A Latent Disease Model to Reduce Detection Bias in Cancer Risk Prediction Studies

In studies of cancer risk, detection bias arises when risk factors are associated with screening patterns, affecting the likelihood and timing of diagnosis. To eliminate detection bias in a screened cohort, we propose modeling the latent onset of cancer and estimating the association between risk factors and onset rather than diagnosis. We apply this framework to estimate the increase in prostate cancer risk associated with black race and family history using data from the SELECT prostate cancer prevention trial, in which men were screened and biopsied according to community practices. A positive family history was associated with a hazard ratio (HR) of prostate cancer onset of 1.8, lower than the corresponding HR of prostate cancer diagnosis (HR = 2.2). This result comports with a finding that men in SELECT with a family history were more likely to be biopsied following a positive PSA test than men with no family history. For black race, the HRs for onset and diagnosis were similar, consistent with similar patterns of screening and biopsy by race. If individual screening and diagnosis histories are available, latent disease modeling can be used to decouple risk of disease from risk of disease diagnosis and reduce detection bias.

Download Full-text

Multi-cohort modeling strategies for scalable globally accessible prostate cancer risk tools

BMC Medical Research Methodology ◽

10.1186/s12874-019-0839-0 ◽

2019 ◽

Vol 19 (1) ◽

Author(s):

Johanna Tolksdorf ◽

Michael W. Kattan ◽

Stephen A. Boorjian ◽

Stephen J. Freedland ◽

Karim Saba ◽

...

Keyword(s):

Prostate Cancer ◽

Risk Factors ◽

Family History ◽

Cancer Risk ◽

Risk Prediction ◽

Prostate Biopsy ◽

Prostate Cancer Risk ◽

High Grade ◽

Prediction Tools ◽

History Of

Abstract Background Online clinical risk prediction tools built on data from multiple cohorts are increasingly being utilized for contemporary doctor-patient decision-making and validation. This report outlines a comprehensive data science strategy for building such tools with application to the Prostate Biopsy Collaborative Group prostate cancer risk prediction tool. Methods We created models for high-grade prostate cancer risk using six established risk factors. The data comprised 8492 prostate biopsies collected from ten institutions, 2 in Europe and 8 across North America. We calculated area under the receiver operating characteristic curve (AUC) for discrimination, the Hosmer-Lemeshow test statistic (HLS) for calibration and the clinical net benefit at risk threshold 15%. We implemented several internal cross-validation schemes to assess the influence of modeling method and individual cohort on validation performance. Results High-grade disease prevalence ranged from 18% in Zurich (1863 biopsies) to 39% in UT Health San Antonio (899 biopsies). Visualization revealed outliers in terms of risk factors, including San Juan VA (51% abnormal digital rectal exam), Durham VA (63% African American), and Zurich (2.8% family history). Exclusion of any cohort did not significantly affect the AUC or HLS, nor did the choice of prediction model (pooled, random-effects, meta-analysis). Excluding the lowest-prevalence Zurich cohort from training sets did not statistically significantly change the validation metrics for any of the individual cohorts, except for Sunnybrook, where the effect on the AUC was minimal. Therefore the final multivariable logistic model was built by pooling the data from all cohorts using logistic regression. Higher prostate-specific antigen and age, abnormal digital rectal exam, African ancestry and a family history of prostate cancer increased risk of high-grade prostate cancer, while a history of a prior negative prostate biopsy decreased risk (all p-values < 0.004). Conclusions We have outlined a multi-cohort model-building internal validation strategy for developing globally accessible and scalable risk prediction tools.

Download Full-text

Common genetic and clinical risk factors: Association with fatal prostate cancer in the Cohort of Swedish Men

10.1101/2020.09.16.20192666 ◽

2020 ◽

Author(s):

Minh-Phuong Huynh-Le ◽

Roshan Karunamuni ◽

Chun Chieh Fan ◽

Wesley K Thompson ◽

Kenneth Muir ◽

...

Keyword(s):

Prostate Cancer ◽

Risk Factors ◽

Family History ◽

Cancer Risk ◽

Risk Stratification ◽

Prostate Cancer Risk ◽

Cancer Death ◽

Clinical Variables ◽

Prostate Cancer Death ◽

Fatal Prostate Cancer

Background: Clinical variables--age, family history, genetics--are used for prostate cancer risk stratification. Recently, polygenic hazard scores (PHS46, PHS166) were validated as associated with age at prostate cancer diagnosis. While polygenic scores are associated with all prostate cancer (not specific for fatal cancers), PHS46 was also associated with age at prostate cancer death. We evaluated if adding PHS to clinical variables improves associations with prostate cancer death. Methods: Genotype/phenotype data were obtained from a nested case-control Cohort of Swedish Men (n=3,279; 2,163 with prostate cancer, 278 prostate cancer deaths). PHS and clinical variables (family history, alcohol intake, smoking, heart disease, hypertension, diabetes, body mass index) were tested via univariable Cox proportional hazards models for association with age at prostate cancer death. Multivariable Cox models with/without PHS were compared with log-likelihood tests. Results: Median age at last follow-up/prostate cancer death were 78.0 (IQR: 72.3-84.1) and 81.4 (75.4-86.3) years, respectively. On univariable analysis, PHS46 (HR 3.41 [95%CI 2.78-4.17]), family history (HR 1.72 [1.46-2.03]), alcohol (HR 1.74 [1.40-2.15]), diabetes (HR 0.53 [0.37-0.75]) were each associated with prostate cancer death. On multivariable analysis, PHS46 (HR 2.45 [1.99-2.97]), family history (HR 1.73 [1.48-2.03]), alcohol (HR 1.45 [1.19-1.76]), diabetes (HR 0.62 [0.42-0.90]) all remained associated with fatal disease. Including PHS46 or PHS166 improved multivariable models for fatal prostate cancer (p<10-15). Conclusions: PHS had the most robust association with fatal prostate cancer in a multivariable model with common risk factors, including family history. Impact: Adding PHS to clinical variables may improve prostate cancer risk stratification strategies.

Download Full-text

Demographic variation and risk factors regarding breast cancer among female in Southern Punjab, Pakistan

Journal of the Pakistan Medical Association ◽

10.47391/jpma.1091 ◽

2021 ◽

pp. 1-16

Author(s):

Waheed Ahmad ◽

Sabika Firasat ◽

Muhammad Sohail Akhtar ◽

Kiran Afshan ◽

Kaukab Jabeen ◽

...

Keyword(s):

Breast Cancer ◽

Risk Factors ◽

Logistic Regression ◽

Family History ◽

Breast Cancer Risk ◽

Cancer Risk ◽

Conditional Logistic Regression ◽

Positive Family History ◽

Study Cohort ◽

Screening Programs

Objective: Breast cancer is a second major cause of female death worldwide. This study aimed to explore epidemiology, clinical profiles and contribution of reproductive and non-reproductive risk factors in breast cancer development among females from South Punjab, Pakistan. Methods: Data was collected through hospitals between October 2017 and March 2018 and study got approval by Bioethical Committee of Quaid-i-Azam University in September, 2017. A total of 163 cases and 163 age-matched controls were recruited through non-probability consecutive sampling method. All histologically confirmed patients irrespective of age, family history, clinical presentation and histopathological type were included in the study as cases. Patients, who were not willing to participate were excluded from the study. Details regarding socio-demographic characteristics, family history of cancer, reproductive health and lifestyle factors were recorded using a structured questionnaire. Conditional logistic regression was performed to calculate odds ratios at 95% confidence intervals for breast cancer by menstrual and reproductive factors after adjustment of potential confounders. Conditional logistic regression was also applied for various demographic and medical risk factors/exposures. Results: We found positive family history and hypertension significantly linked to an increased breast cancer risk (adjusted O.R >1.5, 95% CI, P<0.05) whereas, intense physical activity, increased anthropometric measurements and breastfeeding per child in months were inversely associated with breast cancer risk (adjusted O.R <1.0, 95% CI, P<0.05) in our study cohort. Conclusion: Our study reaffirms contribution of established risk factors for breast cancer, highlights protective factors and necessitates awareness/screening programs to reduce breast cancer burden in upcoming generations. Continuous...

Download Full-text

Common genetic and clinical risk factors: Association with fatal prostate cancer in the Cohort of Swedish Men.

Journal of Clinical Oncology ◽

10.1200/jco.2021.39.6_suppl.65 ◽

2021 ◽

Vol 39 (6_suppl) ◽

pp. 65-65

Author(s):

Minh-Phuong Huynh-Le ◽

Roshan Karunamuni ◽

Chun Chieh Fan ◽

Wesley K Thompson ◽

Kenneth Muir ◽

...

Keyword(s):

Prostate Cancer ◽

Risk Factors ◽

Family History ◽

Cancer Risk ◽

Alcohol Intake ◽

Prostate Cancer Risk ◽

Cancer Death ◽

Clinical Variables ◽

Prostate Cancer Death ◽

Fatal Prostate Cancer

65 Background: Clinical variables (age, family history, and genetics) are commonly used for prostate cancer risk stratification. Recently, polygenic hazard scores (PHS46, PHS166) were validated as associated with age at prostate cancer diagnosis. While polygenic scores, including PHS, are associated with all prostate cancer and are not specific for fatal cancers, PHS46 was also associated with age at prostate cancer death. We evaluated if adding PHS to available clinical variables improves associations with prostate cancer death. Methods: Genotype and phenotype data were obtained from a nested case-control subset (n=3,279; 2,163 were diagnosed with prostate cancer, 278 died of prostate cancer) of the longitudinal, population-based Cohort of Swedish Men. PHS and clinical variables (family history, alcohol intake, smoking, heart disease, hypertension, diabetes history, and body mass index) were independently tested via univariable Cox proportional hazards models for association with age at prostate cancer death. Multivariable Cox models were constructed with clinical variables and PHS. Log-likelihood tests compared models. Results: Median age at last follow-up and at prostate cancer death were 78.0 (IQR: 72.3-84.1) and 81.4 (75.4-86.3) years, respectively. On univariable analysis, PHS46 (HR 3.41 [95% CI 2.78-4.17]), family history (HR 1.72 [1.46-2.03]), alcohol intake (HR 1.74 [1.40-2.15]), and diabetes (HR 0.53 [0.37-0.75]) were each associated with prostate cancer death. A multivariable clinical model including PHS46 improved associations for fatal disease ( p<10−15). On multivariable analysis, PHS46 (HR 2.45 [1.99-2.97]), family history (HR 1.73 [1.48-2.03]), alcohol intake (HR 1.45 [1.19-1.76]), and diabetes (HR 0.62 [0.42-0.90]) all remained associated with prostate cancer death. Similar results were found using the newer PHS166. Conclusions: PHS had the most robust association with fatal prostate cancer in a multivariable model with common clinical risk factors, including family history. Adding PHS to clinical variables may improve individualized prostate cancer risk stratification strategies.

Download Full-text

A common nonsense mutation in EphB2 is associated with prostate cancer risk in African American men with a positive family history

Journal of Medical Genetics ◽

10.1136/jmg.2005.035790 ◽

2006 ◽

Vol 43 (6) ◽

pp. 507-511 ◽

Cited By ~ 38

Author(s):

R A Kittles

Keyword(s):

Prostate Cancer ◽

African American ◽

Family History ◽

Cancer Risk ◽

Nonsense Mutation ◽

African American Men ◽

Positive Family History ◽

Prostate Cancer Risk

Download Full-text

Prostate Cancer Risk with Positive Family History, Normal Prostate Examination Findings, and PSA Less Than 4.0 ng/mL

Urology ◽

10.1016/j.urology.2007.06.1105 ◽

2007 ◽

Vol 70 (4) ◽

pp. 748-752 ◽

Cited By ~ 10

Author(s):

Edith Canby-Hagino ◽

Javier Hernandez ◽

Timothy C. Brand ◽

Dean A. Troyer ◽

Betsy Higgins ◽

...

Keyword(s):

Prostate Cancer ◽

Family History ◽

Cancer Risk ◽

Positive Family History ◽

Prostate Cancer Risk ◽

Normal Prostate

Download Full-text

649: Prostate Cancer Screening in Persons with a Positive Family History

The Journal of Urology ◽

10.1016/s0022-5347(18)37911-4 ◽

2004 ◽

Vol 171 (4S) ◽

pp. 172-173

Author(s):

Kathleen Herkommer ◽

Juergen E. Gschwend ◽

Martina Kron ◽

Richard E. Hautmann ◽

Thomas Paiss

Keyword(s):

Prostate Cancer ◽

Cancer Screening ◽

Family History ◽

Prostate Cancer Screening ◽

Positive Family History

Download Full-text

Prostate Cancer

10.1093/oso/9780190676827.003.0020 ◽

2018 ◽

Author(s):

Kathryn M. Wilson ◽

Lorelei Mucci

Keyword(s):

Prostate Cancer ◽

Risk Factors ◽

Family History ◽

Prostate Specific Antigen ◽

Advanced Prostate Cancer ◽

Advanced Disease ◽

African Ancestry ◽

Specific Antigen ◽

Dietary Factors ◽

Increased Risk

Prostate cancer is among the most commonly diagnosed cancers among men, ranking second in cancer globally and first in Western countries. There are marked variations in incidence globally, and its incidence must be interpreted in the context of diagnostic intensity and screening. The uptake of prostate-specific antigen screening since the 1990s has led to dramatic increases in incidence in many countries, resulting in an increased proportion of indolent cancers that would never have come to light clinically in the absence of screening. Risk factors differ when studying prostate cancer overall versus advanced disease. Older age, African ancestry, and family history are established risk factors for prostate cancer. Obesity and smoking are not associated with risk overall, but are associated with increased risk of advanced prostate cancer. Several additional lifestyle factors, medications, and dietary factors are now emerging as risk factors for advanced disease.

Download Full-text

Family history of prostate cancer in patients with localized prostate cancer: an independent predictor of treatment outcome.

Journal of Clinical Oncology ◽

10.1200/jco.1997.15.4.1478 ◽

1997 ◽

Vol 15 (4) ◽

pp. 1478-1480 ◽

Cited By ~ 69

Author(s):

P A Kupelian ◽

V A Kupelian ◽

J S Witte ◽

R Macklis ◽

E A Klein

Keyword(s):

Prostate Cancer ◽

Treatment Outcome ◽

Family History ◽

Treatment Modality ◽

Proportional Hazards ◽

Positive Family History ◽

Specific Antigen ◽

Tumor Stage ◽

Degree Relative ◽

History Of

PURPOSE To determine if familial prostate cancer patients have a less favorable prognosis than patients with sporadic prostate cancer after treatment for localized disease with either radiotherapy (RT) or radical prostatectomy (RP). PATIENTS AND METHODS One thousand thirty-eight patients treated with either RT (n = 583) or RP (n = 455) were included in this analysis. These patients were noted as having a positive family history if they confirmed the diagnosis of prostate cancer in a first-degree relative. The outcome of interest was biochemical relapse-free survival (bRFS). We used proportional hazards to analyze the effect of the presence of family history and other potential confounding variables (ie, age, treatment modality, stage, biopsy Gleason sum [GS], and initial prostate-specific antigen [iPSA] levels) on treatment outcome. RESULTS Eleven percent of all patients had a positive family history. The 5-year bRFS rates for patients with negative and positive family histories were 52% and 29%, respectively (P < .001). The potential confounders with bRFS rates were iPSA levels, biopsy GS, and clinical tumor stage; treatment modality and age did not appear to be associated with outcome. After adjusting for potential confounders, family history of prostate cancer remained strongly associated with biochemical failure. CONCLUSION This is the first study to demonstrate that the presence of a family history of prostate cancer correlates with treatment outcome in a large unselected series of patients. Our findings suggest that familial prostate cancer may have a more aggressive course than nonfamilial prostate cancer, and that clinical and/or pathologic parameters may not adequately predict this course.

Download Full-text