, and tumor size by testing the effect of including BMI and excluding non-white participants on hazard ratio estimates.
Risk Allele Score Construction
We constructed a summary risk allele score using the log-additive model to estimate unadjusted per copy variant allele odds ratios and hazard ratios for lymph node status, tumor grade, time to recurrence, disease-free survival, and overall survival. We did not include tumor subtype in this analysis due to the complexity of creating a summary score for each subtype. Because the per copy variant allele effects were small in several instances, we considered odds ratios/hazard ratios that fell within the range 0.95 to 1.05 as being too close to null to assign a risk allele score. When this occurred, 0 risk alleles were assigned for all genotypes. If the per copy variant allele odds ratio/hazard ratio for a given SNP was greater than or equal to 1.06, genotypes for that SNP were assigned a score based on the following scheme: homozygous wildtype genotype=0 risk alleles; heterozygous genotype=1 risk allele; homozygous variant genotype=2 risk alleles. If the odds ratio/hazard ratio was less than or equal to 0.94, the coding scheme was reversed: homozygous wildtype=2 risk alleles; heterozygous=1 risk allele; homozygous variant=0 risk alleles. The number of risk alleles for each of the 7 SNPs was then added together for each participant to create the summary risk allele score. The directions of the log-additive odds and hazard ratios for each SNP were not generally similar across the lymph node status, tumor grade and survival analyses, leading to the assignment of different numbers of risk alleles for each outcome. Using the distributions of risk alleles for the lymph node status, tumor grade, and survival analyses, we categorized the number of risk alleles as 5 or less, 6, 7, and 8 or more to create a summary risk allele score. The category “5 or less” risk alleles served as the reference category. We then estimated the odds/hazard ratios for each level of the risk allele score, using logistic and proportional hazards regression as described above. Regression models included the same covariates as described previously for the lymph node status, tumor grade, and survival analyses.
Results
Participant characteristics are shown In Table 1. Younger participants were more likely to have node positive, high grade tumors of the luminal B, HER2-enriched, and triple negative subtypes. Higher educational attainment was associated with decreased likelihood of node positive tumors. Ever users of HRT were more likely to have luminal B and HER2-enriched tumors. There were no other significant differences in demographic and reproductive variables with respect to tumor characteristics. Lymph node metastases were more commonly observed in conjunction with high grade, larger size, ER/PR negative, and HER2 positive tumors.
Table 1.
Participant Characteristics by Lymph Node Status, Tumor Subtype, and Tumor Grade.
| Lymph Node Status2 | Tumor Subtype2 | Tumor Grade2 | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Characteristic1, n (%) | Positive (N=246) | Negative (N=603) | Luminal A (N=596) | Luminal B (N=75) | HER2 (+) (N=34) | Triple (−) (N=138) | Low (N=225) | Moderate (N=371) | High (N=248) |
| Age at diagnosis | |||||||||
| ≤50 | 97 (39.4) | 177 (29.4)* | 174 (29.2) | 29 (38.7) | 15 (44.1) | 51 (37.0) | 62 (27.6) | 115 (31.0) | 93 (37.5) |
| 51-65 | 89 (36.2) | 253 (42.0) | 246 (41.3) | 28 (37.3) | 15 (44.1) | 50 (36.2) | 95 (42.2) | 141 (38.0) | 99 (39.9) |
| ≥66 | 60 (24.4) | 173 (28.7) | 176 (29.5) | 18 (24.0) | 4 (11.8) | 37 (26.8) | 68 (30.2) | 115 (31.0) | 56 (22.6) |
| Race | |||||||||
| White | 171 (90.5) | 463 (94.5) | 450 (94.5) | 60 (92.3) | 19 (86.4) | 103 (91.2) | 172 (96.1) | 277 (93.9) | 184 (90.6) |
| Non-white | 18 (9.5) | 27 (5.5) | 26 (5.5) | 5 (7.7) | 3 (13.6) | 10 (8.8) | 7 (3.9) | 18 (6.1) | 19 (9.4) |
| Missing | 57 | 113 | 120 | 10 | 12 | 25 | 46 | 76 | 45 |
| Education | |||||||||
| High school or less | 79 (41.8) | 166 (34.5) | 180 (38.4) | 19 (29.7) | 8 (36.4) | 39 (34.8) | 63 (36.0) | 109 (37.3) | 76 (37.8) |
| At least some college | 110 (58.2) | 315 (65.5) | 289 (61.6) | 45 (70.3) | 14 (63.6) | 73 (65.2) | 112 (64.0) | 183 (62.7) | 125 (62.2) |
| Missing | 57 | 122 | 127 | 11 | 12 | 26 | 50 | 79 | 47 |
| Menopausal status | |||||||||
| Pre | 72 (38.7) | 157 (32.2) | 150 (31.8) | 29 (44.6) | 7 (31.8) | 39 (34.8) | 58 (32.8) | 92 (31.4) | 77 (38.1) |
| Post | 114 (61.3) | 331 (67.8) | 322 (68.2) | 36 (55.4) | 15 (68.2) | 73 (65.2) | 119 (67.2) | 201 (68.6) | 125 (61.9) |
| Missing | 60 | 115 | 124 | 10 | 12 | 26 | 48 | 78 | 46 |
| HRT use3 | |||||||||
| Never | 68 (60.7) | 170 (53.1) | 183 (58.8) | 15 (42.9) | 4 (26.7) | 39 (54.2)* | 66 (57.4) | 113 (58.2) | 62 (50.4) |
| Ever | 44 (39.3) | 150 (46.9) | 128 (41.2) | 20 (57.1) | 11 (73.3) | 33 (45.8) | 49 (42.6) | 81 (41.8) | 61 (49.6) |
| Missing | 2 | 11 | 11 | 1 | 0 | 1 | 4 | 7 | 2 |
| Parity | |||||||||
| Nulliparous | 28 (15.1) | 94 (19.2) | 84 (17.8) | 17 (26.2) | 3 (13.6) | 14 (12.5) | 34 (19.0) | 51 (17.4) | 35 (17.4) |
| Parous | 158 (84.9) | 395 (80.8) | 389 (82.2) | 48 (73.8) | 19 (86.4) | 98 (87.5) | 145 (81.0) | 242 (82.6) | 166 (82.6) |
| Missing | 60 | 114 | 123 | 10 | 12 | 26 | 46 | 78 | 47 |
| ER status | |||||||||
| Positive | 185 (75.2) | 479 (80.0) | 590 (99.0) | 73 (97.3) | 0 | 0 | 220 (98.2) | 332 (89.5) | 113 (45.7)* |
| Negative | 61 (24.8) | 120 (20.0) | 6 (1.0) | 2 (2.7) | 34 | 138 | 4 (1.8) | 39 (10.5) | 134 (54.3) |
| Missing | 0 | 4 | 0 | 0 | 0 | 0 | 1 | 0 | 1 |
| PR status | |||||||||
| Positive | 154 (62.6) | 420 (70.1)* | 512 (85.9) | 63 (84.0) | 0 | 0 | 200 (89.3) | 281 (75.7) | 91 (36.8)* |
| Negative | 92 (37.4) | 179 (29.9) | 84 (14.1) | 12 (16.0) | 34 | 138 | 24 (10.7) | 90 (24.3) | 156 (63.2) |
| Missing | 0 | 4 | 0 | 0 | 0 | 0 | 1 | 0 | 1 |
| HER2 status | |||||||||
| Positive | 44 (17.9) | 65 (11.0)* | 0 | 75 | 34 | 0 | 4 (1.8) | 46 (12.6) | 55 (22.4)* |
| Negative | 202 (82.1) | 524 (89.0) | 596 | 0 | 0 | 138 | 217 (98.2) | 320 (87.4) | 190 (77.6) |
| Missing | 0 | 14 (2.3) | 0 | 0 | 0 | 0 | 4 | 5 | 3 |
| Lymph node status | |||||||||
| Positive | 246 | - | 163 (27.7) | 23 (30.7) | 21 (61.8) | 39 (28.5)* | 47 (21.4) | 106 (28.8) | 90 (36.6)* |
| Negative | - | 603 | 425 (72.3) | 52 (69.3) | 13 (38.2) | 98 (71.5) | 173 (78.6) | 262 (71.2) | 156 (63.4) |
| Missing | - | - | 8 | 0 | 0 | 1 | 5 | 3 | 2 |
| Tumor subtype | |||||||||
| Luminal A | 163 (66.3) | 425 (72.3)* | 596 | - | - | - | 213 (96.8) | 293 (80.1) | 84 (34.3)* |
| Luminal B | 23 (9.4) | 52 (8.8) | - | 75 | - | - | 4 (1.8) | 37 (10.1) | 32 (13.1) |
| HER2 (+) | 21 (8.5) | 13 (2.2) | - | - | 34 | - | 0 | 9 (2.5) | 23 (9.4) |
| Triple negative | 39 (15.9) | 98 (16.7) | - | - | - | 138 | 3 (1.4) | 27 (7.4) | 106 (43.3) |
| Missing | 0 | 15 | - | - | - | - | 5 | 5 | 3 |
| Tumor grade | |||||||||
| Low | 47 (19.3) | 173 (29.3)* | 213 (36.1) | 4 (5.5) | 0 | 3 (2.2)* | 225 | - | - |
| Moderate | 106 (43.6) | 262 (44.3) | 293 (49.7) | 37 (50.7) | 9 (28.1) | 27 (19.9) | - | 371 | - |
| High | 90 (37.1) | 156 (26.4) | 84 (14.2) | 32 (43.8) | 23 (71.9) | 106 (77.9) | - | - | 248 |
| Missing | 3 | 12 | 6 | 2 | 2 | 2 | - | - | - |
| Tumor size | |||||||||
| Tmi , T1A (≤ 5mm) | 14 (5.7) | 89 (14.8)* | 67 (11.2) | 6 (8.0) | 5 (14.7) | 14 (10.1)* | 37 (16.4) | 40 (10.8) | 18 (7.3)* |
| T1B, T1C (> 5-20mm) | 116 (47.2) | 413 (68.5) | 398 (66.8) | 45 (60.0) | 14 (41.2) | 78 (56.5) | 159 (70.7) | 239 (64.4) | 136 (54.8) |
| T2 (> 20-50mm) | 96 (39.0) | 96 (15.9) | 115 (19.3) | 23 (30.7) | 12 (35.3) | 41 (29.7) | 24 (10.7) | 82 (22.1) | 84 (33.9) |
| T3 (> 50mm) | 20 (8.1) | 5 (0.8) | 16 (2.7) | 1 (1.3) | 3 (8.8) | 5 (3.6) | 5 (2.2) | 10 (2.7) | 10 (4.0) |
| Stage | |||||||||
| 1 | 2 (0.9) | 498 (83.4) | 366 (62.8) | 42 (56.0) | 9 (26.5) | 71 (51.8)* | 156 (72.2) | 225 (61.3) | 111 (45.1)* |
| 2 | 164 (66.9) | 99 (16.6) | 170 (29.2) | 27 (36.0) | 15 (44.1) | 50 (36.5) | 48 (22.2) | 115 (31.3) | 96 (39.0) |
| 3 | 79 (32.2) | 0 | 47 (8.0) | 6 (8.0) | 10 (29.4) | 16 (11.7) | 12 (5.6) | 27 (7.4) | 39 (15.9) |
| Missing | 1 | 6 | 13 | 0 | 0 | 1 | 9 | 4 | 2 |
Race, education, menopausal status, HRT use, and parity were available for 688 participants; age at diagnosis, ER, PR, HER2 status, tumor grade, and tumor size were available for 859 participants.
Lymph node status, subtype, and grade were missing for n=10, n=16, and n=15, respectively.
Excludes premenopausal women.
Significant at α=0.05; p-values were obtained from Chi-squared and Fisher's exact tests as appropriate (missing categories are excluded from p-value calculations).
HRT=hormone replacement therapy; ER=estrogen receptor; PR=progesterone receptor.
SNP and lymph node status associations are presented in Table 2. BRMS1 rs11537993 and SIPA1 rs75894763, rs746429, and rs2306364 were not significantly associated with lymph node metastases. Age-adjusted odds ratios were similar to those obtained in multivariate models (data not shown). BRMS1 rs1052566 heterozygotes were more likely to have node positive tumors (OR=1.58, 95% CI 1.13-2.23), which remained significant after additional adjustment for race, HRT use, education, and menopausal status (OR=1.70, 95% CI 1.13-2.55) and when we limited the analysis to participants with stage 2 and 3 tumors (OR=1.85, 95% CI 1.08-3.18). However, this relationship was not observed among homozygous individuals. Although only marginally significant, participants with at least one copy of the variant SIPA1 rs3741378 allele were less likely to have node positive tumors (OR=0.70, 95% CI 0.48-1.02). While the direction and magnitude persisted following adjustment for additional covariates and limitation to stage 2 and 3 tumors, this association became nonsignificant. Similarly, BRMS1 rs3116068 approached statistical significance only when race, HRT use, education and menopausal status were added as covariates (OR=0.67, 95% CI 0.45-1.01).
Table 2.
Associations of BRMS1 and SIPA1 SNPs with Presence of Lymph Node Metastases at Diagnosis.
| Odds Ratio for Likelihood of Node Positive Tumor at Diagnosis | |||||||
|---|---|---|---|---|---|---|---|
|
|
|||||||
| Genotype | Node positive, n | Node Negative, n | Adjusted1 OR (95% CI) | Adjusted2 OR (95% CI) | Adjusted3 OR (95% CI) | ||
| BRMS1 | rs11537993 | AA | 125 | 293 | 1.00 | 1.00 | 1.00 |
| (Leu67Leu) | AG | 98 | 243 | 0.93 (0.66-1.30) | 1.02 (0.69-1.53) | 0.98 (0.60-1.63) | |
| GG | 20 | 50 | 0.90 (0.49-1.63) | 1.01 (0.53-1.92) | 1.14 (0.44-2.93) | ||
| AG+GG | 0.92 (0.67-1.27) | 1.02 (0.70-1.49) | 1.01 (0.62-1.63) | ||||
| Ptrend | 0.61 | 0.95 | 0.89 | ||||
| rs3116068 | GG | 165 | 365 | 1.00 | 1.00 | 1.00 | |
| (3′ UTR) | AG | 66 | 183 | 0.80 (0.56-1.14) | 0.66 (0.43-1.02) | 0.69 (0.41-1.18) | |
| AA | 10 | 34 | 0.66 (0.30-1.43) | 0.73 (0.31-1.72) | 0.59 (0.18-1.89) | ||
| AG+AA | 0.78 (0.55-1.09) | 0.67 (0.45-1.01) | 0.68 (0.41-1.13) | ||||
| Ptrend | 0.13 | 0.09 | 0.13 | ||||
| rs1052566 | GG | 102 | 285 | 1.00 | 1.00 | 1.00 | |
| (Ala273Val) | AG | 115 | 220 | 1.58 (1.13-2.23) | 1.70 (1.13-2.55) | 1.85 (1.08-3.18) | |
| AA | 21 | 61 | 0.92 (0.51-1.65) | 1.06 (0.54-2.08) | 0.72 (0.33-1.55) | ||
| AG+AA | 1.43 (1.03-1.99) | 1.56 (1.06-2.30) | 1.48 (0.91-2.42) | ||||
| Ptrend | 0.25 | 0.17 | 0.67 | ||||
| SIPA1 | rs75894763 | GG | 237 | 558 | 1.00 | 1.00 | 1.00 |
| (Val621Val) | AG | 5 | 29 | 0.39 (0.14-1.08) | 0.51 (0.16-1.60) | 0.30 (0.09-1.03) | |
| rs746429 | GG | 110 | 267 | 1.00 | 1.00 | 1.00 | |
| (Ala920Ala) | GA | 102 | 239 | 0.99 (0.70-1.40) | 1.06 (0.71-1.59) | 0.86 (0.52-1.43) | |
| AA | 29 | 66 | 1.09 (0.65-1.83) | 1.16 (0.64-2.12) | 1.31 (0.55-3.14) | ||
| GA+AA | 1.01 (0.73-1.40) | 1.08 (0.74-1.58) | 0.93 (0.57-1.51) | ||||
| Ptrend | 0.82 | 0.61 | 0.88 | ||||
| rs3741378 | CC | 187 | 413 | 1.00 | 1.00 | 1.00 | |
| (Ser182Phe) | TC | 49 | 148 | 0.71 (0.48-1.04) | 0.75 (0.47-1.19) | 0.73 (0.41-1.29) | |
| TT | 4 | 13 | 0.62 (0.18-2.10) | 0.61 (0.14-2.67) | 0.42 (0.09-2.05) | ||
| TC+TT | 0.70 (0.48-1.02) | 0.74 (0.47-1.16) | 0.70 (0.40-1.21) | ||||
| Ptrend | 0.07 | 0.18 | 0.15 | ||||
| rs2306364 | GG | 186 | 425 | 1.00 | 1.00 | 1.00 | |
| (Ala342Ala) | AG | 13 | 42 | 0.71 (0.35-1.41) | 0.70 (0.34-1.48) | 0.89 (0.30-2.63) | |
| AA | 42 | 99 | 0.98 (0.64-1.50) | 0.92 (0.56-1.53) | 1.14 (0.58-2.24) | ||
| AG+AA | 0.90 (0.62-1.32) | 0.85 (0.55-1.32) | 1.07 (0.59-1.95) | ||||
| Ptrend | 0.78 | 0.62 | 0.75 | ||||
Unconditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CI) for risk of node positive tumors.
Adjusted for age at diagnosis, ER, PR, tumor size, and tumor grade.
Adjusted for age at diagnosis, ER, PR, HER2, tumor size, tumor grade, race, HRT, education, menopausal status.
Limited to women with stage 2 and 3 breast cancer, adjusted for age at diagnosis, ER/PR status, and tumor grade. Tumor size was not included as a covariate due to sparse data.
SNP associations with tumor grade are presented in Table 3. SIPA1 rs75894763 heterozygous participants were more likely to have high grade tumors (OR=2.62, 95% CI 1.06-6.48), but only after further adjustment for race, HRT use, and menopausal status.
Table 3.
Associations of BRMS1 and SIPA1 SNPs with Tumor Grade.
| Odds Ratio for Likelihood of High Grade Tumor at Diagnosis | ||||||
|---|---|---|---|---|---|---|
|
|
||||||
| Genotype | High Grade, n | Low/Moderate Grade, n | Adjusted1 OR (95% CI) | Adjusted2 OR (95% CI) | ||
| BRMS1 | rs11537993 | AA | 128 | 286 | 1.00 | 1.00 |
| (Leu67Leu) | AG | 92 | 243 | 0.86 (0.58-1.27) | 0.97 (0.62-1.53) | |
| GG | 23 | 47 | 1.23 (0.65-2.35) | 1.36 (0.68-2.70) | ||
| AG+GG | 0.92 (0.64-1.33) | 1.04 (0.68-1.60) | ||||
| Ptrend | 0.97 | 0.55 | ||||
| rs3116068 | GG | 156 | 369 | 1.00 | 1.00 | |
| (3′ UTR) | AG | 70 | 175 | 0.83 (0.55-1.26) | 0.98 (0.61-1.56) | |
| AA | 11 | 32 | 0.82 (0.36-1.89) | 0.67 (0.26-1.75) | ||
| AG+AA | 0.83 (0.56-1.23) | 0.92 (0.59-1.44) | ||||
| Ptrend | 0.38 | 0.54 | ||||
| rs1052566 | GG | 113 | 267 | 1.00 | 1.00 | |
| (Ala273Val) | AG | 92 | 240 | 0.86 (0.58-1.28) | 0.97 (0.61-1.53) | |
| AA | 27 | 55 | 1.39 (0.76-2.57) | 1.63 (0.82-3.25) | ||
| AG+AA | 0.95 (0.66-1.38) | 1.08 (0.70-1.66) | ||||
| Ptrend | 0.67 | 0.33 | ||||
| SIPA1 | rs75894763 | GG | 233 | 553 | 1.00 | 1.00 |
| (Val621Val) | AG | 10 | 23 | 1.97 (0.87-4.47) | 2.62 (1.06-6.48) | |
| rs746429 | GG | 114 | 259 | 1.00 | 1.00 | |
| (Ala920Ala) | GA | 104 | 232 | 1.20 (0.81-1.78) | 1.28 (0.82-2.00) | |
| AA | 21 | 73 | 0.82 (0.44-1.54) | 0.83 (0.41-1.70) | ||
| GA+AA | 1.11 (0.77-1.61) | 1.17 (0.77-1.79) | ||||
| Ptrend | 0.98 | 0.91 | ||||
| rs3741378 | CC | 166 | 429 | 1.00 | 1.00 | |
| (Ser182Phe) | TC | 65 | 127 | 1.44 (0.94-2.20) | 1.46 (0.89-2.39) | |
| TT | 6 | 11 | 0.99 (0.27-3.60) | 1.43 (0.36-5.71) | ||
| TC+TT | 1.40 (0.92-2.11) | 1.46 (0.90-2.35) | ||||
| Ptrend | 0.18 | 0.14 | ||||
| rs2306364 | GG | 187 | 419 | 1.00 | 1.00 | |
| (Ala342Ala) | AG | 13 | 40 | 0.84 (0.40-1.77) | 0.72 (0.32-1.61) | |
| AA | 36 | 103 | 0.93 (0.57-1.53) | 0.93 (0.53-1.63) | ||
| AG+AA | 0.90 (0.58-1.40) | 0.86 (0.52-1.40) | ||||
| Ptrend | 0.71 | 0.67 | ||||
Unconditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CI) for risk of high grade tumors.
Low grade = well-differentiated; Moderate grade = moderately differentiated; High grade = poorly differentiated and undifferentiated.
Adjusted for age at diagnosis, tumor size, lymph node status, ER, PR, and HER2.
Adjusted for age at diagnosis, tumor size, lymph node status, ER, PR, HER2, race, HRT, and menopausal status.
Associations between tumor subtype and genotype are shown in Table 4. Results of age- and tumor size adjusted analyses were similar to the findings presented in Table 4 and are not shown. The BRMS1 rs3116068 homozygous variant genotype was associated with increased likelihood of luminal B tumors (OR=2.50, 95% CI 1.10-5.66), although there was a nonsignificant inverse relationship among heterozygotes. Those who were heterozygous or homozygous were also more likely to have tumors of the HER2-enriched subtype (OR=2.45, 95% CI 1.18-5.06, Ptrend = 0.03). These relationships remained significant after additional adjustment for race, HRT use, and menopausal status (data not shown). Patients homozygous or heterozygous for BRMS1 rs1052566 were less likely to have luminal B tumors (OR=0.59, 95% CI 0.36-0.98, Ptrend = 0.05). These relationships remained marginally significant after further adjustment for race, HRT use, and menopausal status (OR=0.59, 95% CI 0.34-1.02). A nonsignificant decrease in the likelihood of the HER2-enriched subtype was observed among patients heterozygous or homozygous for this variant, which became statistically significant when race, HRT use, and menopausal status were included in the model (OR=0.32, 95% CI 0.12-0.90). Participants homozygous for SIPA1 rs746429 were less likely to have triple negative tumors (OR=0.48, 95% CI 0.24-0.97, Ptrend = 0.04). Similarly, participants either homozygous or heterozygous for SIPA1 rs2306364 were less likely to have triple negative tumors, although this relationship was only borderline significant (OR=0.62, 95% CI 0.38-1.01).
Table 4.
Associations of BRMS1 and SIPA1 SNPs with Tumor Subtype.
| Odds Ratio for Likelihood of Luminal B, HER2 (+), and Triple (-) Subtype, Compared to Luminal A | |||||||||
|---|---|---|---|---|---|---|---|---|---|
|
|
|||||||||
| Genotype | Luminal A, n | Luminal B, n | Adjusted2 OR (95% CI) | HER2 (+), n | Adjusted2 OR (95% CI) | Triple (−), n | Adjusted2 OR (95% CI) | ||
| BRMS1 | rs11537993 | AA | 292 | 33 | 1.00 | 18 | 1.00 | 77 | 1.00 |
| (Leu67Leu) | AG | 246 | 31 | 1.14 (0.68-1.93) | 15 | 1.13 (0.55-2.34) | 48 | 0.74 (0.50-1.11) | |
| GG | 49 | 10 | 1.85 (0.85-4.01) | 1 | 0.36 (0.05-2.82) | 11 | 0.74 (0.50-1.11) | ||
| AG+GG | 1.26 (0.77-2.06) | 0.99 (0.49-2.03) | 0.76 (0.52-1.12) | ||||||
| Ptrend | 0.17 | 0.64 | 0.26 | ||||||
| rs3116068 | GG | 379 | 50 | 1.00 | 15 | 1.00 | 90 | 1.00 | |
| (3′ UTR) | AG | 174 | 16 | 0.71 (0.39-1.28) | 16 | 2.49 (1.18-5.27) | 41 | 1.00 (0.66-1.51) | |
| AA | 28 | 9 | 2.50 (1.10-5.66) | 2 | 2.11 (0.45-10.0) | 5 | 0.76 (0.28-2.04) | ||
| AG+AA | 0.95 (0.57-1.59) | 2.45 (1.18-5.06) | 0.97 (0.65-1.44) | ||||||
| Ptrend | 0.39 | 0.03 | 0.74 | ||||||
| rs1052566 | GG | 268 | 44 | 1.00 | 18 | 1.00 | 58 | 1.00 | |
| (Ala273Val) | AG | 240 | 24 | 0.61 (0.36-1.03) | 13 | 0.70 (0.33-1.49) | 57 | 1.12 (0.74-1.68) | |
| AA | 61 | 6 | 0.55 (0.22-1.36) | 1 | 0.21 (0.03-1.66) | 15 | 1.07 (0.57-2.03) | ||
| AG+AA | 0.59 (0.36-0.98) | 0.60 (0.29-1.25) | 1.11 (0.75-1.63) | ||||||
| Ptrend | 0.05 | 0.09 | 0.69 | ||||||
| SIPA1 | rs75894763 | GG | 559 | 69 | 1.00 | 34 | 1.00 | 135 | 1.00 |
| (Val621Val) | AG | 27 | 5 | 1.47 (0.54-3.99) | 0 | NA | 2 | 0.29 (0.07-1.26) | |
| rs746429 | GG | 255 | 35 | 1.00 | 19 | 1.00 | 69 | 1.00 | |
| (Ala920Ala) | GA | 245 | 29 | 0.85 (0.50-1.44) | 11 | 0.56 (0.26-1.22) | 55 | 0.81 (0.54-1.20) | |
| AA | 75 | 9 | 0.86 (0.39-1.87) | 3 | 0.51 (0.14-1.80) | 10 | 0.48 (0.24-0.97) | ||
| GA+AA | 0.85 (0.52-1.39) | 0.55 (0.27-1.13) | 0.73 (0.50-1.07) | ||||||
| Ptrend | 0.57 | 0.13 | 0.04 | ||||||
| rs3741378 | CC | 429 | 53 | 1.00 | 27 | 1.00 | 93 | 1.00 | |
| (Ser182Phe) | TC | 138 | 17 | 0.99 (0.55-1.78) | 4 | 0.46 (0.16-1.37) | 37 | 1.21 (0.79-1.87) | |
| TT | 9 | 3 | 2.91 (0.75-11.2) | 2 | 5.77 (1.11-29.8) | 3 | 1.57 (0.41-5.98) | ||
| TC+TT | 1.11 (0.64-1.92) | 0.68 (0.27-1.71) | 1.24 (0.81-1.88) | ||||||
| Ptrend | 0.45 | 0.90 | 0.30 | ||||||
| rs2306364 | GG | 421 | 58 | 1.00 | 27 | 1.00 | 107 | 1.00 | |
| (Ala342Ala) | AG | 42 | 6 | 1.02 (0.41-2.51) | 1 | 0.38 (0.05-2.94) | 6 | 0.56 (0.23-1.36) | |
| AA | 109 | 9 | 0.59 (0.28-1.23) | 5 | 0.74 (0.27-2.00) | 18 | 0.64 (0.37-1.11) | ||
| AG+AA | 0.71 (0.39-1.29) | 0.64 (0.26-1.60) | 0.62 (0.38-1.01) | ||||||
| Ptrend | 0.18 | 0.44 | 0.07 | ||||||
Multinomial logistic regression (generalized logit model) was used to estimate odds ratios (OR) and 95% confidence intervals (CI) for risk of luminal B, HER2 (+), or triple negative tumor subtype, using luminal A as the comparison group.
Luminal A = ER and/or PR positive, HER2 negative; Luminal B = ER and/or PR positive, HER2 positive; HER2 enriched subtype (HER2 (+)) = ER and PR negative, HER2 positive; Triple negative (Triple (−)) = ER, PR, and HER2 negative.
Adjusted for age at diagnosis, tumor size, and lymph node status.
Associations with survival outcomes are shown in Table 5. The median follow-up time was 45 months (range 4-101 months), during which 58 recurrences and 70 deaths from all causes occurred. When adjusted for age only, the heterozygous genotype of BRMS1 rs3116068 was associated with poorer overall survival (HR=1.65, 95% CI 1.02-2.68), but this association became nonsignificant when additional covariates were included. We did not observe any other significant associations. Results were unchanged when additional covariates (chemotherapy, hormone therapy, education, age at menarche, and menopausal status) were included (data not shown).
Table 5.
BRMS1 and SIPA1 SNP Associations with Time to Recurrence, Disease-Free Survival, and Overall Survival.
| Genotype | No. Events, TTR | TTR, Adjusted1 HR (95% CI) | No. Events, DFS | DFS, Adjusted1 HR (95% CI) | No. Events, OS | OS, Adjusted1 HR (95% CI) | ||
|---|---|---|---|---|---|---|---|---|
| BRMS1 | rs11537993 | AA | 27 | 1.00 | 39 | 1.00 | 28 | 1.00 |
| (Leu67Leu) | AG | 21 | 1.05 (0.59-1.89) | 35 | 1.12 (0.70-1.79) | 28 | 1.10 (0.64-1.89) | |
| GG | 5 | 1.24 (0.47-3.30) | 7 | 1.18 (0.52-2.69) | 6 | 1.30 (0.53-3.20) | ||
| AG+GG | 1.08 (0.62-1.89) | 1.13 (0.72-1.77) | 1.13 (0.67-1.89) | |||||
| Ptrend | 0.69 | 0.59 | 0.56 | |||||
| rs3116068 | GG | 35 | 1.00 | 48 | 1.00 | 33 | 1.00 | |
| (3′ UTR) | AG | 15 | 0.77 (0.41-1.43) | 29 | 1.04 (0.65-1.67) | 26 | 1.44 (0.85-2.45) | |
| AA | 3 | 0.90 (0.26-3.13) | 3 | 0.54 (0.16-1.82) | 1 | 0.33 (0.04-2.49) | ||
| AG+AA | 0.78 (0.43-1.42) | 0.97 (0.61-1.55) | 1.31 (0.77-2.22) | |||||
| Ptrend | 0.51 | 0.60 | 0.76 | |||||
| rs1052566 | GG | 26 | 1.00 | 41 | 1.00 | 33 | 1.00 | |
| (Ala273Val) | AG | 22 | 1.05 (0.59-1.87) | 33 | 0.99 (0.62-1.57) | 23 | 0.86 (0.50-1.48) | |
| AA | 4 | 1.01 (0.35-2.96) | 5 | 0.97 (0.38-2.48) | 3 | 0.81 (0.24-2.67) | ||
| AG+AA | 1.04 (0.60-1.81) | 0.98 (0.63-1.54) | 0.85 (0.51-1.44) | |||||
| Ptrend | 0.92 | 0.93 | 0.55 | |||||
| SIPA1 | rs75894763 | GG | 51 | 1.00 | 78 | 1.00 | 60 | 1.00 |
| (Val621Val) | AG | 2 | 1.61 (0.38-6.89) | 3 | 1.57 (0.48-5.11) | 2 | 1.35 (0.32-5.75) | |
| rs746429 | GG | 24 | 1.00 | 37 | 1.00 | 28 | 1.00 | |
| (Ala920Ala) | GA | 25 | 1.20 (0.67-2.14) | 38 | 1.15 (0.72-1.83) | 30 | 1.24 (0.73-2.10) | |
| AA | 5 | 0.94 (0.36-2.50) | 7 | 0.86 (0.38-1.94) | 4 | 0.69 (0.24-1.99) | ||
| GA+AA | 1.14 (0.66-1.99) | 1.09 (0.70-1.70) | 1.13 (0.67-1.89) | |||||
| Ptrend | 0.84 | 0.98 | 0.94 | |||||
| rs3741378 | CC | 37 | 1.00 | 58 | 1.00 | 45 | 1.00 | |
| (Ser182Phe) | TC | 17 | 1.28 (0.71-2.30) | 24 | 1.09 (0.67-1.77) | 16 | 0.86 (0.48-1.55) | |
| TT | 0 | NA | 0 | NA | 1 | 1.16 (0.16-8.63) | ||
| TC+TT | 1.20 (0.67-2.16) | 1.03 (0.63-1.67) | 0.87 (0.49-1.55) | |||||
| Ptrend | NA | NA | 0.70 | |||||
| rs2306364 | GG | 41 | 1.00 | 61 | 1.00 | 48 | 1.00 | |
| (Ala342Ala) | AG | 5 | 1.16 (0.45-3.01) | 7 | 1.20 (0.54-2.71) | 4 | 0.92 (0.32-2.64) | |
| AA | 8 | 0.82 (0.38-1.75) | 13 | 0.91 (0.50-1.66) | 9 | 0.80 (0.39-1.65) | ||
| AG+AA | 0.92 (0.49-1.73) | 0.99 (0.59-1.66) | 0.83 (0.45-1.56) | |||||
| Ptrend | 0.67 | 0.84 | 0.54 |
Hazard ratios (HR) and 95% confidence intervals (CI) were estimated using Cox proportional hazards regression.
Adjusted for age at diagnosis, ER, PR, tumor size, tumor grade, radiation treatment, and Charlson comorbidity score.
TTR=Time to recurrence (time from diagnosis to date of first recurrence or date of last follow-up); DFS=Disease-free survival (time from diagnosis to date of first recurrence, death, or last follow-up); OS=Overall survival (time from diagnosis to date of death or date of last follow-up).
Patients whose recurrence status indicated that they were never disease free were excluded from time to recurrence and disease-free survival analyses, but were included in overall survival analyses.
One of our hypotheses is that these SNPs would be related to lymph node status. Because nodal status is also related to survival outcomes, it can be hypothesized that this variable is part of the causal pathway, and therefore adjustment for nodal status could mask true SNP-survival associations. To test whether these SNPs could affect vital status or recurrence through a pathway independent of lymph node status, we included nodal status in a model containing age, ER, PR, radiation, tumor grade, and tumor size as covariates. There was no change in the hazard ratio estimates for any of the three outcomes (data not shown).
Results of the summary risk allele score analysis are shown in Table 6. Having 8 or more risk alleles was associated with significantly greater likelihood of having a node positive tumor (OR=2.14 95% CI 1.18-3.86). There was evidence of a dose-response pattern, with increasing numbers of risk alleles associated with increased likelihood of node positivity (Ptrend = 0.002). There were no significant associations with tumor grade or the three survival outcomes.
Table 6.
Risk Allele Score Associations with Lymph Node Status, Tumor Grade, Time to Recurrence, Disease-Free Survival, and Overall Survival.
| Lymph Node Status | Tumor Grade | Time to Recurrence | Disease-Free Survival | Overall Survival | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Number of Alleles | Positive (n) | Negative (n) | OR1 (95% CI) | High (n) | Low/Mod (n) | OR2 (95% CI) | N* | HR3 (95% CI) | N* | HR3 (95% CI) | N* | HR3 (95% CI) |
| 5 or less | 18 | 73 | 1.00 | 153 | 362 | 1.00 | 19 | 1.00 | 15 | 1.00 | 34 | 1.00 |
| 6 | 50 | 148 | 1.31 (0.69-2.50) | 53 | 150 | 0.85 (0.54-1.32) | 21 | 1.83 (0.96-3.48) | 32 | 1.84 (0.98-3.46) | 20 | 1.19 (0.67-2.10) |
| 7 | 44 | 117 | 1.38 (0.71-2.67) | 33 | 61 | 1.19 (0.67-2.10) | 12 | 1.22 (0.59-2.54) | 15 | 0.86 (0.41-1.82) | 8 | 1.74 (0.79-3.83) |
| 8 or more | 131 | 251 | 2.14 (1.18-3.86) | 4 | 6 | 1.28 (0.28-5.91) | 2 | 0.50 (0.11-2.17) | 20 | 1.13 (0.57-2.24) | 0 | NA |
| Ptrend | 0.002 | 0.78 | 0.82 | 0.60 | 0.48 | |||||||
N is the number of participants in each risk allele score category.
N*is the number of events in each risk allele score category.
Adjusted for age at diagnosis, ER, PR, tumor size, and tumor grade.
Adjusted for age at diagnosis, ER, PR, HER2, tumor size, and lymph node status.
Adjusted for age at diagnosis, ER, PR, tumor size, tumor grade, radiation treatment, and Charlson comorbidity score.
In sensitivity analyses, nonwhite participants were excluded for all study outcomes, which did not alter our findings (data not shown).
Discussion
Our data suggest that 2 SNPs in the BRMS1 gene, rs1052566 and rs3116068, may be associated with lymph node status and tumor subtype, and that SNPs in the SIPA1 gene may be associated with tumor grade and subtype. We also found that a summary score, composed of the number of “at risk” alleles for each of the 7 BRMS1 and SIPA1 SNPs analyzed, was significantly associated with lymph node status.
To our knowledge, associations between SNPs in BRMS1 and breast tumor characteristics and prognosis have not been previously evaluated. BRMS1 has multiple functions, including transcriptional regulation via NF-κB signaling pathways [26, 27], chromatin modification [26], interactions with histone deacetylase complexes [27], and transcriptional repression of anti-apoptotic genes [12]. Previous studies have correlated decreased BRMS1 gene expression with breast tumor aggressiveness. Reduced mRNA and protein expression in breast tumors has been associated with PR-negative, HER2-positive tumors, as well as younger age at diagnosis [13, 14], but not with lymph node metastases [14, 15]. One study found that BRMS1 mRNA was reduced in brain metastases of breast cancer patients [16]. In survival analyses, both increased and decreased BRMS1 gene expression have been correlated with reduced survival [14, 17], while loss of BRMS1 protein expression has been associated with reduced survival only in patients with ER-negative, HER2-positive tumors [13].
We observed several relationships between SIPA1 SNPs rs746429, rs3741378, and rs2306364 and tumor subtype, although there are too few participants with variant alleles to draw strong conclusions. SIPA1 encodes a GTPase-activating protein and affects expression of extracellular matrix genes [19]. SIPA1 SNPs rs931127, rs3741378, and rs746429 have not been shown to be associated with overall survival [22], similar to our findings. SIPA1 rs3741378 has been correlated with increased risk of ER/PR-negative tumors, while rs746429 has been correlated with increased risk of node positive breast tumors [21]. In our study, rs3741378 was associated with an increased risk of HER2-enriched tumors, which are ER/PR-negative.
Our study used data and samples collected under the DBBR's standardized protocol and had relatively large sample size, but our analyses were limited by the small number of outcomes and participants with the variant genotypes. We also examined only a few SNPs in each gene. Our ability to adjust for socioeconomic and reproductive covariates was limited by missing epidemiologic questionnaire data, although these factors are unlikely to be strong confounders as SNPs are generally unlikely to be associated with them. Distributions of tumor characteristics and treatment were similar between participants with and without questionnaire data.
We compared the characteristics of participants with complete data to the characteristics of those with incomplete data to assess the possibility of bias. In general, participants with incomplete data were more likely to be postmenopausal, non-white, to have smaller tumors, and to have received hormonal treatment, but were less likely to have a family history of breast cancer and to have received radiation and chemotherapy. Age, ER, PR, HER2, tumor grade, comorbidity score, parity, age at menarche, and history of benign breast disease were generally similar. This indicates that the participants who dropped out of our analyses had less aggressive tumors than those who were included, but were similar with respect to other possible risk factors. Genotype frequencies were also not significantly different between groups, suggesting that a substantial bias is unlikely to be present.
The median follow-up time in this study was 45 months (3.8 years), which may have been too short to observe associations with recurrence and survival if there is a true effect of the SNPs on these outcomes. We did not correct for multiple comparisons and have not performed a replication study. The types of breast cancer patients treated at our institution and community-based facilities could be different with respect to tumor characteristics, possibly leading to our study population having a greater proportion of aggressive tumor characteristics than would be expected from the source population, which could affect the generalizability of our findings. It is possible, although less likely, that there could be differences with respect to genotype as well.
Conclusions
In conclusion, we showed that SNPs in BRMS1 and SIPA1 may be associated with tumor characteristics related to prognosis. Additional studies are needed to validate these findings and further investigate relationships between genetic variation in metastasis modifying genes and the metastatic phenotype. Understanding the biology of metastasis and identifying biomarkers of recurrence are necessary to improve prediction of the subset of patients who will experience metastatic relapse, particularly since treatment for breast cancer often results in significant patient morbidity. While many women will never progress to metastatic disease, identifying factors associated with metastatic recurrence is critical to achieving effective, efficient therapy for breast cancer patients.
Acknowledgments
This research was supported by a gift from the Jayne and Philip Hubbell Family. MRR is supported by DAMD W81XWH-11-1-0024, a predoctoral fellowship from the Department of Defense Breast Cancer Research Program, and previously supported by R25-CA11395102, an NCI predoctoral award. CBA is a recipient of funding from the Breast Cancer Research Foundation. Biospecimens and data were obtained from the RPCI DataBank and Biorepository, and genotyping was performed in the RPCI Genomics Core Facility, both of which are Cancer Center Support Grant Shared Resources supported by National Institutes of Health P30 CA016056-27. The funding agents played no role in any part of the study, including the design, collection, analysis and interpretation of the data, in the writing of the manuscript, or the decision to submit the manuscript for publication.
Abbreviations
- BMI
Body Mass Index
- BRMS1
Breast Cancer Metastasis Suppressor 1
- CI
Confidence Interval
- DBBR
DataBank and BioRepository
- DCIS
Ductal Carcinoma In Situ
- DFS
Disease Free Survival
- ER
Estrogen Receptor
- HER2
Human Epidermal Growth Factor Receptor 2
- HR
Hazard Ratio
- HRT
Hormone Replacement Therapy
- NCCN
National Comprehensive Cancer Network
- OR
Odds Ratio
- OS
Overall Survival
- PR
Progesterone Receptor
- RPCI
Roswell Park Cancer Institute
- SIPA1
Signal-Induced Proliferation-Associated 1
- SNP
Single Nucleotide Polymorphism
- TTR
Time to Recurrence
Footnotes
Conflict of Interests and Ethical Standards: The authors declare that they have no conflict of interests. This work complies with all ethical standards and current laws of the United States.
Contributor Information
Chi-Chen Hong, Email: Chi-Chen.Hong@RoswellPark.org.
Stephen B. Edge, Email: Stephen.Edge@RoswellPark.org.
Song Yao, Email: Song.Yao@RoswellPark.org.
Wiam Bshara, Email: Wiam.Bshara@RoswellPark.org.
Jo L. Freudenheim, Email: JFreuden@Buffalo.edu.
Christine B. Ambrosone, Email: Christine.Ambrosone@Roswellpark.org.
References
- 1.Cancer Facts and Figures 2013. American Cancer Society. 2013 [Google Scholar]
- 2.Higgins MJ, Wolff AC. Oncology. Vol. 22. Williston Park, NY: 2008. Therapeutic options in the management of metastatic breast cancer; pp. 614–23. discussion 623, 627–9. [PubMed] [Google Scholar]
- 3.Nguyen DX, Bos PD, Massagué J. Metastasis: from dissemination to organ-specific colonization. Nat Rev Cancer. 2009;9:274–84. doi: 10.1038/nrc2622. [DOI] [PubMed] [Google Scholar]
- 4.Nathanson SD, Kwon D, Kapke A, Alford SH, Chitale D. The role of lymph node metastasis in the systemic dissemination of breast cancer. Ann Surg Oncol. 2009;16:3396–405. doi: 10.1245/s10434-009-0659-2. [DOI] [PubMed] [Google Scholar]
- 5.Tresserra F, Rodriguez I, García-Yuste M, Grases PJ, Ara C, Fabregas R. Tumor size and lymph node status in multifocal breast cancer. Breast J. 2007;13:68–71. doi: 10.1111/j.1524-4741.2006.00365.x. [DOI] [PubMed] [Google Scholar]
- 6.Koscielny S, Arriagada R, Adolfsson J, Fornander T, Bergh J. Impact of tumour size on axillary involvement and distant dissemination in breast cancer. Brit J Cancer. 2009;101:902–7. doi: 10.1038/sj.bjc.6605221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Payne SJL, Bowen RL, Jones JL, Wells Ca. Predictive markers in breast cancer--the present. Histopathology. 2008;52:82–90. doi: 10.1111/j.1365-2559.2007.02897.x. [DOI] [PubMed] [Google Scholar]
- 8.Stafford LJ, Vaidya KS, Welch DR. Metastasis suppressors genes in cancer. Int J Biochem Cell B. 2008;40:874–91. doi: 10.1016/j.biocel.2007.12.016. [DOI] [PubMed] [Google Scholar]
- 9.Hedley BD, Vaidya KS, Phadke P, MacKenzie L, Dales DW, Postenka CO, MacDonald IC, Chambers AF. BRMS1 suppresses breast cancer metastasis in multiple experimental models of metastasis by reducing solitary cell survival and inhibiting growth initiation. Clin Exp Metastas. 2008;25:727–40. doi: 10.1007/s10585-008-9184-0. [DOI] [PubMed] [Google Scholar]
- 10.Seraj MJ, Samant RS, Verderame MF, Welch DR. Functional Evidence for a Novel Human Breast Carcinoma Metastasis Suppressor , BRMS1 , Encoded at Chromosome 11q13. Cancer Res. 2000;60:2764–2769. [PubMed] [Google Scholar]
- 11.Phadke Pa, Vaidya KS, Nash KT, Hurst DR, Welch DR. BRMS1 suppresses breast cancer experimental metastasis to multiple organs by inhibiting several steps of the metastatic process. Am J Path. 2008;172:809–17. doi: 10.2353/ajpath.2008.070772. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Liu Y, Smith PW, Jones DR. Breast cancer metastasis suppressor 1 functions as a corepressor by enhancing histone deacetylase 1-mediated deacetylation of RelA/p65 and promoting apoptosis. Mol Cell Biol. 2006;26:8683–96. doi: 10.1128/MCB.00940-06. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hicks DG, Yoder BJ, Short S, Tarr S, Prescott N, Crowe JP, Dawson AE, Budd GT, Sizemore S, Cicek M, Choueiri TK, Tubbs RR, Gaile D, Nowak N, Accavitti-Loper MA, Frost AR, Welch DR, Casey G. Loss of breast cancer metastasis suppressor 1 protein expression predicts reduced disease-free survival in subsets of breast cancer patients. Clin Cancer Res. 2006;12:6702–8. doi: 10.1158/1078-0432.CCR-06-0635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhang Z, Yamashita H, Toyama T, Yamamoto Y, Kawasoe T, Iwase H. Reduced expression of the breast cancer metastasis suppressor 1 mRNA is correlated with poor progress in breast cancer. Clin Cancer Res. 2006;12:6410–4. doi: 10.1158/1078-0432.CCR-06-1347. [DOI] [PubMed] [Google Scholar]
- 15.Kelly LM, Buggy Y, Hill A, O'Donovan N, Duggan C, McDermott EW, O'Higgins NJ, Young L, Duffy MJ. Expression of the breast cancer metastasis suppressor gene, BRMS1, in human breast carcinoma: lack of correlation with metastasis to axillary lymph nodes. Tumour Biol. 2005;26:213–6. doi: 10.1159/000086955. [DOI] [PubMed] [Google Scholar]
- 16.Stark AM, Tongers K, Maass N, Mehdorn HM, Held-Feindt J. Reduced metastasis-suppressor gene mRNA-expression in breast cancer brain metastases. J Cancer Res Clin. 2005;131:191–8. doi: 10.1007/s00432-004-0629-9. [DOI] [PubMed] [Google Scholar]
- 17.Lombardi G, Di Cristofano C, Capodanno A, Iorio MC, Aretini P, Isola P, Tancredi M, Collecchi P, Naccarato AG, Porta RP, Bevilacqua G, Caligo MA. High level of messenger RNA for BRMS1 in primary breast carcinomas is associated with poor prognosis. Int J Cancer. 2007;120:1169–78. doi: 10.1002/ijc.22379. [DOI] [PubMed] [Google Scholar]
- 18.Tsukamoto N, Hattori M, Yang H, Bos JL, Minato N. Rap1 GTPase-activating protein SPA-1 negatively regulates cell adhesion. J Biol Chem. 1999;274:18463–9. doi: 10.1074/jbc.274.26.18463. [DOI] [PubMed] [Google Scholar]
- 19.Crawford NPS, Walker RC, Lukes L, Officewala JS, Williams RW, Hunter KW. The Diasporin Pathway: a tumor progression-related transcriptional network that predicts breast cancer survival. Clin Exp Metastas. 2008;25:357–69. doi: 10.1007/s10585-008-9146-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Park YG, Zhao X, Lesueur F, Lowy DR, Lancaster M, Pharoah P, Qian X, Hunter KW, Yang H, R D. Sipa1 is a candidate for the metastasis efficiency modifier locus Mtes1. Nat Genet. 2005;37:1055–1062. doi: 10.1038/ng1635. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Crawford NPS, Ziogas A, Peel DJ, Hess J, Anton-Culver H, Hunter KW. Germline polymorphisms in SIPA1 are associated with metastasis and other indicators of poor prognosis in breast cancer. Breast Cancer Res. 2006;8:R16. doi: 10.1186/bcr1389. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Gaudet MM, Hunter K, Pharoah P, Dunning AM, Driver K, Sherman M, Peplonska B, Brinton LA, Lissowska J, Chanock S, Garcia-Closas M. Genetic variation in SIPA1 in relation to breast cancer risk and survival after breast cancer diagnosis. Int J Cancer. 2009;124:1716–20. doi: 10.1002/ijc.23919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hsieh SM, Look MP, Sieuwerts AM, Foekens Ja, Hunter KW. Distinct inherited metastasis susceptibility exists for different breast cancer subtypes: a prognosis study. Breast Cancer Res. 2009;11:R75. doi: 10.1186/bcr2412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ambrosone CB, Nesline MK, Davis W. Establishing a cancer center data bank and biorepository for multidisciplinary research. Cancer Epidem Biomar. 2006;15:1575–7. doi: 10.1158/1055-9965.EPI-06-0628. [DOI] [PubMed] [Google Scholar]
- 25.Sherry ST, Ward MH, Kholodov M, Baker J, Phan L, Smigielski EM, Sirotkin K. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001;29:308–11. doi: 10.1093/nar/29.1.308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Meehan WJ, Welch DR. Breast cancer metastasis suppressor 1: update. Clin Exp Metastas. 2003;20:45–50. doi: 10.1023/a:1022542519586. [DOI] [PubMed] [Google Scholar]
- 27.Meehan WJ, Samant RS, Hopper JE, Carrozza MJ, Shevde LA, Workman JL, Eckert KA, Verderame MF, Welch DR. Breast cancer metastasis suppressor 1 (BRMS1) forms complexes with retinoblastoma-binding protein 1 (RBP1) and the mSin3 histone deacetylase complex and represses transcription. J Biol Chem. 2004;279:1562–9. doi: 10.1074/jbc.M307969200. [DOI] [PubMed] [Google Scholar]