The Cancer Genome Atlas (TCGA) is one of the largest biorepositories of digital histology. Deep learning (DL) models have been trained on TCGA to predict numerous features directly from histology, ...including survival, gene expression patterns, and driver mutations. However, we demonstrate that these features vary substantially across tissue submitting sites in TCGA for over 3,000 patients with six cancer subtypes. Additionally, we show that histologic image differences between submitting sites can easily be identified with DL. Site detection remains possible despite commonly used color normalization and augmentation methods, and we quantify the image characteristics constituting this site-specific digital histology signature. We demonstrate that these site-specific signatures lead to biased accuracy for prediction of features including survival, genomic mutations, and tumor stage. Furthermore, ethnicity can also be inferred from site-specific signatures, which must be accounted for to ensure equitable application of DL. These site-specific signatures can lead to overoptimistic estimates of model performance, and we propose a quadratic programming method that abrogates this bias by ensuring models are not trained and validated on samples from the same site.
In 2012, the US Preventive Services Task Force (USPSTF) discouraged prostate-specific antigen (PSA) -based prostate cancer screening. Previous USPSTF recommendations did not appreciably alter ...prostate cancer screening. Therefore, we designed a trend analysis to determine the population-based impact of the 2012 recommendation.
The nationally representative National Health Interview Survey was used to estimate the proportion of men age 40 years and older who saw a physician and were screened for prostate cancer in 2013. An externally validated 9-year mortality index was used to analyze screening rates based on remaining life expectancy. Screening rates from 2005, 2010, and 2013 were compared using logistic regression.
PSA-based screening did not significantly change from 2010 to 2013 among 40- to 49-year-old men (from 12.5% to 11.2%; P = .4). Screening rates significantly declined in men age 50 to 59 years (from 33.2% to 24.8%; P < .01), age 60 to 74 years (from 51.2% to 43.6%; P < .01), and age 75 years or older (from 43.9% to 37.1%; P = .03). A large percentage of men were screened for prostate cancer despite a high risk (> 52%) of 9-year mortality, including approximately one third of men older than age 75 years. Approximately 1.4 million men age 65 years or older with a high risk (> 52%) of 9-year mortality were screened in 2013.
Prostate cancer screening significantly declined among men older than age 50 years after the 2012 USPSTF guideline discouraging PSA-based screening. A significant proportion of men continue to be screened despite a high risk of 9-year mortality, including one third of men age 75 years and older.
Although Wnt/β-catenin pathway activation has been implicated in mouse models of breast cancer, there is contradictory evidence regarding its importance in human breast cancer. In this study, ...invasive and in situ breast cancer tissue microarrays containing luminal A, luminal B, human epidermal growth factor receptor 2 (HER2)+ /ER− and basal-like breast cancers were analyzed for β-catenin subcellular localization. We demonstrate that nuclear and cytosolic accumulation of β-catenin, a read-out of Wnt pathway activation, was enriched in basal-like breast cancers. In contrast, membrane-associated β-catenin was observed in all breast cancer subtypes, and its expression decreased with tumor progression. Moreover, nuclear and cytosolic localization of β-catenin was associated with other markers of the basal-like phenotype, including nuclear hormone receptor and HER2 negativity, cytokeratin 5/6 and vimentin expression, and stem cell enrichment. Importantly, this subcellular localization of β-catenin was associated with a poor outcome and is more frequently observed in tumors from black patients. In addition, β-catenin accumulation was more often observed in basal-like in situ carcinomas than other in situ subtypes, suggesting that activation of this pathway might be an early event in basal-like tumor development. Collectively, these data indicate that Wnt/β-catenin activation is an important feature of basal-like breast cancers and is predictive of worse overall survival, suggesting that it may be an attractive pharmacological target for this aggressive breast cancer subtype.
Among 695 women with clear-cell carcinoma of the vagina and cervix who were followed for a median of nearly 23 years, 5-year survival was 86% among women with documented diethylstilbestrol exposure ...and 81% among those without such exposure.
The aim of the current study was to determine the prevalence and clinical predictors of germline cancer susceptibility mutations in patients with malignant mesothelioma (MM).
We performed targeted ...capture and next-generation sequencing of 85 cancer susceptibility genes on germline DNA from 198 patients with pleural, peritoneal, and tunica vaginalis MM.
Twenty-four germline mutations were identified in 13 genes in 23 (12%) of 198 patients. BAP1 mutations were the most common (n = 6; 25%). The remaining were in genes involved in DNA damage sensing and repair (n = 14), oxygen sensing (n = 2), endosome trafficking (n = 1), and cell growth (n = 1). Pleural site (odds ratio OR, 0.23; 95% CI, 0.10 to 0.58; P < .01), asbestos exposure (OR, 0.28; 95% CI, 0.11 to 0.72; P < .01), and older age (OR, 0.95; 95% CI, 0.92 to 0.99; P = .01) were associated with decreased odds of carrying a germline mutation, whereas having a second cancer diagnosis (OR, 3.33; 95% CI, 1.22 to 9.07; P = .02) significantly increased the odds. The odds of carrying a mutation in BAP1 (OR, 1,658; 95% CI, 199 to 76,224; P < .001), BRCA2 (OR, 5; 95% CI, 1.0 to 14.7; P = .03), CDKN2A (OR, 53; 95% CI, 6 to 249; P < .001), TMEM127 (OR, 88; 95% CI, 1.7 to 1,105; P = .01), VHL (OR, 51; 95% CI, 1.1 to 453; P = .02), and WT1 (OR, 20; 95% CI, 0.5 to 135; P = .049) were significantly higher in MM cases than in a noncancer control population. Tumor sequencing identified mutations in a homologous recombination pathway gene in 52% (n = 29 of 54).
A significant proportion of patients with MM carry germline mutations in cancer susceptibility genes, especially those with peritoneal MM, minimal asbestos exposure, young age, and a second cancer diagnosis. These data support clinical germline genetic testing for patients with MM and provide a rationale for additional investigation of the homologous recombination pathway in MM.
Effective targeting of cancer stem cells is necessary and important for eradicating cancer and reducing metastasis-related mortality. Understanding of cancer stemness-related signaling pathways at ...the molecular level will help control cancer and stop metastasis in the clinic.
By analyzing miRNA profiles and functions in cancer development, we aimed to identify regulators of breast tumor stemness and metastasis in human xenograft models
and examined their effects on self-renewal and invasion of breast cancer cells
To discover the direct targets and essential signaling pathways responsible for miRNA functions in breast cancer progression, we performed microarray analysis and target gene prediction in combination with functional studies on candidate genes (overexpression rescues and pheno-copying knockdowns).
In this study, we report that hsa-miR-206 suppresses breast tumor stemness and metastasis by inhibiting both self-renewal and invasion. We identified that among the candidate targets, twinfilin (
) rescues the miR-206 phenotype in invasion by enhancing the actin cytoskeleton dynamics and the activity of the mesenchymal lineage transcription factors, megakaryoblastic leukemia (translocation) 1 (MKL1), and serum response factor (SRF). MKL1 and SRF were further demonstrated to promote the expression of
, which is essential for miR-206's function in inhibiting both invasion and stemness of breast cancer.
The identification of the miR-206/TWF1/MKL1-SRF/IL11 signaling pathway sheds lights on the understanding of breast cancer initiation and progression, unveils new therapeutic targets, and facilitates innovative drug development to control cancer and block metastasis.
.
Although several transcriptome-wide association studies (TWASs) have been performed to identify genes associated with overall breast cancer (BC) risk, only a few TWAS have explored the differences in ...estrogen receptor-positive (ER+) and estrogen receptor-negative (ER-) breast cancer. Additionally, these studies were based on gene expression prediction models trained primarily in breast tissue, and they did not account for alternative splicing of genes.
In this study, we utilized two approaches to perform multi-tissue TWASs of breast cancer by ER subtype: (1) an expression-based TWAS that combined TWAS signals for each gene across multiple tissues and (2) a splicing-based TWAS that combined TWAS signals of all excised introns for each gene across tissues. To perform this TWAS, we utilized summary statistics for ER + BC from the Breast Cancer Association Consortium (BCAC) and for ER- BC from a meta-analysis of BCAC and the Consortium of Investigators of Modifiers of BRCA1 and BRCA2 (CIMBA).
In total, we identified 230 genes in 86 loci that were associated with ER + BC and 66 genes in 29 loci that were associated with ER- BC at a Bonferroni threshold of significance. Of these genes, 2 genes associated with ER + BC at the 1q21.1 locus were located at least 1 Mb from published GWAS hits. For several well-studied tumor suppressor genes such as TP53 and CHEK2 which have historically been thought to impact BC risk through rare, penetrant mutations, we discovered that common variants, which modulate gene expression, may additionally contribute to ER + or ER- etiology.
Our study comprehensively examined how differences in common variation contribute to molecular differences between ER + and ER- BC and introduces a novel, splicing-based framework that can be used in future TWAS studies.
Women exposed to diethylstilbestrol (DES) in utero are at increased risk for the development of vaginal and cervical clear cell adenocarcinoma (CCA) at younger age. It is unknown if a second peak ...will occur in later life, the ages when CCA developed spontaneously in the pre-DES era. The complete epidemiologic curve of CCA has not been reported, yet.
We reviewed 720 cases of CCA from the CCA registry at the University of Chicago through 2014. Incidence rates and cumulative risks for CCA were calculated based on white women born in the U.S. from 1948 through 1971.
In 420 CCA cases there was documented evidence of prenatal DES exposure. 80% were among those between ages 15 and 31 but some occurred as late as age 55. A small second peak occurred around age 42. The risk of DES-related CCA was highest in the 1951–1956 birth cohort and this birth cohort effect closely correlated with DES prescriptions over time in the U.S. (r=0.98, P=0.005). By age 50, the cumulative risk of CCA was 1 per 750 exposed women. CCA cases without evidence of DES exposure had similar ages, year of diagnosis, and birth cohort patterns as the documented DES-exposed cases, suggesting that some negative cases were exposed. Their inclusion raises the cumulative risk of CCA to 1 per 520.
With the largest data available, our results confirmed the association between prenatal DES exposure and clear cell adenocarcinoma. The study also refines the risks of DES-related CCA.
•The rate of DES-related CCA peaked at age 19; there was a second peak at age 42.•DES-related CCA has occurred in those as old as 55years.•CCA risk across birth cohorts was closely correlated with DES prescription over time.•The cumulative risk of CCA up to the age of 50 is 1 per 750 exposed.•There is evidence that many of the cases with negative histories were exposed.
To evaluate weight change patterns over time following the diagnosis of breast cancer and to examine the association of post-diagnosis weight change and survival outcomes in Black and White patients.
...The study included 2888 women diagnosed with non-metastatic breast cancer in 2000-2017 in Chicago. Longitudinal repeated measures of weight and height were collected, along with a questionnaire survey including questions on body size. Multilevel mixed-effects models were used to examine changes in body mass index (BMI). Delayed entry Cox proportional hazards models were used to investigate the impacts of changing slope of BMI on survival outcomes.
At diagnosis, most patients were overweight or obese with a mean BMI of 27.5 kg/m
and 31.5 kg/m
for Blacks and Whites, respectively. Notably, about 45% of the patients had cachexia before death and substantial weight loss started about 30 months before death. In multivariable-adjusted analyses, compared to stable weight, BMI loss (> 0.5 kg/m
/year) showed greater than 2-fold increased risk in overall survival (hazard ratio HR = 2.60, 95% CI 1.88-3.59), breast cancer-specific survival (HR = 3.05, 95% CI 1.91-4.86), and disease-free survival (HR = 2.12, 95% CI 1.52-2.96). The associations were not modified by race, age at diagnosis, and pre-diagnostic weight. BMI gain (> 0.5 kg/m
/year) was also related to worse survival, but the effect was weak (HR = 1.60, 95% CI 1.10-2.33 for overall survival).
BMI loss is a strong predictor of worse breast cancer outcomes. Growing prevalence of obesity may hide diagnosis of cancer cachexia, which can occur in a large proportion of breast cancer patients long before death.
Recent reports have shown that the breast cancer incidence rate in the US stabilized after a sharp reduction in 2002 and 2003. It is important to continue monitoring breast cancer incidence rates ...according to age group, race/ethnicity, estrogen receptor (ER) status, and tumor stage. Age-standardized breast cancer incidence rates were calculated using data from the surveillance, epidemiology, and end results 18 registries from 2000 to 2009, for 677,774 female breast cancer patients aged 20 and above. Jointpoint regression models were used to fit a series of joined straight lines on a log scale to annual age-standardized rates. The incidence rates of all breast cancer significantly increased for non-Hispanic blacks from 2005 to 2009 (annual percentage change, APC = 2.0 %,
p
= 0.01) and Asian/Pacific Islanders from 2000 to 2009 (APC = 1.2 %,
p
= 0.02). Since 2004, incidence rates in women aged 40–49 years significantly increased for most racial/ethnic groups (overall APC = 1.1 %,
p
= 0.001). The incidence rate of carcinoma in situ significantly increased in all racial/ethnic groups, with an APC range from 2.3 to 3.0 % (
p
< 0.005). The localized breast cancer incidence significantly increased in non-Hispanic blacks (APC = 1.3 %,
p
= 0.004) and Asians (APC = 1.2 %,
p
= 0.03). ER-positive breast cancer significantly increased in almost all age/race sub-groups after 2005 (APC by race: non-Hispanic whites 1.5 %, non-Hispanic blacks 4.3 %, Asian/Pacific Islanders 1.7 %, and Hispanics 1.8 %; all
p
values <0.05), while ER-negative breast cancer decreased in most sub-groups (APC by race: non-Hispanic whites—3.9 %, non-Hispanic blacks—3.7 %, Asian/Pacific Islanders—1.5 %, and Hispanics—4.3 %; all
p
values <0.05). Recently the incidence of breast cancer appears to be increasing in certain subgroups, including ER-positive, early-stage breast cancers, in particular among non-Hispanic blacks and Asian/Pacific Islanders. Further studies are warranted to examine possible reasons for these changes, such as changes in mammography screening methods and risk factors prevalence.